This page contains discussions that have been archived from Village pump (technical). Please do not edit the contents of this page. If you wish to revive any of these discussions, either start a new thread or use the talk page associated with that topic.
Hi @Monorodo:, you will need to be more detailed in what your challenge is. First, only administrators can delete pages. Please see the edit notice above ("Welcome to the village pump for technical issues.") and provide details about what is occurring. Thank you, — xaosfluxTalk22:04, 4 April 2018 (UTC)
{{Infobox football biography
| name = Adam Ståhl
| image =
| fullname = Adam Mikael Sven Ståhl
| birth_date = {{birth date and age|1994|10|8}}
| birth_place = [[Malmö]], Sweden
| height = {{convert|1.86|m|abbr=on}}
| position = [[Midfielder]]
| currentclub = [[Kardemir Karabükspor]]
| clubnumber = 18
}}
any word
###
We need just the Infobox and navbox and <title>New France</title>
<page>
<title>New France</title>
<ns>0</ns>
<id>159828</id>
<revision>
<id>822862933</id>
{{Infobox former country
|native_name = ''Nouvelle-France''
|conventional_long_name = New France
|common_name = New France
}}
'''New France''' ({{lang-fr|Nouvelle-France}}) was the area [[French colonization of the Americas|colonized]] by [[France]]
{{Navboxes
|title=Articles related to New France
|list1=
{{Canadian colonies}}
{{Canadian history}}
{{Former French colonies}}
{{European Colonization of North America}}
}}
{{DEFAULTSORT:New France}}
[[Category:Colonial United States (French)| ]]
I am use this regex (?=\{Infobox former conuntry)(\{([^{}]|(?1))*\})
this regex select only infobox.
now I want select all except navbox and infobox? what is the best regex ? thanksMonorodo (talk) 10:21, 7 April 2018 (UTC)
@Monorodo: If you use a regex, your matching is likely to fail if the order of the parameters changes, or if there is a template inside one of the parameters, etc. The best way to do this is to use an XML parser to get the wikitext of the whole page, and then to use a wikitext parser to get the information from the wikitext. If you're using Python, you can use xml.etree.ElementTree and MWParserFromHell. Best — Mr. Stradivarius♪ talk ♪01:24, 8 April 2018 (UTC)
regex for wikitable in xml file
{| class="wikitable"
|-
!colspan=9|[[Football at the Islamic Solidarity Games|Football at the Islamic Solidarity Games record]]
|-
|-
!Year
!Round
!Position
!{{Tooltip|Pld|Games played}}
!{{Tooltip|W|Won}}
!{{Tooltip|D|Drawn}}
!{{Tooltip|L|Lost}}
!{{Tooltip|GF|Goals for}}
!{{Tooltip|GA|Goals against}}
|-
|{{flagicon|Saudi Arabia}} [[Football at the 2005 Islamic Solidarity Games|2005]]||[[File:Silver medal icon.svg|16px]] ''Silver Medal''||2nd||5||2||2||1||4||2
|-
|{{flagicon|Iran}} [[2010 Islamic Solidarity Games|2010]]||colspan=8|''Cancelled''
|-
|{{flagicon|Indonesia}} [[Football at the 2013 Islamic Solidarity Games|2013]]||[[File:Gold medal icon.svg|16px]] '''Gold Medal'''||1st||4||3||0||1||6||3
|-
|{{flagicon|Azerbaijan}} [[Football at the 2017 Islamic Solidarity Games|2017]]||Group Stage||5th||3||1||2||0||2||1
|-
|{{flagicon|Turkey}} [[2021 Islamic Solidarity Games|2021]]||colspan=8|To Be Determined
|-
!Total||1 Title||3/3||12||6||4||2||12||6
|}
If you can make some assumptions about the input, this can be greatly simplified. regex in itself only matches a pattern, perhaps putting the pieces into variables that can be used later. Do you want to code this in javascript, perhaps using the "replace" function? Then you can run it from your browser on request. I myself have coded this sort of thing by first clearing all variable values when encountering the xml code for the start of an object that is a row in your table, then when matching the codes signalling the start of field values, match everything to the end and putting that into a variable, then at the end xml tag for the structure print out your info in the table format. consider the /m for multiline, or /g for global repeated action on the string. Graeme Bartlett (talk) 00:39, 11 April 2018 (UTC)
@Monorodo: It guess it'll be a lot easier to help you, if you tell use, what exactly you are trying to archive. So:
What exactly do you need? Wikitext? HTML? Or just the content?
And what for? Another wiki(-article)? HTML? Word? A Database?
Since parsing nested templates in WikiText is a pain, probably the easiest way is not to use the Wikitext at all but it's HTML-representation. Thanks to the Document Object Model it is quite easy to access specific nodes (and there content) via JavaScript (or any other HTML-capable programming language).
E.g. the following JS gives you all the tables in the Morocco_national_football_team-Artikel:
== See also ==
* [[Wallachia]]
== Gallery ==
<gallery>
File:Curtea de Arges.E.jpg|St. Nicholas Princely Church
File:Catedrala Curtea de Arges.jpg|Curtea de Argeș Cathedral in an 1880 engraving
</gallery>
Bold and italic buttons in edit toolbar obscuring edit window text (Monobook, IE11)
On Monobook (IE11) the bold and italic buttons in the edit toolbar have suddenly started showing up for me as oversized squares that obscure the beginning of the top line of the text in the edit window. On Vector I still see the odd squares around the B and the I, but they don't protrude into the edit window. Is this happening for anyone else? Double sharp (talk) 14:19, 13 April 2018 (UTC)
It works for me. Does it happen if you log out and edit with MonoBook? What about safemode? Try to clear your cache. Do you see the single letters B and I or boxes spelling out "Bold" and "Italics"? I get rectangular hover boxes with the full words when I hover over the buttons but they don't obscure the edit area. PrimeHunter (talk) 15:51, 13 April 2018 (UTC)
MediaWiki tweaks may sometimes change the result. If you don't want to remove all of it then you can experiment by previewing User:Double sharp/common.css with variations of it. If you are in a special editor for CSS pages then click the <> icon at the top left to change to the normal editor so you can see the effect during normal editing. PrimeHunter (talk) 10:36, 14 April 2018 (UTC)
Line 10 in the first example was missing a close bracket, and line 10 and 11 were missing semicolons. Line 12 in vector.js had an extra bracket. I've removed/added them, but I can't test it to see if it worked - you'll need to confirm if they were the issues. - Bilby (talk) 08:40, 14 April 2018 (UTC)
If you try to access an archive that doesn't exist, like if I have two talk archive pages and you try to access Archive_3, which would normally be redlinked, Wikipedia says, "Post a message to Nick Levinson/Archive 3." It also says, "No messages have been posted for this user yet." Of course, they have been posted, but to the current talk page, not into the redlinked archive. I'm not sure if any users have nonexistent archives on their watchlists and would ever find out if any message were posted into a nonexistent archive, at least until the recipient tried to create a new archive, perhaps months or years later. (Per instructions, I submitted this to MediaWiki Phabricator, but I was referred back to here, so perhaps the VP(T) instructions should be edited, but that would be a separate issue.) Nick Levinson (talk) 00:26, 8 April 2018 (UTC)
It was only on browsers identifying as "desktop", and not spiking on mobile. So likely the result of one computer gone crazy reloading, accidentally or deliberately. Maybe some person testing some code let it get out of control in a loop. Or perhaps something slipped on a keyboard,"r" and "ctrl" keys. Graeme Bartlett (talk) 00:47, 11 April 2018 (UTC)
Just to add to what I said earlier, even when I change the x value "'78–'99" to "1978–1999", the issue remains. The bar appears in the correct spot if I simply say "1978" or "1999", but that's not what I want it to say. Master of Time(talk)23:32, 14 April 2018 (UTC)
Looking at the documentation at {{Graph:Chart}} and mw:Extension:Graph I can't see any notes on how the x values work. One of the demos starts with an example of a simple bar chart. Changing the first x value from 1 to 9 makes the item for 9 appear after the others. That is, the graph extension sorts the (x,y) pairs by the x value, putting strings after numbers. Yurik could explain what is going on. Johnuniq (talk) 00:54, 15 April 2018 (UTC)
I just came over here to report the same thing myself - several deletions I just attempted gave the same error "A database query error has occurred. This may indicate a bug in the software." and "[Wsx78gpAMF0AAEo1QGwAAABE] 2018-04-10 08:56:15: Fatal exception of type "Wikimedia\Rdbms\DBQueryTimeoutError"" (with the "Wsx78gpAMF0AAEo1QGwAAABE" part different in each case). Boing! said Zebedee (talk) 08:58, 10 April 2018 (UTC)
(edit conflict) Was in the process of writing out the exact same message. I've tried deleting different namespaces and with different reasons but to no avail. I'm getting the exact same error message (except the string at the start which is different every time). Anarchyte (work | talk)09:00, 10 April 2018 (UTC)
It seems some regular database maintenance, that should normally have no impact on edits made deletes fail for 40 minutes. Database administrators will research why this happened and put measures in place to prevent from happening again. Apologies for the problems caused, this was a really bad thing to happen. We will share our findings here, for the time being, you may want to review edits from 9:40 to 10:19 UTCCorrection: 8:40 to 9:19 UTC in case some to-delete pages/editions were missed. --JCrespo (WMF) (talk) 09:36, 10 April 2018 (UTC)
I once discovered the hard way that if you start an OPTIMIZE TABLE on a mysql database, and then kill it with CTRL+C, it leaves the table in an unusable state. As Boing! said, things break, it's just life, and thanks for the prompt investigation. Ritchie333(talk)(cont)09:53, 10 April 2018 (UTC)
In a decades-long career as a user of large computer systems in industry, I always found it hard to distinguish between successful hacking and failed maintenance, when maintainers don’t manage to test the effects of their changes, or to note problems and notify users appropriately that the problem has been detected and is being fixed. Edison (talk) 14:30, 16 April 2018 (UTC)
Reminds me of another problem with portals, that often people copy from articles without attribution. Anyhow, I've fixed it by reverting Legacypac's edit here. Also legacypac, atleast use {{#lsth:Article}} to transclude only the lead - the entire article shouldn't be dumped in the front of a portal lead Galobtter (pingó mió) 08:18, 16 April 2018 (UTC)
Was not aware of that bit of code. I did not like the content as it was presented. I'd rather redirect the entire portal to the associated topic Legacypac (talk) 15:11, 16 April 2018 (UTC)
Latest tech news from the Wikimedia technical community. Please tell other users about these changes. Not all changes will affect you. Translations are available.
Changes later this week
Profiling statistics for an abuse filter tell how often edits match the filter. The statistics for the abuse filters were reset after 10000 actions. Wikis can now decide to reset it more or less often. They can file a phabricator task to do so. [1]
Abuse filters will now treat integers and floats more precisely. For example, 5/2 was rounded down to 2 but will now be 2.5 and 2*4 will be the integer 8 and not the floating-point number 8.0. Division values are the only ones changed. For the rest only strict comparisons (=== and !==) will be affected leaving the values unchanged. [2][3]
The new version of MediaWiki will be on test wikis and MediaWiki.org from 17 April. It will be on non-Wikipedia wikis and some Wikipedias from 18 April. It will be on all wikis from 19 April (calendar).
Meetings
You can join the next meeting with the Editing team. During the meeting, you can tell developers which bugs you think are the most important. The meeting will be on 17 April at 18:30 (UTC). See how to join.
You can join the technical advice meeting on IRC. During the meeting, volunteer developers can ask for advice. The meeting will be on 18 April at 15:00 (UTC). See how to join.
Future changes
The new PDF renderer could not create PDFs from books. Books are in this case collections of pages on a Wikimedia wiki. PediaPress will take over development of the books-to-PDF function. [4]
Pywikibot will no longer support Python 2.7.2 and 2.7.3. [5]
This is proposed quite frequently, but there are scalability issues that prevent it from being rolled out more fully. You may wish to see the discussion from last month about enabling two-factor authentication for more information. --Deskana (talk) 16:35, 12 April 2018 (UTC)
The main issue that I see is that the current implementation has no backup. With most 2FA systems, there are other ways to get in to your account if you lose your authenticator software or the seed code (Google has a system that can call you with a login code, banks have customer service people who can ask you security questions or, if worst comes to worst, snail mail you to verify your identity, etc.). Unfortunately, most of those methods wouldn't work for Wikipedia since it would require the WMF storing your personal information, which they don't want to do (Wikipedia could use email as a second factor, but there are security issues with email). Instead, you have to open a bug report on Phabricator, wait for a developer to see it, hope that your IP hasn't changed so a checkuser can verify your ID, and then the developer has to manually edit the database to remove 2FA from your account--Ahecht (TALK PAGE) 22:57, 12 April 2018 (UTC)
General question: does anyone know of any particular reasons why making the MonoBook skin responsive for mobile (sidebar moved to bottom, navigation in general a bit collapsed, has proper scaling, looks like this, screenshots: phab:F16577440, phab:F16577439) would be a bad idea? As much as I'd argue that all skins should just work across devices and scale appropriately, it's come up that some people may not want this. Do folks here expect such to be the case, or something that should be considered a blocker? -— Isarra༆19:02, 14 April 2018 (UTC)
More accurate would be to say I already... did it, really, though it's a fairly small one. The patch exists and just needs some fixes before we merge it, but it also makes sense to verify that it's not likely to be a problem in practice beforehand as well. -— Isarra༆21:45, 14 April 2018 (UTC)
I use Timeless presently and if I didn't have at least one sidebar for any medium-size or longer article I would be sad. --Izno (talk) 20:52, 14 April 2018 (UTC)
It'd still have the sidebar at normal resolutions, as the desktop layout is unchanged. The linked example has it live on my personal wiki, so if you make it narrow you can see it collapses into the mobile layout at 850px. -— Isarra༆21:45, 14 April 2018 (UTC)
Right, but some people either a) edit on mobile or b) edit with a half-width screen. Actually, this is something about Timeless that bothers me come to think of it. :) --Izno (talk) 22:11, 14 April 2018 (UTC)
(And it bothers me more because I only have access to the tools in one place--the top... maybe I don't need them? --Izno (talk) 22:14, 14 April 2018 (UTC))
We should speak more about this, because it sounds like you might be hitting on something that could well be quite important in general for this stuff. What's the issue, exactly, that's been coming up? That it's just too wide that the layout changes? Is it causing problems specifically for the editing action, or just usage in general as an editor? -— Isarra༆12:44, 17 April 2018 (UTC)
Really the idea here is to make ALL the skins work across devices. Timeless, MonoBook, Vector... -— Isarra༆21:46, 14 April 2018 (UTC)
You probably got pinged because Atsme's user page got transcluded by accident. Probably because of the image with this caption: ""In Memorandum: An Editor's Hand after Scrolling down EEng's Talk Page." User:Eman235/talk 5:40 am, 29 May 2017 (UTC−5)". Anarchyte (work | talk)06:30, 14 April 2018 (UTC)
@Eman235: Yes, Mr. Daniel Plainview (talk·contribs) wrote {{user:Atsme|Atsme}} presumably as a typo for either {{user|Atsme|Atsme}} or [[user:Atsme|Atsme]]. The first is only one character different, the second four; but whichever it was intended, it makes all the world of difference. So, everybody whose user page is linked from user:Atsme will have been notified for the same edit. --Redrose64 🌹 (talk) 09:11, 14 April 2018 (UTC)
LOL well that's a smidgen embarrassing. Sorry about the confusion. Still kind of trying to work out the best way to ping other users without the colon added by default. Mr. Daniel Plainview (talk) 15:08, 16 April 2018 (UTC)
It's pretty janky, but I did write a user script to help with this; it puts a drop-down below the edit box that gives you an option to insert a ping for you for anyone whose name is already linked in the section. Installation is to add mw.loader.load("/w/index.php?title=User:Writ Keeper/Scripts/autoping.js&action=raw&ctype=text/javascript"); to your common.js page. Like I said, it's a little janky, though; the user detection isn't perfect, and it doesn't always insert the ping in the right place. Still might be of use though. Writ Keeper⚇♔14:07, 17 April 2018 (UTC)
@Writ Keeper: I'm having trouble understanding how it works... presumably it scans for usernames and encloses these in [[User:username|username]]. It's something to do with a variable called this but I can't find where that is set, nor even declared (I guess it's a string?); the first use is as this.selectedIndex but even that isn't set up anywhere. --Redrose64 🌹 (talk) 18:09, 17 April 2018 (UTC)
@Redrose64: Not quite. What it does is it scans for usernames that are already linked in the section (as identified by the regex /\[\[User:([^|#\/\]]+)/g), which will usually be either people's pings or signatures. It then puts those names in a dropdown menu right above the edit summary box; when one of the usernames in the dropdown is chosen, the script will insert a ping, in the form @[[User:<username>|<username>]]: into the edit box at the cursor's last location. (If the cursor hasn't been in the editbox yet, or if the edit box has lost focus, the placement can still get pretty wonky.) The this that you see is the dropdown menu itself, such that this.selectedIndex is a reference to the dropdown entry chosen (i.e. the index that was selected). It's a Java-style this keyword; it refers to the object currently being operated on. In this context, it refers to the DOM object from the jQuery selector $("#pingMenu"), since the code is in that object's "change" event callback. (I spend a loooot of time in the jQuery API.) Writ Keeper⚇♔18:49, 17 April 2018 (UTC)
Writ Keeper: In which case, I don't see the point. Notifications are generated by links to a user page, as here. Changing such links to templates will make no difference at all as to whether a notification is sent or not. --Redrose64 🌹 (talk) 19:29, 17 April 2018 (UTC)
@Redrose64: I...think you still might not get how it works? The point is that you don't have to type out a user wiki link or remember a template format or (especially) copy and paste a user who has wacky characters in their name, you just click on a dropdown and the script writes it out for you. It doesn't change anything to anything else, it inserts something. Like, if I was too lazy to type out Redrose64, but wanted to ping you, I just choose the dropdown and the script writes out: @[[User:Redrose64|Redrose64]]: Writ Keeper⚇♔20:01, 17 April 2018 (UTC)
This did not notify me. It is fairly well known that the user link and signature must be added in the same post, but it is less well understood that it must all be done be a new post, not as an amendment to an existing post. That is also one of the reasons that I am sceptical about your script. If you link a few examples of where your script has been used, I should be able to determine if a notification would have been sent. --Redrose64 🌹 (talk) 20:43, 17 April 2018 (UTC)
@Redrose64: Ha, when he wrote about not wanting a colon, I interpreted that as not having to type one, like he did above for the accidental userpage transclusion. LOL. Home Lander (talk) 13:57, 17 April 2018 (UTC)
Still down. Petscan is an invaluable maintenance tool: what is the best way to contact the WMF engineering team about this? -- The Anome (talk) 09:52, 18 April 2018 (UTC)
It's back now. It would still be useful to know where to report problems in future; and interesting to know why it went down for so long. William Avery (talk) 11:49, 18 April 2018 (UTC)
Well some searching often gives you the answer. In this case Wikipedia:PetScan comes up quite quickly, which gives you all the information you need. —TheDJ (talk • contribs)
@TheDJ: - I've got to disagree with you there. While there is reference that this is run by "Wikimedia Foundation's cloud computing environment", the "issues" link goes to the third party bitbucket.org reporting service. That page also says it is an "external tool", and yes it is not part of core, but if it is operated by WMF it is in the enterprise scope. So from just reading the page you referenced, this appears that there is some paid support from the foundation that perhaps should be engaged via phabricator - but like with what I think are too many services, there is no clear direction on how users should intake trouble reports and what to expect once it is submitted; or if after-action reports are going to be produced. This type of confusion is unfortunately common with tooling, especially knowing what aspects of outages are expected to be handled under an SLO by paid staff and which aspects are dependent on volunteers. — xaosfluxTalk14:58, 18 April 2018 (UTC)
Note, I'm really just trying to think of a way forward on issue reporting - perhaps we need some big templates (report an issue) that can be applied to the most popular tools - and they can link to or include specific directions for the bigger apps. — xaosfluxTalk15:04, 18 April 2018 (UTC)
And most people would have no way to know if something isn't working because an underlying system is malfunctioning. — xaosfluxTalk15:56, 18 April 2018 (UTC)
Replication lag should now be fixed on 2 out of 3 servers. We try to keep the web server as updated as possible, but the analytics ones, due to long running queries, sometimes it is simply not possible. If queries are short, you can use the web server. Due to the current topology and the nature of the changes (ROW based replication is more reliable, but also less flexible with schema changes), sometimes replication lag has to happen. The smaller tables that will be introduced will minimize in the future the schema change time. Also, there is ongoing plans to make redundant, not only the servers, as there is now, but the replication channel T190704, that should also minimize (but not make fully disappear) lagging due to maintenance- we are only waiting on new hardware to arrive for it. --JCrespo (WMF) (talk) 07:52, 19 April 2018 (UTC)
Invisible stub category?
For several days the redirect Romani people in Slovenia has been appearing in the list at Category:Stubs. There is no visible stub template or category in the redirect "article", nor anything in the history of the redirect which seems to show either of these. Articles from which the {{stub}} template (or the incorrect Category:Stubs) is removed (usually replaced by a specific stub template) normally disappear from the category listing instantly. Any ideas? I'd just like to stop it appearing when I'm stub-sorting. PamD07:48, 19 April 2018 (UTC)
It's not there now. That may be because the article was, coincidentally, just edited and that would have purged cached information. Some background is at WP:PURGE but an easy way to purge would be to edit the article then click Publish without making any change and with no edit summary. Nothing will be recorded in the edit history but categories will be updated. Johnuniq (talk) 08:18, 19 April 2018 (UTC)
A purge updates the page itself but not category pages and WhatLinksHere. A null edit updates all of it. Johnuniq described a null edit but called it a purge. PrimeHunter (talk)
Bluelink RGB color
Can someone link to the (formal) definition of the wikilink RGB colors (for example, the blue of the in blue wikilink)? I want to do contrast checks. - DePiep (talk) 09:58, 19 April 2018 (UTC)
The Fb cl header navbar template that I use on many of the pages I've created has stopped working, in spite of the fact that it has not yet been deleted. Why is this happening? Also, if it is deleted, what would be the easiest way to replace my tables? Birdsgeek (talk)
@Birdsgeek: I moved the <noinclude> code to in front of the TfD tag and purged the cache on one of the articles on which it is used, and it seems to have made a difference. Is it behaving normally now? Home Lander (talk) 15:18, 19 April 2018 (UTC)
Hi. About WP:USERNAME. Whatever their fancy colors styling duh, I run into this problem: I cannot copy/paste their username. So I propose: Username must be c/p friendly (css wise). - DePiep (talk) —Preceding undated comment added 20:15, 15 April 2018 (UTC)
What are you referring to? "fancy colors styling" sounds like your issue is with customized signatures and not with the username policy. If you think a signature should display the real username so you can copy-paste from the rendered signature instead of the mandatory wikilink in the source then you can post a suggestion to Wikipedia talk:Signatures. If you see a signature without the username in the source then point the user to WP:SIGLINK. If you enable "Navigation popups" at Special:Preferences#mw-prefsection-gadgets then you can copy-paste the username from a popup when you hover over a signature. PrimeHunter (talk) 21:51, 15 April 2018 (UTC)
I say: Similar to {{DISPLAYTITLE:}}: the username in a sign must be copy/paste-able whatever the formatting (styling). Sure that could require a guideline change. - DePiep (talk) 20:44, 17 April 2018 (UTC)
I doubt there would be a phab id; it's not a technical problem, and I don't think there could be a technical solution, short of removing the ability to customize signatures. In what context do you need to copy and paste a user's name?Writ Keeper⚇♔21:12, 17 April 2018 (UTC)
hmm, in communication (think talkpage). It helps when the reply-to name is the same as the post-sign-name (duh). For *all* readers of a discussion. Today, one cannot trust searching a sign-name (because the user-name may be different & hidden). - DePiep (talk) 21:19, 17 April 2018 (UTC)
@DePiep: Please clarify. Is your complaint (1) that some rendered signatures do not show the username, or (2) that the usernames in some rendered signatures are not c/p-able? ―Mandruss☎21:55, 17 April 2018 (UTC)
1. Why must I? 2. Does it help reading, searching the talkpage? 3. Username can be hidden in irrelevant but disturbing code. (see source code from example #1; this is a simple example because it has no color & talkpage mixups). My question: did you yourself never met this situation? - DePiep (talk) 22:50, 17 April 2018 (UTC)
Andy gives the username following the userpage link. Hey I agree with you in principle, I think the signature p&g should be improved and enforced. As a practical matter in the current environment, established editors are not going to change their sigs unless given a choice between that and a block, which is never going to happen. You might have some luck with a few new editors, the ones who haven't been here long enough to know that the community doesn't really care about compliance with signature p&g. Once they learn better, I suspect many of them will simply change back to what they had before. I think any improvement to the p&g is largely a waste of time. ―Mandruss☎22:54, 17 April 2018 (UTC)
So now you say like: 'established users won't give up their privilege'. Sure, if that is how Wikipedia works. I repeat: usernames in talkpages should be open&clear this way, because it is a website. - DePiep (talk) 18:14, 19 April 2018 (UTC)
When you hit submit it's just redirecting to that wiki's Special:AutoblockList. The old tool didn't actually work anyway, at least at the time we retired it. What is the expected behaviour? Would the username pertain to the originally blocked user, or the blocking admin? For privacy reasons we can't reveal the account/IP that was autoblocked. — MusikAnimaltalk19:50, 19 April 2018 (UTC)
Mapping content of different language articles to English
I find that Russian articles frequently deviate from English articles - to a point where false or misleading information is disseminated.
For example, Echinacea (corn flower) has no reference in English that it may be used to isolate/purge radio-isotopes from the human body, yet this claim is made in the Russian version.
So, there should be a simple translation program used (Google) to compare languages of the same article for words used that are not within the English (correct and trustworthy) version. A red flag would be raised and the article should be reviewed by a competent language volunteer to assure the reliability and accuracy of foreign language entries.
Thanks! — Preceding unsigned comment added by NM Remote (talk • contribs) 05:13, 12 April 2018 (UTC)
I'm not surprised by the claim; it's one of hundreds of miraculous-seeming claims for that plant. What you wouldn't find is any WP:MEDRS-qualifying sources that support such a claim. User:Jytdog or User:Alexbrn, do you know any editors who can read Russian and might be interested in looking into this report? The idea of being able to automatically identify significant differences might be a good one (e.g., it could help us find new content, so we could expand our articles), but in the short-term, it's possible that we could find someone to review that article and make sure that it meets that community's standards. WhatamIdoing (talk) 23:24, 19 April 2018 (UTC)
Thanks for the invitation but I don't read Russian and my hands are overfull dealing with woo-pushers here. Sorry. Jytdog (talk) 23:30, 19 April 2018 (UTC)
commons.wikimedia.org Log in problem
"There seems to be a problem with your login session; this action has been canceled as a precaution against session hijacking. Please resubmit the form."
Hi. I have a few requests relating to maintaining good articles and was hoping I could get some help from the technically minded folk here.
Update articles that have been passed as a good article, but do not have the {{good article}} template. Legobot does this after a review, but I have come across a few that do not have it.[7][8][9]
Removing the {{good article}} template from articles that do not have it. This occurs usually when editors start a new article and copy paste from an existing good one.
Updating pages listed at Wikipedia:Good articles/all. This means removing ones that are no longer good articles and adding those that reviewers forgot to add.
We have Category:Good articles which lists article pages with the GA icon and Category:Wikipedia good articles which lists the talk pages with a passed review {{GA}}. There is a discrepancy of 25 between the two. There are about 2000 entries missing from Wikipedia:Good articles/all so that is where most of the work will lie. If it was possible to do it in petscan or some other tool that would be great as I could then run it myself in the future. Cheers AIRcorn(talk)22:27, 19 April 2018 (UTC)
@JJMC89: Thanks a lot. That's perfect. I have AWB, but didn't even know about list comparer. I should be able to update the mismatches page myself in the future. AIRcorn(talk)05:24, 20 April 2018 (UTC)
Sometimes References are split up in two or more subsections, each with its own sub-heading, say: ===Notes=== and ==-Citations===. Some editors happen not to like to have these sub-heading visible in the TOC.
Which is the correct way to proceed:
{{TOC limit|2}}, which of course affects all subsections of the same level,
or
{{fakeheader|level=3|Notes}}, which has a punctual scope, but maybe also other technical drawbacks?
No, the response is DONT use Pseudo headings (like the guidelines tell you). Secondly don't fret so much about what goes into the ToC, it's an automated system, you can't control everything, just because you want to (which is why it's easier to use than LaTeX). And then be prepared for most people just doing whatever they want anyways. —TheDJ (talk • contribs) 22:40, 19 April 2018 (UTC)
Okay, that is the most sensible thing I've read on Wikipedia in a very long time. Thank you for the smile, TheDJ - not to mention the reality check. Risker (talk) 00:38, 20 April 2018 (UTC)
@FR30799386: the problem is not the inputbox, it's the divs around it. You want a responsive gridlayout, for which you can best use flexbox layouting or display:table-cell —TheDJ (talk • contribs) 22:04, 20 April 2018 (UTC)
Checking "watch" box in edit action box doesn't put article on watch list
For some time now, my checking the "Watch this page" box in the edit action box doesn't accomplish what it used to, and I have to click on the star in the top-of-page tab bar to put an unwatched page on my watchlist. I'm using Firefox 59.0.2 (32-bit) under Windows 7. Dhtwiki (talk) 23:19, 19 April 2018 (UTC)
I have Firefox 59.0.2 (32-bit), but Windows 10. Clicking the "watch" box in the edit action box just worked for me. Are you using NoScript? If so, try disabling NoScript, and then checking the watch box. The latest version of NoScript interferes with much that it used not to. In my case, when I need to, I disable NoScript, change what I need and save, then re-check NoScript. — Maile (talk) 23:27, 19 April 2018 (UTC)
I used to, but no longer, use NoScript. However, I have been using Private Browsing, which may be the cause. I'll look into that. Evidently, the feature is working for others. Thanks for the reply. Dhtwiki (talk) 03:39, 20 April 2018 (UTC)
@Dhtwiki: I use Opera 36, no private browsing, no noscript. If I edit a page (or section) and check "Watch this page" and save, the top of the page doesn't update the watch star (tab if you use MonoBook skin); but if I reload the page (or go to page history) and check the tabs at the top, it shows as being watched; it also gets listed if I go straight to my watchlist after saving. To my mind there is some sort of caching delay such that an older version of the page is being displayed. --Redrose64 🌹 (talk) 07:02, 20 April 2018 (UTC)
I had a couple of user-warning templates to give out tonight, the usual cause of my putting new pages on my watchlist. I checked the box, then when the watch star showed white after I published my edit, I refreshed the page and the star showed blue. So, mystery sort-of solved. Thank you for pointing out the possibility of, what?, more aggressive caching on Firefox's part? Dhtwiki (talk) 03:56, 21 April 2018 (UTC)
links
Recently, when a cursor is over a link, there is a photo and a summary of the article. This is particularly bothersome when looking at an editor's contributions. It is also not helpful with articles as it can block a large portion of the text of the article.
I'm getting an error of: PHP fatal error: Argument 1 passed to EchoEventPresentationModel::getTruncatedTitleText() must be an instance of Title, null given when I visit this page. It's the same on desktop and mobile. The buttons in the toolbar work fine SmartSE (talk) 22:51, 18 April 2018 (UTC)
In its present form, Special:RandomInCategory is not very useful because it does not support subcategories. If, for example, you want a random scientist, you could try typing in Scientists for the category; but Category:Scientists has 25 subcategories and 17 articles, all of which are there because someone did not bother to classify them properly. Not a very representative sample! Are there any plans to expand its functionality? If not, is there an alternative way of finding random articles? RockMagnetist(talk)19:10, 20 April 2018 (UTC)
I had looked at randomlink.js and didn't think that it would be useful for finding articles in a category; but now it occurs to me that one could run CatScan V2.0β and then use randomlink.js on the output. RockMagnetist(talk)16:01, 21 April 2018 (UTC)
API calls "by hand" within Pywikibot: no public-facing method... why?
# TODO: expand support to other parameters of action=parse?defget_parsed_page(self,page):"""Retrieve parsed text of the page using action=parse."""req=self._simple_request(action='parse',page=page)data=req.submit()assert'parse'indata,"API parse response lacks 'parse' key"assert'text'indata['parse'],"API parse response lacks 'text' key"parsed_text=data['parse']['text']['*']returnparsed_text
The TODO note lets me think that (for now) PWB implements only a small subset of the action=parse API, and in particular not what I wanted, but please let me know if that is wrong. However, that snippet hints at low-level methods that allow a dirty hack to make arbitrary API calls; the output of the API's https://en.wikipedia.org/w/api.php?action=parse&prop=sections&oldid=837538913 can be obtained via
importpywikibot# Define subfunction for arbitrary API callsdefmanual_API_call(site,**kwargs):"""Make API request by giving parameters 'by hand'."""request=site._simple_request(**kwargs)returnrequest.submit()# Pass parameters by handdeffind_sections(site,pagerevision):params={'action':'parse','prop':'sections','format':'json','formatversion':2,'oldid':pagerevision,}returnmanual_API_call(site,**params)print(find_sections(pywikibot.Site(),837538913))# same output as web request https://en.wikipedia.org/w/api.php?action=parse&prop=sections&oldid=837538913
Is there a PWB equivalent to the low-level API call manual_API_call defined above? If no, why not (since all the machinery is in place)? Defining it myself does not look great to me, since we are calling a "private" method of the pywikibot.APISite class, and this looks like bad practice to me. For instance, if _simple_request gets renamed or changed in the future, the above snippet will stop working, but the rename/change itself will not be marked "breaking" by PWB devs (since that is not a public method). TigraanClick here to contact me15:28, 21 April 2018 (UTC)
@Tigraan: I don't know if there is an equivalent to your manual_API_call, but you can avoid using the private method. Add frompywikibot.data.apiimportRequest after line 1, and change line 5 to request=Request.create_simple(site,**kwargs). — JJMC89 (T·C) 18:29, 21 April 2018 (UTC)
I may have missed some doc because from what I can read I followed all the steps. Details with ssh sessions on Toolforge etc:
Simple session - login on Toolforge as the tool account, run login.main(), abort (ctrl-C) when credentials are requested
Linux tools-bastion-03 3.13.0-139-generic #188-Ubuntu SMP Tue Jan 9 14:43:09 UTC 2018 x86_64
Ubuntu 14.04.5 LTS
tools-bastion-03 is a Tool Labs bastion (toollabs::bastion)
======================================================================
_______ _____ _____ _______ _____ ______ ______ _______
| | | | | | |______ | | |_____/ | ____ |______
| |_____| |_____| |_____ | |_____| | \_ |_____| |______
======================================================================
This is a server of the tools Cloud VPS project, the home of community
managed bots, webservices, and tools supporting the Wikimedia movement.
Use of this system is subject to the Toolforge Terms of Use,
Code of Conduct, and Privacy Policies:
- https://tools.wmflabs.org/?Rules
General guidance and help can be found at:
- https://tools.wmflabs.org/?Help
The last Puppet run was at Sun Apr 15 19:12:08 UTC 2018 (1 minutes ago).
Last login: Sun Apr 15 18:47:04 2018 from 213-245-111-42.rev.numericable.fr
tigraan@tools-bastion-03:~$ become muninnbot
tools.muninnbot@tools-bastion-03:~$ python
Python 2.7.6 (default, Nov 23 2017, 15:49:48)
[GCC 4.8.4] on linux2
Type "help", "copyright", "credits" or "license" for more information.
>>> from scripts import login
>>> login.main()
Password for user Muninnbot on wikipedia:en (no characters will be shown): Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/data/project/shared/pywikipedia/core/scripts/login.py", line 179, in main
site.login(sysop, autocreate=autocreate)
File "/data/project/shared/pywikipedia/core/pywikibot/site.py", line 2090, in login
if loginMan.login(retry=True, autocreate=autocreate):
File "/data/project/shared/pywikipedia/core/pywikibot/login.py", line 308, in login
password=True)
File "/data/project/shared/pywikipedia/core/pywikibot/bot.py", line 407, in input
data = ui.input(question, password=password, default=default, force=force)
File "/data/project/shared/pywikipedia/core/pywikibot/userinterfaces/terminal_interface_base.py", line 292, in input
text = self._input_reraise_cntl_c(password)
File "/data/project/shared/pywikipedia/core/pywikibot/userinterfaces/terminal_interface_base.py", line 310, in _input_reraise_cntl_c
raise QuitKeyboardInterrupt()
pywikibot.bot_choice.QuitKeyboardInterrupt
Same problem, but with a non-interactive session (so it aborts because stdin is EOF)
As the tool account, I crontab'd to make the following Python test run on the grid, i.e. in the same conditions as the bot should run:
#!/usr/bin/env python3# -*- coding: utf-8 -*-importdatetimefromscriptsimportloginimportsysprint(datetime.datetime.utcnow(),' - test executed')print('Python PATH is:',sys.path)print('I will attempt to login via PWB')login.main()print('login.main() did not freeze')
results in the following cron-0.out file:
2018-04-15 18:50:22.191869 - test executed
Python PATH is: ['/mnt/nfs/labstore-secondary-tools-project/muninnbot/Teahouse-bot/scripts', '/data/project/shared/pywikipedia/core', '/data/project/shared/pywikipedia/core/externals/httplib2', '/shared/pywikipedia/core/scripts', '/usr/lib/python3.4', '/usr/lib/python3.4/plat-x86_64-linux-gnu', '/usr/lib/python3.4/lib-dynload', '/usr/local/lib/python3.4/dist-packages', '/usr/lib/python3/dist-packages']
I will attempt to login via PWB
<class 'EOFError'>
(The script's last line is never reached)
...and the following cron-0.err file:
Password for user Muninnbot on wikipedia:en (no characters will be shown): WARNING: /usr/lib/python3.4/getpass.py:92: GetPassWarning: Can not control echo on the terminal.
passwd = fallback_getpass(prompt, stream)
Warning: Password input may be echoed.
Traceback (most recent call last):
File "/usr/lib/python3.4/getpass.py", line 70, in unix_getpass
old = termios.tcgetattr(fd) # a copy to save
termios.error: (25, 'Inappropriate ioctl for device')
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
File "/data/project/muninnbot/Teahouse-bot/scripts/test.py", line 12, in <module>
login.main()
File "/data/project/shared/pywikipedia/core/scripts/login.py", line 179, in main
site.login(sysop, autocreate=autocreate)
File "/data/project/shared/pywikipedia/core/pywikibot/site.py", line 2090, in login
if loginMan.login(retry=True, autocreate=autocreate):
File "/data/project/shared/pywikipedia/core/pywikibot/login.py", line 308, in login
password=True)
File "/data/project/shared/pywikipedia/core/pywikibot/bot.py", line 407, in input
data = ui.input(question, password=password, default=default, force=force)
File "/data/project/shared/pywikipedia/core/pywikibot/userinterfaces/terminal_interface_base.py", line 292, in input
text = self._input_reraise_cntl_c(password)
File "/data/project/shared/pywikipedia/core/pywikibot/userinterfaces/terminal_interface_base.py", line 306, in _input_reraise_cntl_c
text = getpass.getpass('')
File "/usr/lib/python3.4/getpass.py", line 92, in unix_getpass
passwd = fallback_getpass(prompt, stream)
File "/usr/lib/python3.4/getpass.py", line 127, in fallback_getpass
return _raw_input(prompt, stream)
File "/usr/lib/python3.4/getpass.py", line 149, in _raw_input
raise EOFError
EOFError
CRITICAL: Closing network session.
I believe I have a correct setup for my password file
...i.e., I have a password file in the correct format, and user-config.py points to it:
(Password redacted of course - but yes, I am sure I put the correct password in the file, I manage to log on en-wp manually by copy-pasting from the file)
@JJMC89: the exact same problem occurs with your code snippet (I had already tried using the pywikibot.Site().login() but did not keep the logs so I wanted to make sure to execute your exact snippet). It asks for the password (so it is a no-go in non-interactive session), and the stack trace points to similar places.
details for the stack trace but it is similar to above
Linux tools-bastion-03 3.13.0-139-generic #188-Ubuntu SMP Tue Jan 9 14:43:09 UTC 2018 x86_64
Ubuntu 14.04.5 LTS
tools-bastion-03 is a Tool Labs bastion (toollabs::bastion)
======================================================================
_______ _____ _____ _______ _____ ______ ______ _______
| | | | | | |______ | | |_____/ | ____ |______
| |_____| |_____| |_____ | |_____| | \_ |_____| |______
======================================================================
This is a server of the tools Cloud VPS project, the home of community
managed bots, webservices, and tools supporting the Wikimedia movement.
Use of this system is subject to the Toolforge Terms of Use,
Code of Conduct, and Privacy Policies:
- https://tools.wmflabs.org/?Rules
General guidance and help can be found at:
- https://tools.wmflabs.org/?Help
The last Puppet run was at Mon Apr 16 17:42:40 UTC 2018 (21 minutes ago).
Last login: Sun Apr 15 19:42:21 2018 from 213-245-111-42.rev.numericable.fr
tigraan@tools-bastion-03:~$ become muninnbot
tools.muninnbot@tools-bastion-03:~$ python3
Python 3.4.3 (default, Nov 28 2017, 16:41:13)
[GCC 4.8.4] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import pywikibot
>>> site = pywikibot.Site()
>>> site
APISite("en", "wikipedia")
>>> site.login()
Password for user Muninnbot on wikipedia:en (no characters will be shown): Traceback (most recent call last):
File "/data/project/shared/pywikipedia/core/pywikibot/userinterfaces/terminal_interface_base.py", line 306, in _input_reraise_cntl_c
text = getpass.getpass('')
File "/usr/lib/python3.4/getpass.py", line 78, in unix_getpass
passwd = _raw_input(prompt, stream, input=input)
File "/usr/lib/python3.4/getpass.py", line 147, in _raw_input
line = input.readline()
KeyboardInterrupt
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/data/project/shared/pywikipedia/core/pywikibot/site.py", line 2090, in login
if loginMan.login(retry=True, autocreate=autocreate):
File "/data/project/shared/pywikipedia/core/pywikibot/login.py", line 308, in login
password=True)
File "/data/project/shared/pywikipedia/core/pywikibot/bot.py", line 407, in input
data = ui.input(question, password=password, default=default, force=force)
File "/data/project/shared/pywikipedia/core/pywikibot/userinterfaces/terminal_interface_base.py", line 292, in input
text = self._input_reraise_cntl_c(password)
File "/data/project/shared/pywikipedia/core/pywikibot/userinterfaces/terminal_interface_base.py", line 310, in _input_reraise_cntl_c
raise QuitKeyboardInterrupt()
pywikibot.bot_choice.QuitKeyboardInterrupt
>>> site.logged_in()
False
@Tigraan: My guess is that you're using the password for that account that you use for the web interface or the file permissions of the password file aren't correct. You should use OAuth (or BotBasswords). If you add me as a maintainer, I can poke around. — JJMC89 (T·C) 02:52, 17 April 2018 (UTC)
@JJMC89: File permissions allow reading (see "I believe I have the correct setup for the password file" in the OP), unless Pywikibot does something funky (e.g. run from another account) but I do use the same password as for the web interface. Is that wrong? I will try OAuth instead in any case. Will keep you updated. TigraanClick here to contact me08:22, 17 April 2018 (UTC)
@Tigraan: The passwords file in my directory is 700 (executable). It doesn't use another account, but there have been a number of breaking changes to the API action=login. Basically Pywikibot must use OAuth or BotPasswords instead of the username/password used for the web interface (interactive login). — JJMC89 (T·C) 14:29, 17 April 2018 (UTC)
On List of songs in Guitar Hero, the reference list at the bottom of the page is numbered differently than the in-line reference tags (it skips #7). I think the cutoff has something to do with the group Note, and I'm not sure if this error has come because of the new-looking tooltips. For some odd reason I can't seem to recreate this on any page other than other Guitar Hero songs lists. Hameltion (talk, contribs) 18:38, 21 April 2018 (UTC)
No, PrimeHunter, I think you are confusing defining LST blocks via template (not possible as you have proven) and calling LST via template, which there is no reason shouldn't be allowed by template. {{3x|p}}ery (talk) 14:19, 21 April 2018 (UTC)
returns "String Module Error: Match not found". If your goal is to pass the article's title using unnamed parameter 1, Sandbox163 does this, if you want a regular expression that always matches the lead text excluding Twinkle tags, infoboxs, and the likes, that is proving to be a difficult undertaking. I am going to continue testing for a solution. I wish you the best in the mean time.--John Cline (talk) 23:50, 21 April 2018 (UTC)
See my sandbox for a LUA module that might lead you to a solution. The module fetches everything from the beginning of an article to the first == section header. This works ok for Aviation because all of the hats are wrapped in <noinclude>...</noinclude> tags; try another article and the results might not be what you want.
I tried something sort of similar to your ".+\n(.-'''.+)". It sort of worked but left out the lead's image which {{#lsth:Aviation}} includes. When I try your pattern in my module, I get nil when used in mw.ustring.match (content, ".+\n(.-'''.+)"); or the whole article minus the lead image and hat templates when used in string.match (content, ".+\n(.-'''.+)");. It is not clear to me why the mw.ustring.match() matches nothing. I have seen this before. The latter is because there is nothing in your pattern to stop string.match() from matching and consuming characters except that it runs out of characters to match and consume.
When hovering over a blue linked title using popups the snippet rendered is similar to the snippet I believe we're trying to call. Does anyone know how popups retrieves its data (in case it can be used)?--John Cline (talk) 13:02, 22 April 2018 (UTC)
Looks to me like an API call that returns article content in json format. From that, popups creates the html for rendering. See MediaWiki:Gadget-popups.js and look for function loadAPIPreview.
So I was checking my created articles list, and one thing I noticed is that the article Yurika Endō is in it. The thing is: I didn't create the page, the article had recently undergone a history merge with a userspace draft that I wrote last month. What a weird bug: is there a way to make this more accurate? Narutolovehinata5tccsdnew14:12, 22 April 2018 (UTC)
What's happening here is there are two revisions (1, 2) that have a parent revision ID of 0, when this is normally only the case with the first revision. In my opinion it's a bug with how histories are merged, but I can't say for sure. For XTools, I think we can just find the oldest revision with a parent ID of 0, which in this case would correctly attribute the article creation to Juandmarco. I'll look into fixing this soon. — MusikAnimaltalk15:22, 22 April 2018 (UTC)
@Koavf:, it shows as properly vertical for me. Sometimes clearing your browser cache fixes stuff like this. If not, try looking at it in another browser and/or quitting and restarting your browser. BlackcurrantTea (talk) 08:28, 19 April 2018 (UTC)
If you have rollback rights, you might be familiar with this problem: You want to thank someone, but you accidentally click the rollback link – since no confirmation is needed for rollbacks, this can lead to very unpleasant misunderstandings. A top wish from the German Technical Wishlist asked to solve this problem.
NB that this solution isn't just the illustrated confirmation box--it also entails removing the rollback link by default from all list views (page histories, watchlists, user contributions, basically anything that isn't the shown diff page). It's significantly more disruptive than I thought at first glance. Writ Keeper⚇♔13:58, 19 April 2018 (UTC)
Removing the rollback link from all list views is potentially very disruptive - such as when tracking a persistent vandalism-only user. This would not be acceptable, in my opinion. Home Lander (talk) 15:27, 19 April 2018 (UTC)
Agree. Although, maybe having an inline 2nd confirmation (such as the 'thank' feature), would be better than a pop-up? Rehman15:31, 19 April 2018 (UTC)
Well there would be a preference. However indeed - part of why I requested rollback is precisely because it shows up in lists and allows quick reversion there, unlike Twinkle. Galobtter (pingó mió) 15:40, 19 April 2018 (UTC)
Removing the "thank" function would also solve the problem of confusing the "thank" and "rollback" links. How many times do people accidentally click "rollback" without noticing? (If you notice, you just rollback your own edit, which takes 1 second and completely solves the problem). —Kusma (t·c) 09:01, 23 April 2018 (UTC)
Latest tech news from the Wikimedia technical community. Please tell other users about these changes. Not all changes will affect you. Translations are available.
The iOS and Android apps now have synced reading lists. This means you can save articles to a private list that can be seen on your other devices if you use the apps.
The visual editor and the 2017 wikitext ask you to write an edit summary after you press Publish. This button now also shows an ellipsis. This is to show that pressing Publish is not the last step. [11]
Changes later this week
The new version of MediaWiki will be on test wikis and MediaWiki.org from 24 April. It will be on non-Wikipedia wikis and some Wikipedias from 25 April. It will be on all wikis from 26 April (calendar).
Meetings
You can join the next meeting with the Editing team. During the meeting, you can tell developers which bugs you think are the most important. The meeting will be on 24 April at 18:30 (UTC). See how to join.
You can join the technical advice meeting on IRC. During the meeting, volunteer developers can ask for advice. The meeting will be on 25 April at 15:00 (UTC). See how to join.
Future changes
<mapframe> will come to most Wikipedias in May. This means that you can put interactive maps in the articles. Nine Wikipedias that use a strict version of flagged revisions will not get this feature in May. [12]
The rollback function could change. This was a German community request. All editors with rollback rights can leave feedback on the proposed solution. The last day to leave feedback is 4 May (UTC).
Is it technically possible to prevent the sitewide ability to move a page to a subpage of a nonexistent page? If not, can a filter be used to flag such a move when it has occurred?--John Cline (talk) 04:53, 21 April 2018 (UTC)
Probably not, and that wouldn't be a good idea anyway. Talk:/r doesn't currently exist (although it would be perfectly valid to create it), but that should not prevent someone from moving subreddits whose articles are named incorrectly and have talk pages to the correct place. {{3x|p}}ery (talk) 13:31, 21 April 2018 (UTC)
I am primarily concerned with people moving their talk page, along with its history, into an /Archive and then recreating the page anew. It disjoints the page's history and makes searching for earlier edits difficult. Archiving should be accomplished by cut and paste and unfortunately, administrators are the leading culprits in the counter intuitive method of moving pages to /Archive instead. I can find examples later, I am leaving right now, but I've seen this done quite often and don't think it's much of a little known practice.--John Cline (talk) 00:26, 22 April 2018 (UTC)
The docs show that using move to archive a talk page is not recommended. However, moving was a documented method in September 2012 and many admins are old timers from long before then. It is irritating to watch a talk page where moving is used because you end up with many archive pages on your watchlist however it would be awkward getting people to change their ways. Johnuniq (talk) 01:49, 22 April 2018 (UTC)
I don't think that this method would stop people from moving their talk pages anyway. If you move User_talk:Example to User_talk:Example/Archive, then User_talk:Example exists. WhatamIdoing (talk) 17:22, 23 April 2018 (UTC)
Yes, thank you, I agree. It oft happens that one's "brilliant idea" isn't nearly as brilliant as one earlier believed. I know that sinking feeling, well; imagining 3 or 4 more reasons why it wouldn't actually work. Perhaps considering methods that won't work is the path to ideas that will? If so, I am keen in knowing I've done my part, perhaps even well. Thank you again.--John Cline (talk) 18:20, 23 April 2018 (UTC)
Ok, It has corrected, I cleared my temp files, cookies and cache, closed my open tabs, logged out, shut down, rebooted in safe mode (but accidentally didn't safeboot with networking), shut down again, started windows normally, and the top edit tab worked normally for me logged out, I logged in and it continues working. So I don't know what, if anything, I may have done to corrected things. Let me say, I did recently change a thing, and I may soon revert back to my former configuration, but let me ask if anyone knows of 1.1.1.1, known bugs with that service, and whether it could have contributed? I appreciate the help you guys so freely give, I'd have been technically lost long ago without you all! BTW Izno, what did you mean regarding what editor I am using?--John Cline (talk) 17:11, 23 April 2018 (UTC)
Ok, I understand. Over the years, I have experimented with this and that, I probably have things enabled that I never use, and depreciated stuff that don't even work. As far as gadgets, I settled primarily on popups. I have something turned on that opens the edit window if I double tap any part of the page except if a blue link is there, which launches instead; that's about it. I used Huggle until it stopped allowing me to login, and AWB stopped working when I deleted Explorer (foolheatedly). It's probably my machine about to blink out; its been a good one, for a good while. Still wondering if somebody can speak about 1.1.1.1, That is the only thing substantially different that I have recently done. I stumbled on Draft:1.1.1.1, and ended up using the service. Best.--John Cline (talk) 21:10, 23 April 2018 (UTC)
FWIW, using 1.1.1.1 as your dns resolver couldn't possibly cause this problem (Possibly it might cause you to not visit the most geographically close server to you, which could add a couple miliseconds to rendering time, particularly for not logged in users. If cloudflare was malicious (which they obviously aren't), they could in theory use the 1.1.1.1 dns server to prevent you from accessing wikipedia entirely). Bawolff (talk) 09:42, 24 April 2018 (UTC)
Wiki sending to a dangerous site?
−I noticed that many citations on a wikipedia page (https://en.wikipedia.org/wiki/Rajneesh) use a site called webcitation.org. My security software (Norton) is red-alerting this as a "known dangerous site"; I tried linking via a couple of the original URLs (i.e. without the webcitation.org prefix) and they worked fine that way. I started to correct them on the page -- which is a page I didn't even want to spend any of my time on -- but then thought 1) what if this is a problem on other pages as well; 2) what if it's one of those actions the bots do automatically and deliberately, like those I see with "external links modified" and 3) (the bottom line) I don't know nearly enough to mess with this stuff. So I cancelled my changes and came here. I know so little, I'm not even at all sure where I should be bringing this up. Hope someone who knows what they're doing will make the necessary changes, or send this request to the appropriate department, or -- flush this whole note down the toilet. Whatever... alacarte (talk) 22:49, 23 April 2018 (UTC)
Your security software seems to have a false flag set for the URL. WebCite is a known safe archiving service on the Internet, and we make extensive use of it on Wikipedia. Maybe GreenC has some more information. --Izno (talk) 00:42, 24 April 2018 (UTC)
WebCite is the second most common web archiving site on English Wikipedia, behind web.archive.org .. there is nothing inherently wrong with it, but there might be pages archived there that are dangerous which is why Norton is flagging the entire webcite domain since it sees the source URL in the path portion of the webcitation URL. -- GreenC01:10, 24 April 2018 (UTC)
@GreenC, Professor alacarte, and Izno: I've noticed that WebCite tends to directly load some third-party content, like Google Analytics tracking, instead of loading that content from an archived copy (which is what the Wayback Machine and archive.is do). This loading of content could be related to the security software, but I'm not sure. Jc86035 (talk) 11:58, 24 April 2018 (UTC)
Cursor insertion point
In the last month or so, editing has become much more difficult. if I backspace to the front of the line, instead of the cursor moving to the end of the previous line, it moves to a random place in the previous line. Hawkeye7(discuss)22:10, 23 April 2018 (UTC)
Browser if Firefox and the skin is Vector, which I think is the default. I don't have any new beta features switched on. Suppose I move the insert point to after "previous" in the line above, which is wrapped. I then add two blank lines and "xxxx" I then backspace over the xxxx and the blank lines. The cursor winds up in the word "ends". Works okay in safemode though. Hawkeye7(discuss)00:40, 24 April 2018 (UTC)
Is there a reason why cells are no longer highlighted when hovered over in tables using the mw-datatable attribute? Per Help:Table#mw-datatableWhen a cursor hovers over the table, that row over which the cursor is on will be highlighted. Sadly this no longer works and there is no highlighting. 78.18.91.209 (talk) 18:20, 24 April 2018 (UTC)
Maybe it's the OS I'm using. I'm currently viewing Wikipedia in desktop mode on an iPad running iOS 10. Before yesterday whenever I tapped a cell it would be highlighted, yet that no longer happens for some reason. 78.18.91.209 (talk) 10:37, 25 April 2018 (UTC)
How do you transclude a lead only?
I've run into a bit of a mystery...
This works:
{{#section:Donald Trump|Lead text}}
This doesn't work:
{{#section:Physical geography|Lead text}}
<section begin=Lead text />
...
<section end=Lead text />
At Help:Magic words, search for "#section" to find a link to documentation. I hope transclusions like this are rare because it is very unlikely that the same text is really useful in two places, and it is likely that future edits will either break the transclusion or make it inappropriate. Johnuniq (talk) 23:43, 14 April 2018 (UTC)
But it picks up hatnotes too. Getting closer though. I'd like a method that doesn't require marking up the source page being transcluded, and if it is needed, makes it as minimally intrusive as possible. — The Transhumanist00:08, 15 April 2018 (UTC)
Galobtter, Can this technique be extended to transcluding the content of a short description template?, Something in the line of {{#invoke:String|match|pattern={{short description|= {{#lsth:Water}}}} · · · Peter (Southwood)(talk): 08:04, 17 April 2018 (UTC)
Hmm, @Pbsouthwood:{{#invoke:String|match|pattern=<div class="shortdescription.-</div>|s = {{#lsth:Water}}}}. Not running the code here, since it adds the short description to the village pump (that can be fixed, depending on your use case- what are you trying to do this for?) Galobtter (pingó mió) 08:47, 17 April 2018 (UTC)
Galobtter, there are uses as an annotation in some lists, such as indexes, where transcluding the short description after the link would be very informative. All that would be needed is the text from the unique short description in an article. Not needed for disambiguations etc where the same short description applies to several articles/pages. I have an idea that this would be helpful for creating automated indexes, possibly for outline lists too, maybe also as an option in category lists. Really anywhere that a short description after a link might be useful. Could also have applications in hovercards. This sort of application would make the short descriptions really useful inside Wikipedia instead of just being a pain in the butt forced on us to avoid being described by Wikidata.· · · Peter (Southwood)(talk): 09:35, 17 April 2018 (UTC)
How hard would it be to set up ORES for the Draft: space?
Hi all
So I've tried recently to find articles within the Draft: space from Wikipedia:Drafts to work on but I have found it very difficult because I have to go to each one individually to check what is in it.
I think that having a page with an ORES rating would make this much easier. E.g pages with an ORES rating of 0.9 are probably more likely to only need a small amount of work to get them to being publishable than an article with a rating of 0.4.
Would this be difficult to set up? I would be very happy to work on the documentation around it if someone technical could set up the actual rating system.
Hi John Cummings! I don't think it would be that hard. I'd like to see the Page Curation Tool start integrating ORES predictions. I think the most difficult work to make that happen is to make ORES predictions available inside of MediaWiki so tools like that can query them in a straightforward way. We've just finished work to make those predictions available. See Phab:T175757 (pending deployment). We'll be working to get that "draftquality" and "drafttopic" predictions in there soon too -- probably with a month. These predictions will allow you to quickly identify new drafts that are likely to be spam (G11), attack (G10), or vandalism (G3) (draftquality) and to categorize the rest by topic so you can review the drafts that align with your interests/expertise -- maybe even prioritizing the highest quality drafts first (wp10/articlequality). Once these are available in MediaWiki's internal databases, we need to find places to surface the predictions and to filter by them. Where would you want see the predictions surfaced? I think once we decide on a location to surface a prediction, we can build a better estimate of how much dev work it'll take. --Halfak (WMF) (talk) 21:49, 22 March 2018 (UTC)
Thanks very much @Halfak (WMF): and @SQL:. @SQL I never saw this before, very interesting :) @Halfak, this is very timely, I guess I'd really like to see Wikipedia:Drafts become a place for people to rescue drafts from drafts hell and get turned into articles, to become more of a place to collaborate on articles before (not only after) they are published. I think this will help less effort be wasted when people create a draft or when an Article for Deletion is moved to draft space. Do you have an example of what an ORES page might look like? I'm aware that there may be a very high volume of drafts so displaying them all may be quite a large page..... Thanks again both. John Cummings (talk) 15:33, 23 March 2018 (UTC)
I've been thinking about tagging draft talkpages with ORES wp10 predictions, and maybe draftquality as well. It wouldn't be terribly hard. SQLQuery me!03:25, 24 March 2018 (UTC)
SQL cool! Nice to see you experimenting with that. I think we need to find a good way to get this in front of people. E.g. if we could implement a sort-of backlog browser that could take advantage of some of these machine predictions and other filter criteria, we could help people to "rescue drafts from drafts hell and get turned into articles". John Cummings, I can't tell you how excited I am that you've got an eye for this too. I wonder if we could work with SQL to develop a sort-able list using the draft quality/wp10 models now. If we have a few people try to use it, that will help us better understand what is missing -- and what would be important for the draft hell backlog browser tool. :) I'm hoping to have the draft topic model deployed and well documented in time for the Wikimedia Hackathon in Barcelona. Either of you interested in attending or maybe just hacking with us remotely? --Halfak (WMF) (talk) 20:46, 26 March 2018 (UTC)
@Halfak (WMF): and @SQL: I'm not a programmer but very eager to help to get this off the ground where I can (I guess working on usability, beta testing etc), ley me know where I can help :) John Cummings (talk) 21:42, 26 March 2018 (UTC)
@SQL:, this looks amazing, one suggestion I have is to have table sorted by OK% as the default. Is is possible to have it updated more commonly than daily? Just thinking if it becomes popular then people may trip over each other? John Cummings (talk) 21:45, 26 March 2018 (UTC)
@SQL: super, is it possible to add definitions of the table columns at the top? I'm not clear what they all mean exactly. No idea how often they should update, is there something similar we could look at and see how often they update? John Cummings (talk) 09:52, 27 March 2018 (UTC)
@SQL: amazing work. My only other thoughts are about the order of the columns, I think the OK% column is probably the most useful so should go before the other % measurements (it appears to be a summary of the other percentages?). Just an observation that having one very long file name in the the Article column makes that line very long. These are all a bit of nitpicking rather than big things, I feel like its ready to be used when a short introduction is added to explain the purpose of the list, something like:
This list (generated by ORES) rates draft articles to help users find suitable articles to improve and then publish.
Any suggestions on the best way to integrate this into the Wikipedia:Drafts page? I'm aware its a very long list...perhaps a translusion with a scrolling section or a collapsable section?
Two things - one, I was thinking of removing "Vandalism" and "Attack" all together - they aren't really useful in this context. Two - maybe 'top 25 possible copyvios' / 'top 25 spam' / bottom (and/or top) 25 OK tables transcluded onto WP:DRAFTS? SQLQuery me!02:52, 29 March 2018 (UTC)
@SQL:, I think removing vandalism and attack are a good idea, I've been looking at the spam articles and many of them appear to be of notable subjects, just written in a tone that doesn't fit with Wikipedia so I'm not sure how useful that label is. Transcluding the top 25 would be a great idea, I guess if one gets popular you could upgrade to top 50... Let me know if there's anything I can do to help, once its live I'll be sure to let people know about it :) John Cummings (talk) 07:57, 29 March 2018 (UTC)
Hi all -- I'm Marshall Miller; I'm a new product manager on the Community Tech team with DannyH at WMF. As part of the follow-up to ACTRIAL, the Community Tech team is going to have some bandwidth over the next couple months to take on an improvement to the AfC process. I've been following along with the conversation on this talk page and elsewhere to learn about the biggest challenges facing AfC and the ideas for improving it.
The idea of integrating ORES scores or Copyvio checks in a streamlined fashion into the AfC workflow is one of the ideas being discussed on the talk page of that summary linked above, so I wanted to let you know in case you want to participate in that conversation over the course of the next week as we consolidate around some of the top ideas for improvements that WMF could help with.
Responded on the talk page. I am not an experienced articles for creation reviewer, so please be wary of some of my comments that might be misleading or counter constructive as a result. --Gryllida (talk) 04:02, 5 April 2018 (UTC)
the tables seem very useful. I want to track what actually happens to the articles there. Is there any automatic way of providing a historical record of what gets added/removed, or snapshots at intervals? (If not, I may do it manually). DGG ( talk ) 20:23, 11 April 2018 (UTC)
@SQL:, Wikipedia:Drafts looks great :), I just did a bit of testing with a few volunteers and we have a few requests of changes:
Please can the list update more frequently or have a manual update button (its not clear on how regularly the page updates).
Is it possible to link to longer lists than top 25 for each section? E.g Top 250? One of the testers wanted to start using the draft space as their place to create articles and getting to the Top 25 is a bit of a dark art.
The term 'OK drafts' is a bit confusing, is saying 'highest quality' and 'lowest quality' a good alternative?
@John Cummings: - Absolutely on all 3 points. It was updating daily, I've bumped that to ~every 6 hours. I can make some more subtables, possibly tomorrow. I don't think the OK% reflects the quality of the draft, but I'm not 100% sure. mw:ORES#Curation_support reads like it might be the inverse of the likelyhood of being CSD'ed for being spam/vandalism/attack. SQLQuery me!01:41, 14 April 2018 (UTC)
@SQL:, great stuff :), I'll think some more about the naming, maybe just 'highest rated', 'lowest rated' and 'potential spam'? Top 25 OK is very underwhelming :) John Cummings (talk) 13:33, 15 April 2018 (UTC)
Great chart but simply overwhelmed for transclusion on this explanatory supplement page arrived at by many many talks. Pls dont make our editors looking to derive serviceable information from the page to have to scroll though a huge list at the start of a parent type article. --Moxy (talk) 04:35, 21 April 2018 (UTC){{quotation }}
The irony is that Aaron is talking about this earlier in the thread with John Cummings but they don't appear to be aware of the other discussions on the same topics in in other venues where Marshall Miller is also leading some discussions. I'm never sure how the work at the WMF overlaps, or if indeed it does, or even if we are discussing these issues at too many different venues.. That said, the simple answer is that it should dovetail. This is one of the very reasons why the New Pages Feed should incorporate the flow of Drafts - because AfC doesn't have such a flow overview. So it's all a bit chicken-egg. AFAICS, many new page reviewers AFAICS don't appear to be bothered to use most of the excellent article meta information that is displayed in the feed, which is a great shame, so I'm not sure if ORES would help much. It might even lead them to making even more superficial checks on the new pages. Too much reliance on AI technology sometimes turns out to be a net negative to accurate productivity. That said, overall, NPPers do indeed do a somewhat better job than the AfCers - for now. With the correct training in these apps however, perhaps we would meet our goals, but even the tutorial at WP:NPP seems to be a challenge for some. On a more positive note, I believe there is finally a fully automatic replacement for the sorely missed Coren Search Bot under development which would help enormously.Kudpung กุดผึ้ง (talk) 05:42, 21 April 2018 (UTC)
@Moxy:, can you describe your concern here so that we can have a discussion about if it is possible to address this? Simply removing a useful tool people are using because you don't like it is disruptive editing, please leave the tool on the page until we have discussed. Thanks, John Cummings (talk) 15:33, 21 April 2018 (UTC)
Suggestion: we change the tables to collapsed so that they are easy to find and access but do not take up room for people who are not using them. John Cummings (talk) 15:37, 21 April 2018 (UTC)
Er, no, John Cummings, it's fairly standard practice - suggest you self-revert until there is consensus to include. Moxy is right: whether the tool is useful or not, it overwhelms the informational page, and it really isn't necessary to transclude it there at all for people to use it. Nikkimaria (talk) 15:43, 21 April 2018 (UTC)
I suggest something like this, it maintains all of the functionality of the uncollapsed version and addresses the concerns of taking up space on the page John Cummings (talk) 16:11, 21 April 2018 (UTC)
@Nikkimaria: Ah, sorry, forgot to change the section headings to <h4>, if the problem persists (I'm sure that's a way of avoiding it making section headings but it doesn't seem to have filtered through) I can just change it to big or something else. Other than the formatting issue I feel like this resolves the issue, what do you think? John Cummings (talk) 23:21, 21 April 2018 (UTC)
As mentioned above simply overwhelming for an information page. Guess I will have to make a new page WP:Finding drafts for this table and explain what the chart is about and more info about the topic. Will do this in the next few days. This does raise a concern.....do we have these walls of links on other info type pages. If so perhaps we should explain accessibility concerns about this on our MOS pages. It's disruptive to have a wall of links leading to editors not getting serviceable information..... we know statistically editors simply won't scroll through all that mess wall of text to get any information *Which_parts_of_an_article_do_readers_read....we structure our help pages in a certain format for a reason Moxy (talk) 01:40, 22 April 2018 (UTC)
Thanks for bringing my attention back to the conversation here, Kudpung. Aaron and I have been talking a lot about this in the last couple weeks, so he and I are in sync on this front. SQL and John Cummings -- what you have been putting together so far is really great, and the Community Tech team is thinking on our end what we can learn from it as we mull over the potential improvements to AfC summarized here, which sound similar to what you're doing. We'll be able to build some more detail around our thoughts this coming week as people come back from Wikimedia Conference. -- MMiller (WMF) (talk) 15:26, 21 April 2018 (UTC)
Thank you, Marshall for pointing me to this. Aaron, SQL, John Cummings, and Gryllida, these lists are really excellent. However, and as mentioned above by DGG, doing this post facto they represent a lot of work for anyone wishing to work through them manually. AFAICS, AfC still has no recognised work flow in the sense that one exists for WP:NPP as shown in the chart by Insertcleverphrasehere, which while complex and somewhat intimidating, has received positive comments by serveral users, or the more simplified one at File:New Page Review Flow Chart.jpg. IMO, the meta information provided by ORES would be very useful if it were highlighted in the Special:NewPagesFeed if submitted drafts were to be incorporated there with a user defined option in the preferences to select drafts only. More reason to consider migrating the functions of AfC into the NewPagesFeed and Curation UIs. Despite the resistance of Community Tech to prioritise the upgrades required for these UIs, IMO this would be a more rational deployment of WMF resources.Kudpung กุดผึ้ง (talk) 01:41, 22 April 2018 (UTC)
@Kudpung:, Marshall, Aaron, SQL, DGG, Moxy, and Gryllida are you happy for me to re-add the list in the form below until a longer term solution is potentially found through the suggestions above? This solution reduces the size of the list when people aren't using it to 2 lines and does not effect the table of contents. I hope this proof of concept helps to show the value of having something more developed and so make it more likely to have resources allocated to making it happen. John Cummings (talk) 07:23, 22 April 2018 (UTC)
I understand Moxy's concerns. My concern is more: Who will go to Wikipedia:Drafts and see and use the list even if it is collapsed? Perhaps it could be on a dedicated sub-page such as, just for example, Wikipedia:Drafts/ORES. That said, it would need to be linked to from AfC (and/or) the page patrolling tools. However, that would be the same parallel problem with CopyPatrol.Halfak (WMF) has said he would like to see ORES integrated into the Curation system, but I think he means into new arriving articles. I can certainly imaging that meta information being included in the entries in the Special:NewPagesFeed. So I ask again, what is the problem with giving AfC the opportunity to share that feed for their reviewing? It would seem to me to be a ready-made solution (bar the engineering). Kudpung กุดผึ้ง (talk) 08:53, 22 April 2018 (UTC)
Does ORES provide WikiProjects tags? If ORES was used to add WikiProject tags to any articles which do not already have them, that could be interesting. I am really interested in knowing what percentage of the main namespace is tagged with a WikiProject and what is not. --Gryllida (talk) 04:29, 24 April 2018 (UTC)
I don't know. If a bot were sent around to re-assesses all of the redirects as being redirects (and disambiguation pages as being dab pages), then you could probably calculate it. However, I think it's safe to assume that most articles are tagged, possibly 90% or even more. Clicking Special:Random ten times gave me ten tagged artices. Whatamidoing (WMF) (talk) 22:30, 25 April 2018 (UTC)
Any auto-hyphenation yet
I wonder if yet any way to auto-hyphenate long words on browsers, as done 40 years ago by soft hyphens (versus "required hyphens"), to now wrap long words on narrow windows or smaller hand-held devices. See Danish sports page: "1927–28 Danmarksmesterskabsturneringen". I have used "<wbr>" to auto-wrap non-hyphen text, but want to auto-hyphenate to append "-" in each mid-word wrap. No hurry on this. -Wikid77 (talk) 00:49, 18 April 2018 More: As I recall, the soft-hyphen "­" worked fine on either computer browsers or mobile phones, but there were concerns for some screenreader devices to handle Wikipedia pages. However, I was thinking the newer screenreaders could handle (skip) the "­" characters when processing the hyphenated text. Alternatively, I wonder if soft hyphens were to be limited to one per long word, as a compromise to reduce problems in screenreaders, that limit might allow more use of soft hyphens in WP text. -Wikid77 (talk) 04:06, 18 April 2018 (UTC)
i think the primary concern is readability of the wikitext honestly. Anyway, hyphenation is primary a layouting problem, and with HTML that makes it a responsibility of the browser, not of wikipedia. It seems that browsers still dont recognize syllables in arbitrary languages and thus still dont support this as a layouting option. Then again, i feel that people are increasingly confusing the web with a typesetting platform sometimes. Its not, the web is designed to be dynamic, not to be an optimal typesetting language. Thats what latex is for. You cant even dictate what font or what size of rendersurface a person is using, so dictating hyphenation is even harder and shy’ing every syllable of every word seems an excessive solution to a minor problem to me. —TheDJ (talk • contribs) 06:24, 18 April 2018 (UTC)
The relevant W3C Working Draft was last updated on 22 August 2017 - just 8 months ago. It should not be considered stable, and is not a finished work - that won't occur until it is promoted to W3C Recommendation. Nutshell: don't rely on it. --Redrose64 🌹 (talk) 13:28, 18 April 2018 (UTC)
Just wanted to put a word in that I definitely noticed, as have some of my colleagues and friends. I understand that in the case of something similar to "Danmarksmesterskabsturneringen" that soft hyphens would be helpful, but hyphenating words of all lengths across line breaks really impacts the readability of an article. I'm not speaking with any knowledge about the different technical solutions to this, just purely feedback on how it affects some more casual Wikipedia users. Peloneoustc03:23, 25 April 2018 (UTC)
I think we should make 'final' decision after 2 weeks to a month or so. Having a solid amount of feedback will be very useful. —TheDJ (talk • contribs) 10:53, 25 April 2018 (UTC)
That sounds good to me. I'll keep an eye out on this page, so if the conversation or feedback about this moves to another location, I'd appreciate a link here to that. Peloneoustc03:45, 26 April 2018 (UTC)
@Xaosflux: thanks. I added my comment at phaprecator task. I wrote here just to make sure that there is no local mediawiki page need to be modified according to thia issue. Regards--مصعب (talk) 17:45, 26 April 2018 (UTC)
Neither of these replies have helped in any way. The way that I read the linked pages is that: either both of these MassMessage posts should have triggered the tag; or neither of them should have. How can it be that one triggers yet the other doesn't? --Redrose64 🌹 (talk) 18:16, 26 April 2018 (UTC)
The two diffs are posted to different user pages; one of the users could've triggered a few more conditions than the other, based on their username or edit count or whatever the filters on enwiki happen to check against. The condition limit takes into account short-circuit evaluation, so if a hypothetical filter has a condition that says (article_text rlike 'Phi.*') & <other stuff>, then the Whispyhistory edit will only accrue one condition (the rlike) and no more, because the rlike comparison failed and the filter can stop there, whereas the Philafrenzy edit will keep going, since it matched. Writ Keeper⚇♔18:29, 26 April 2018 (UTC)
We have been warned many times over the last year or so, but HTML Tidy seems to have been turned off. This means that pages with unclosed (or badly-closed) markup may now show large amounts of corruption. Examples:
It seems to have been inadvertently enabled in all wikimedia wikis and certainly caused https://phabricator.wikimedia.org/T192855, and probably other reports there. It will probably be reverted as soon as the developers notice it. It makes the case for simply leaving it as is considering that there hasn't been that much of "drama" related to the early deploy.
@Redrose64:, @Galobtter:, @Whatamidoing (WMF): Click on the edit link from the lint errors listing which will take you to the regular edit form with the wikitext section that is the source of the problem. You can then replace the ?action=edit part of the URL with ?action=parsermigration-edit and that will both give you the side-by-side preview and also continue to highlight the problematic wikitext section. For example, here is that link for the basketball season page. In this case, the problem is with '''{{Nowrap|[[Galatasaray Medical Park (men's basketball)|Galatasaray Medical Park]]''' {{flagicon|TUR}}}} where the quote tags are improperly nested across the nowrap-span. As you can tell, in some cases, there is no rendering difference since the fixups by Remex and Tidy are effectively the same, but this is hard to determine automatically, so we flag all potential cases. SSastry (WMF) (talk) 22:29, 24 April 2018 (UTC)
You can also get quick access to this side-by-side comparison through your prefs. Go to Special:Preferences#mw-prefsection-editing and choose "Enable parser migration tool". Then look for a new option in the sidebar. NB that today this is in the "General" group at the top, and that during the next deployment train, it will move to the bottom of the page, in a new section called "Developer tools". Whatamidoing (WMF) (talk) 22:23, 25 April 2018 (UTC)
Two thoughts:
Most of the noticeable problems were on talk pages. The editors working on this problem are mostly focused on the mainspace. It might be a good idea to double-check your talk page and/or any other talk pages you care about. Just add ?action=parsermigration-edit to the end of the regular (/wiki/ style) URL, scroll down, and see if anything looks broken. Also, I want to thank User:Anomalocaris in particular for individually contacting editors whose sigs were broken, starting months ago, because getting those fixed will reduce the number of problems on active discussion pages such as this one.
When this change happens (in about a month or two), any resulting formatting problems won't be visible until the page is re-parsed. This means that the problems will become visible after someone edits the page, when the page is purged, or when the cache expires. I expect, as a result of that delay, that people are going to see problems when someone makes an unrelated edit, and that this page is going to see questions along the lines of "I fixed this simple typo at the top, and now something's wrong at the end of the page, and reverting doesn't fix it". Whatamidoing (WMF) (talk) 19:03, 26 April 2018 (UTC)
Strange interwiki problem
Could you open the article "Anybody Seen My Baby?" and try clicking on the "Bielorussian (taraškievica)" interwiki (the first one)? When I click on it, I am being sent to a page that doesn't have a question mark in its title (to a page that doesn't actually exist). But when I go to the corresponding Wikidata item and click on the same interwiki ("be_x_old") there, it works correctly. --Moscow Connection (talk) 13:57, 27 April 2018 (UTC)
Something seems off...we should link directly to the correct wiki location, not to the redirect. Also, the redirect should retain the full page title, not chop it off (I bet it's treating the ? as start of a query string). FACE WITH TEARS OF JOY[u+1F602]16:20, 27 April 2018 (UTC)
Hi. When I search in Google, find a Wikipedia article, instead of the English version, I'm taken to Google Translate for "Terjemahan". -SusanLesch (talk) 14:48, 27 April 2018 (UTC)
User:Moscow Connection, Thank you! Then the mysterious language may be Indonesian. My husband's machine on the same connection doesn't do this so TheDJ might be right, too. It happens for any search, like "Minneapolis." -SusanLesch (talk) 21:51, 27 April 2018 (UTC)
Can someone have a look at Category:1710s and click on the subcategories for each year. For me, on all the years, they expand to a message, saying: "Problem loading data. Please wait a moment and try again." The portion in italics is blue linked, but dead for me. Thanks.--John Cline (talk) 05:49, 28 April 2018 (UTC)
Everything works for me. All the categories are there, I can browse them without any problem whatsoever. --Moscow Connection (talk) 06:15, 28 April 2018 (UTC) Update. I didn't understand the topic starter's message correctly. Yes, I can browse the subcategories, but if (instead of opening them) I just click on the arrows to expand them, I, too, see "Problem loading data. Please wait a moment and try again." --Moscow Connection (talk) 06:51, 28 April 2018 (UTC)
Some portals would like to show an excerpt from a page chosen from a category. Is there any way for a module or template to list the members of a category? title:getContent() on a category page just returns the "This is a list of foos" blurb, not the actual members. mw.site.stats.pagesInCategory counts the pages but doesn't reveal the titles. getContent doesn't work on Special:RandomInCategory, and transcluding it just produces a link rather than the text. Surely we don't need to write a bot to periodically copy the category page to a standard page (which we can then read easily)? Certes (talk) 01:32, 28 April 2018 (UTC)
@Certes: You can get a listing of category members via Special:WhatLinksHere/Category:Name, if it is restricted to namespace 0 (articles). This can be done via transclusion, e.g. {{Special:WhatLinksHere/Category:Australia|namespace=0|limit=200}}, and so is available in Lua via frame:preprocess(). That's as far as I got in Module:Sandbox/Evad37/randomInCat, but it should be possible to do some pattern-matching magic to extract a random title from the whatlinkshere list. - Evad37[talk]04:39, 28 April 2018 (UTC)
No, actually, after looking into it a little. Returns a strip marker instead of the list, and per the reference manual can longer access the list: mw.text.unstrip no longer reveals the HTML behind special page transclusion, <ref> tags, and so on as it did in earlier versions of Scribunto.Galobtter (pingó mió) 06:45, 28 April 2018 (UTC)
Galobtter beat my reply but I'll post it anyway. That's interesting because putting that into Special:ExpandTemplates shows a bunch of HTML. By contrast, trying {{Special:WhatLinksHere/Category:Australia|namespace=0|limit=200}} in ExpandTemplates shows merely a single line:
[[:Special:WhatLinksHere/Category:Australia]]
However, the following shows that a module gets a strip marker rather than anything useful:
{{#invoke:dump|dumphtml|1= {{#categorytree:Countries in Africa|hideroot|namespaces=Main}} }} →
You can put that in a template and it works just the same, I believe that some WikiProjects do this - possibly using the alternative syntax <categorytree hideroot=on namespaces=Main>Countries in Africa</categorytree>. A third valid syntax is {{#tag:categorytree|Countries in Africa|hideroot=on|namespaces=Main}}. --Redrose64 🌹 (talk) 11:27, 28 April 2018 (UTC)
The original question is about how to get a random article from a category (list the category members in a lua readable way), hence trying to get the output readable in lua so a random member can be selected. Galobtter (pingó mió) 11:31, 28 April 2018 (UTC)
Thanks for all the replies so far. #categorytree looks promising but I can't make it give me the titles within a module for further parsing. frame:preprocess returns the string "nil" (not Lua's nil value), and frame:expandTemplate gives an error (quite reasonably, since it's not a template). Attempting to preprocess a dummy page containing "<categorytree...", or to expandTemplate a template with that text, return the string "nil" too. Further suggestions are very welcome. Certes (talk) 12:24, 28 April 2018 (UTC)
This is another use-case for an interface between the MediaWiki API and Lua, even if only some API entry points. My only suggestion is a cron bot to post the data where Lua can pick it up. This 2014 thread and this 2016 comment by User:Mr. Stradivarius confirms Lua does not have access to category members. -- GreenC14:33, 28 April 2018 (UTC)
When editing, I'm now seeing two mysterious extra checkmarks just below the proper check boxes. They first appeared a few days ago, but vanished within minutes. Now they're back: what are they for? MonoBook, Opera 36, Windows XP. --Redrose64 🌹 (talk) 20:58, 23 April 2018 (UTC)
I have difficulty trying other browsers. I can't upgrade from XP since the hardware won't be able to handle anything more sophisticated, so I'm confined to using XP-compatible browsers. Safari hasn't been updated since about 2012; IE8 is rejected by MediaWiki; Chrome hasn't been updated since April 2016. Firefox still gets the occasional update but is horribly slow - three minutes to follow a link and six minutes to go back again, which is prohibitive for watchlist checking. That leaves Opera. --Redrose64 🌹 (talk) 23:31, 23 April 2018 (UTC)
If you have hardware that old, have you tried Ubuntu or another Linux OS? Their requirements should be significantly less than Windows (which is in the realm of 4 GB RAM just for the OS right now...). --Izno (talk) 00:19, 24 April 2018 (UTC)
In that vein, I've had a lot of luck using Neverware Cloudready (Chromium OS - turns your laptop / desktop into a chromebook / chromebox), booted from a thumbdrive in older machines. SQLQuery me!00:29, 24 April 2018 (UTC)
These extra checkmarks also show up for me on my phone in the desktop version (Android 5.1.1, Samsung Browser 3, Monobook). Double sharp (talk) 02:35, 24 April 2018 (UTC)
@Redrose64: I looked into this, and it was a temporary issue when deploying a new version of MediaWiki – on last Thursday and last Monday, for about five minutes each, Wikipedia was using incompatible versions of HTML and CSS for the checkboxes, causing this weirdness on MonoBook (and causing the checkboxes to be very slightly misaligned on Vector and other skins). I wrote a longer description on the Phabricator task: phab:T192951#4155800. I think we understand why this happened and it shouldn't happen again. (Thank you for including the screenshot with your bug report, it was actually surprisingly helpful!) Matma Rextalk22:54, 24 April 2018 (UTC)
Also, while Opera 36 is probably the best browser you could be using if you still have to use Windows XP (and these weren't the cause of this problem), I really would recommend upgrading your hardware. You can get a decent used laptop (e.g. Thinkpad T500) for $100 or less, it should comfortably run Windows 7/10 (I have one, use it for testing the sites on old hardware sometimes). Matma Rextalk23:04, 24 April 2018 (UTC)
@Matma Rex: Please disregard the above – I figured out that the persistence of the problem was caused by the line $(".oo-ui-checkboxInputWidget").removeClass("oo-ui-checkboxInputWidget") in my common.js page. I've removed this line and everything is now appearing as it should. Also, my thanks to @TheDJ: for suggesting that .oo-ui-checkboxInputWidget .oo-ui-checkboxInputWidget-checkIcon.oo-ui-iconElement-icon could be the problem, which prompted me to try doing this! Double sharp (talk) 16:30, 28 April 2018 (UTC)
Checkboxes obscuring start of text
Seems to be a recent development: when I edit (standard wikitext mode) the two checkboxes in front of "this is a minor edit " and "watch this page" are now too big in all directions, which is most noticeable because they now obscure part of the "T" and part of the "W" of these two texts. Fram (talk) 13:36, 25 April 2018 (UTC)
Thanks. Safemode=1 puts it just next to the boxes (though still extremely close. When I enlarge my screen to 200% (I normally work at 100%), I can't see any space between the left side of the T or W and the box, which isn't visually attractive. But, like I said, it is no longer obscured in safemode, so some gadget must cause the issue (I have no css or other fancy stuff added, only enabled or disabled things in preferences, most recently the "show the wikidata description" one). Fram (talk) 13:53, 25 April 2018 (UTC)
There's a recent spate of seven consecutive malformed edit requests at Talk:Grand Theft Auto: San Andreas from various IPs. All of them are empty and as such weren't implemented. My assumption is that anons don't know how to properly file an edit request, but could someone check what it's all about? Brandmeistertalk22:36, 26 April 2018 (UTC)
The IP's are from around the World so I guess nothing special is going on. If you click the "View source" tab or try to edit a protected page without permission then there is a "Submit an edit request" button. The form of the empty requests with dated headings show they all clicked the button and then saved without writing anything. This is quite common, it happened during nearly three months and the article had 117,383 views in the past 30 days alone, and the top has a box with an "improve this article" edit link which leads to the edit request button for IP's. Some editors just revert empty edit requests. PrimeHunter (talk) 23:26, 26 April 2018 (UTC)
Some blank requests are people trying to edit the article, but not understanding, others are trying to get around the protection (it is not uncommon for an IP who has left a blank SPER to delete the pp template, hoping that doing so will grant them access). With many people's attention spans being so short, the instructions (seen here) are too long, the IP just wants to make an edit, not read a lot of instructions. I tend to leave a short instruction for the first/latest, empty SPER, which, sometimes, leads to a proper request. - Arjayay (talk) 09:43, 27 April 2018 (UTC)
Howdy all, I use WP:EPH to assist with edit request which when clicking Remove Request states: "Are you sure you want to completely remove this edit request from the page? In general, this should not be done to good-faith requests, even if they are blank, unless they are duplicates of other requests by the same user, etc." in a popup. So that's the line I follow. I feel so long as pages have an archive it'd be no harm to give them the template x/y response and if really necessary remove them in bulk at a later date. Personally I think it is better to help that 1 in 5 that genuinely wanted to improve an article for the cost of a little more talk page spam. — IVORKDiscuss12:50, 27 April 2018 (UTC)
@Arjayay:it is not uncommon for an IP who has left a blank SPER to delete the pp template, er, how can they remove the template if they can't edit the article? Please provide diffs of where this has happened.
However, as explained above WP:EPH states: "Are you sure you want to completely remove this edit request from the page? In general, this should not be done to good-faith requests, even if they are blank" so the current advice is mutually contradictory. - Arjayay (talk) 20:56, 27 April 2018 (UTC)
IP edit requests have become common in the past few years often with POINTy or otherwise vandal-like or trolling aspects. IMO the best solution is to delete them rather than littering the talk page which encourages more "requests". -- GreenC20:44, 27 April 2018 (UTC)
Agreed. Please delete empty or suspected trolling edit requests. Look at it from the perspective of the other person: they see a button, they click it, they don't actually have a constructive proposal so some of them close the tab page and we see nothing, while others publish their thoughts. Such junk attracts other fluff and should not be encouraged. Johnuniq (talk) 00:08, 28 April 2018 (UTC)
Yeah that is what I do, just litters the talk page for no benefit, it is a test edit and thus is reasonable to revert. I do usually leave an explanation in the edit summary and revert the edit so they get a ping with the message. Galobtter (pingó mió) 11:21, 28 April 2018 (UTC)
Is there any way for PAGESINCATEGORY to show how many pages that appear in both of two categories?
Hi
Is there any way for PAGESINCATEGORY to display pages that appear in both of two categories? I don't mean adding the sums of two categories together, I mean showing only the pages that appear in Category A and Category B.
John Cummings specifically asked for PAGESINCATEGORY so I suspect he wants wiki code which automatically updates it in a wiki page. This is not possible. PrimeHunter (talk) 21:46, 27 April 2018 (UTC)
I'm not sure if {{PAGESINCATEGORY}} will do what I think you are attempting and I couldn't find an existing template doing it. If I did understand you, however, I worked up a template that might do some of what your seem to be trying to do (hopefully). I've used Category:1711 births and Category:1711 deaths for a demonstration and they plug in to the template with the following syntax: {{Ct2|1711 births|1711 deaths|mode=pages}} and render as:
@John Cline:, sorry for the late reply, thanks so much for all the effort :) I didn't explain myself very well, I'm just looking for something that will simply give me a number rather that a list (see the In Progress section on this page. I'm sure your work will be useful for many people, I encourage you to turn it into a template :) John Cummings (talk) 09:06, 30 April 2018 (UTC)
Latest tech news from the Wikimedia technical community. Please tell other users about these changes. Not all changes will affect you. Translations are available.
Recent changes
The parameter for unpatrolled edits in recent changes filters changed name. You might need to update saved filters and links. [14]
Problems
We are migrating wikis from Tidy to Remex. Because of a bug the 250 wikis which do not yet use Remex were switched on 23 April. This is two months early. This meant that pages with broken wikitext showed wrongly to readers. The bug was undone the next day. You can help fix broken wikitext to avoid this problem when your wiki switches. Tidy will be removed on all wikis before July 2018. You can follow the process on Phabricator. [15]
When an administrator blocks someone they will have a calendar they can use to choose when the block ends. This is to make it easier to pick a specific date. [17]
You can soon turn on the Performance Inspector in the Editing section in your preferences. It shows information about the performance of pages. This could be the size of modules in the page, how many CSS selectors are defined on the page and how many are used, or the size of the images on the page. This tool is intended to help editors fix pages that load slowly. [18]
There is a new abuse filter function called equals_to_any. You can use it to check if its first argument is equal (===) to any of the following ones. For example you can use it to check if the page's namespace is amongst a given set of values in a more compact way than you could earlier. You can read more on mediawiki.org.
The new version of MediaWiki will be on test wikis and MediaWiki.org from 1 May. It will be on non-Wikipedia wikis and some Wikipedias from 2 May. It will be on all wikis from 3 May (calendar).
Meetings
You can join the next meeting with the Editing team. During the meeting, you can tell developers which bugs you think are the most important. The meeting will be on 1 May at 18:30 (UTC). See how to join.
You can join the technical advice meeting on IRC. During the meeting, volunteer developers can ask for advice. The meeting will be on 2 May at 15:00 (UTC). See how to join.
Future changes
The Wikimedia Cloud Services team is working on a new project called Toolhub. The goal is to make it easier for Wikimedians to discover software tools they can use. You can leave feedback on the talk page or email jharewikimedia.org to leave private feedback.
All wikis with fewer than 100 high-priority linter errors in all namespaces will switch to use the Remex parsing library. This is to replace Tidy. It will happen on 2 May. Other wikis will be recommended to switch soon when they have fixed the errors that must be fixed. Wikibooks wikis with fewer than 100 high-priority linter errors in the main namespace will switch on 9 May. Tidy will be removed on all wikis before July 2018. [19][20][21]
Something very odd appears to be happening. The initial discussion was at WT:CS1. It looks to me as though MediaWiki has become somewhat schizophrenic in how it handles <bdi>...</bdi> in wikilinks. The cs1|2 templates use <bdi>...</bdi> to wrap titles for languages that use non-Latn scripts. <bdi>...</bdi> conveniently isolates unspecified text direction and allows the browser to determine how to correctly handle the wrapped text. This functionality in the cs1|2 templates is not new; it was introduced 11 October 2014.
As I began writing this, the page source for Help talk:Citation Style 1 had this:
<p><a rel="nofollow" class="external text" href="http://cpc.people.com.cn/BIG5/64162/64172/85037/85038/7104292.html">"Lǐ Dàzhāo de Zuòyòumíng" <bdi lang="zh">李大釗的座右銘</bdi></a></p>
I then clicked the edit link for the above-named section, clicked Show preview, right-clicked View page source, and got this from the source:
<p><a rel="nofollow" class="external text" href="http://cpc.people.com.cn/BIG5/64162/64172/85037/85038/7104292.html">"Lǐ Dàzhāo de Zuòyòumíng"</a></p>
<bdi lang="zh">李大釗的座右銘</bdi>
Then to be perverse, I clicked Publish changes (without changing anything), fetched a new page source where I found this:
<p><a rel="nofollow" class="external text" href="http://cpc.people.com.cn/BIG5/64162/64172/85037/85038/7104292.html">"Lǐ Dàzhāo de Zuòyòumíng"</a></p>
<bdi lang="zh">李大釗的座右銘</bdi>
The first version above is the desired version. I'm pretty sure that this isn't a browser bug (I'm using latest version Chrome) because the html is so obviously different in the three inspections which leads me to suspect something in how the Wiki source is rendered into html.
I checked the content of the <bdi>...</bdi> just to be sure that there is nothing odd encoded there. There isn't. All of those characters belong to Unicode CJK Unified Ideographs (U+4E00–U+9FEA).
The parser is parsing it just fine. Tidy is screwing around with it based on some other broken code elsewhere. Open the section with parsermigration-edit and you see that the right hand side is rendering correctly, and that is with Remex. When I parse this page locally with Parsoid with linting enabled, I see 3 linter complaints:
So, my speculation is: if you fix the missing-end-tag for the p-tag on the page around character offsets 36381 - 36601, my guess is you should probably see expected behavior with Tidy as well. Or, wait for Tidy to be switched (end of June). SSastry (WMF) (talk) 14:31, 30 April 2018 (UTC)
I confess that I don't understand the meaning of the linter output. However, I was able to glean sufficient information to lead me to a fix at Module:Citation/CS1/Configuration/sandbox for an extraneous </close> tag. That particular error has been in place since 14 October 2015. So, thanks for the clue.
I left this out: Even with the linter errors fixed higher up in the page, when editing the section, those errors are out of scope for an html rendering via Show preview. For the examples above, I was editing the section, not the whole page. This leads me to think that the linter errors are not the cause of the <bdi>...</bdi> problem because the linter did not find any errors in the section.
You are correct. The broken behavior is back - which is weird. Parsoid's output looks good. So, you can reduce it down to a single external link. And, if you keep hitting 'Show Preview' repeatedly, in some cases, the rendering looks good, and in other case, it doesn't. So, this almost feels non-deterministic. I tried this on metawiki (which has Tidy) and I see the same behavior. I tried this on mediawiki (which has Remex) and I don't see this behavior. This is mysterious indeed and I am not sure why Tidy would do different things for different runs. Will file a ticket just to record this (just in case this is not Tidy-specific), but, it is unlikely we will try to fix it if we know it is Tidy-specific since Tidy is going away in 2 months. SSastry (WMF) (talk) 16:49, 30 April 2018 (UTC)
Same here as well. I am using an up-to-date Chrome 66 and Windows 10, 64-bit. However the other Twinkle functions still works, it's just only the rollback links that are missing... theinstantmatrix (talk) 20:59, 27 April 2018 (UTC)
For me is it is missing in Wikitext view but is back (in a new form) in VisEd mode. I'm using Monobook skin. It seems to have started with the introduction of the new "browse history" widget. Was this introduction trailed somewhere? It passed me by if it was. The good news is the browse history header has buttons making it easy to switch to VisEd. SpinningSpark21:20, 27 April 2018 (UTC)
If I turn off "Visual differences" in preferences and purge the cache then the problem goes away (but the "browse history" feature stays on). SpinningSpark22:53, 27 April 2018 (UTC)
I turned off "Automatically enable all new beta features" and "Visual differences" and the issue is resolved for now. It appears that this issue is being looked into. Classicwiki (talk) If you reply here, please ping me.23:58, 27 April 2018 (UTC)
I'm sorry about this, I didn't anticipate that adding the "header" to visual diffs (which we did last week) would cause issues with other scripts. I've just proposed a patch to do this differently (in a way that should interact better with other tools like Twinkle), but it might take a day or two before we get it reviewed and deployed. In the meantime, if you want to keep using visual diffs and Twinkle, note that the "[restore this version]" link is actually present, but only when you switch to the "Visual" tab (using the buttons in top-right corner of the diff). Sorry! Matma Rextalk21:20, 30 April 2018 (UTC)
Is anyone aware of the default implementation of Navigation popups for unregistered users? We've had a couple of complaints to OTRS in the last few days and I'm not aware of any discussion or announcement saying that the function would be enabled automatically. Can anyone point me to the discussion where this was agreed upon, so that I can direct OTRS enquiries to the correct information? Cheers, Yunshui雲水08:36, 1 May 2018 (UTC)
Are you requesting a new gadget with this feature? You can write the namespace and a colon in the search bar, e.g. wp:searching where wp: is an alias for Wikipedia:. You can click "Advanced" in search results pages to select one or more namespaces with a click. The below in your common JavaScript will add a "Search" link to the top of pages, going directly to Advanced search with "Article" selected. PrimeHunter (talk) 11:01, 1 May 2018 (UTC)
Do you guys have any plan to maintain, especially refactor, some of the modules? A keyword is "code smell", and then complexity… Some of the code is even worse than mine. Jeblad (talk) 21:06, 1 May 2018 (UTC)
The same could be said of many templates, noticeboards, and articles. The problem with refactoring a module (or a complex template) is that unless the time is spent to get fully immersed in defining the problem and solution, it is very easy to overlook gotchas, and testcases are usually absent or insufficient. At any rate, feel free to nominate a couple of modules that are used as needing refactoring. Johnuniq (talk) 00:58, 2 May 2018 (UTC)
MySQL and Toolforge
I'm trying to connect to MySQL at Toolforge via HeidiSQL, but after filling login form I get error "Can't connect to MySQL server on 'localhost' (10061)". But I don't see anywhere any reference to localhost. Probably I simply did something silly. Some help needed :) --Edgars2007 (talk/contribs) 16:43, 29 April 2018 (UTC)
You set up the SSH tunnel incorrectly. You want your computer <-> toolforge bastion <-> wiki replica. In the SSH tunnel you attempted to connect to wiki replica directly, which isn't possible and the enwiki.analytics.db.svc.eqiad.wmflabs will not resolve. Specify a bastion host instead, like login.tools.wmflabs.org with port 22 for ssh. --Zhuyifei1999 (talk) 14:42, 1 May 2018 (UTC)
As for why it's connecting to localhost but failed, the ssh tunnel is supposed to port-forward the port 3306 of enwiki.analytics.db.svc.eqiad.wmflabs to a local port on your computer, which you specified as 3306 as well. i.e. it will open a port 3306 on your computer (localhost) and relay all traffic to and from wiki replicas through the ssh connection. Since the configuration for ssh tunnel is broken, the tunnel cannot be set up, and the port will not open. The program will therefore fail to connect to the port. --Zhuyifei1999 (talk) 15:00, 1 May 2018 (UTC)
Is your private key encrypted? If so try run an SSH agent like PuTTY's pagent.exe to have the key loaded.
The password of the SSH tunnel might have to be blank.
What is this plink.exe timeout? Try setting it to a higher value like a minute?
If no combinations of this works I'm out of ideas. (And since I don't use Windows I'm unable to test it for myself) --Zhuyifei1999 (talk) 17:01, 2 May 2018 (UTC)
I notice that Wikimedia software doesn't play midi files. What would be the likelihood of adding an extension which converts the file to ogg/mp3 on the backend, allowing modern browsers to play the file without a download? Magog the Ogre (t • c) 18:30, 29 April 2018 (UTC)
I understand that having to open sound files separately is also a problem for Wikivoyage, Wikibooks, and Wikiversity contributors. Just being able to play the audio file in a phrasebook or other language-learning content, without having to open a separate page, would be very handy. Maybe something to remember for next year's m:Community Wishlist? Whatamidoing (WMF) (talk) 17:55, 30 April 2018 (UTC)
MIDI files ain't audio files, but 'sequence' files (think of them as 'advanced' sheet music); there is no 'canonical' translation from a set of instructions for musical instruments to an audio waveform. Yes, certain softwares are able to do the synthesis and play the music (TiMidity++ is one such software in the FOSS world), but given the complexity, it's best to store both the 'instructions' (for what notes on what instruments are being played) and the waveform (for a canonical audible representation of the music).
And no, MIDI can't store pronunciations so it should not be relevant in a phrasebook or other language-learning content --Zhuyifei1999 (talk) 17:53, 1 May 2018 (UTC)
The iPhone doesn't support MIDI playback. I tried MIDI.js which looks to be an in-browser synthesizer, but it did not work either. — Dispenser10:26, 2 May 2018 (UTC)
I noticed a recent change to the icons for featured content. In most places, including the top-right corners of articles, the icon in use is this: (this matches the first gallery icon below), but in some places it recently changed to this: (this matches the second gallery icon below). There is a clear difference between the two when placed side-by-side. This would not be a problem, except the change has not been implemented across the board (see the table in Template:icon). The commons links for the files themselves are here:
In my opinion, there should be consistency with which featured icons are used and to that end, I believe the former icon above should replace the latter icon above, as the former has a higher definition and looks better in icon form. Jackdude101talkcont04:20, 2 May 2018 (UTC)
I also think that the icons should be consistent, although this has nothing to do with the resolution of the images, since they're SVGs and can be resized virtually infinitely; I have no strong preference for either icon. Jc86035's alternate account (talk) 10:54, 2 May 2018 (UTC)
In that case, agreed - if there is no clear difference between the two, but the former is used more widely then we should standardise on it. Richard061218:59, 2 May 2018 (UTC)
Reference archiving bot
Are there any bots dedicated to archive URLs in reference templates such as "cite web", in a platform in the like of Web Archive? Has this been tried? Leefeni de Karik (talk) 03:21, 3 May 2018 (UTC)
Location of Wikimania 2018—Cape Southern Sun Hotel
English Wikipedians can now use the mapframe function to embed maps right on a page. The map at right (which shows the location of this year’s Wikimania conference) demonstrates the functionality. Mapframe was requested by the English community via RfC last year and the request was reconfirmed this spring. If you you’re new to mapframe, this Kartographer help page shows how to use it.
Expect more significant new features in the coming weeks, with the implementation of the long-awaited map internationalization features. After internationalization, maps will display in the content language of the wiki they’re published on. Until then, they will continue to display in the language of the territory mapped. (You can experiment with internationalized maps on testwiki now—here’s a page of examples.)
The credits aren't watermarked into the image (see this image or this one). They're positioned on top of the bottom-right corner of the image by the browser, obscuring a corner of the image. If you download the image, it won't contain credits or anything like that. --Roan Kattouw (WMF) (talk) 04:48, 1 May 2018 (UTC)
Oh, I thought you meant the placement/floating issue of the map compared to the other elements. If you want an area highlighted you need to use a line or shape feature rather than coordinates. And |plain=yes will set frameless mode, removing the outer frame and caption. Or if you want more complicated things (Template:Maplink#Things_this_template_can't_do), you might have to use the <mapframe>...</mapframe> tags directly. - Evad37[talk]04:41, 1 May 2018 (UTC)
This is getting closer:
{{maplink|frame=yes|plain=yes|frame-width=250|frame-height=280|type=shape-inverse|zoom=7|frame-lat=29.185|frame-long=76.225|stroke-width=3|id=Q1174}} - Evad37[talk]04:54, 1 May 2018 (UTC)
Added a smaller stroke-width. Before adopting into infoboxes, the template should really support adding map data from multiple objects – e.g. an outline of an area plus one or more markers at various points. To do so in a sane way really require it to be rewritten in Lua - Evad37[talk]09:30, 1 May 2018 (UTC)
What is the easy way to localize a name in OSM? Is there any plans? Many places doesn't have Bengali localization, so last week i tried to have add some Bengali name. After so many steps i finally managed to add a translation & i took one minute to complete just for one place. আফতাব (talk) 22:21, 1 May 2018 (UTC)
I've edited this page a lot, which can be seen at [24]. But it seems to me that something in the 2 "top ten" piecharts is somewhat off, my % doesn't match my piece of pie. Any good reason for this? Gråbergs Gråa Sång (talk) 17:35, 3 May 2018 (UTC)
I believe the percentages are percent out of total — so for example, you made 51 edits out of a total of 103, or 49.51& — but the pie charts are out of the top 10 — so 51 out of 90, or 56.67%. If you mouseover the pie chart slices, they display numbers (51, 9, etc.) not percentages. It's a little weird, perhaps, but it certainly makes the charts nicer. ~ Amory(u • t • c)18:02, 3 May 2018 (UTC)
The character icons are all changed. I don't think the Template Wizard people are responsible because I have been following the implementation of that tool, and this change is more recent. L3X1◊distænt write◊14:13, 3 May 2018 (UTC)
@L3X1: can you be more specific about "character icons", perhaps a screen shot about which component you are referring to? — xaosfluxTalk14:21, 3 May 2018 (UTC)
Hello everyone! I'm Trevor, a product manager at the Wikimedia Foundation. My team, the Anti-Harassment Tools team, is working on a project to build page, namespace, category, and upload blocking. We see the full site block as a sledgehammer, when some user misconduct situations call more a more tactical response.
In general if you search google in a foreign language it will return results in that language. For one of your examples, "Гуьржех" the #2 hit is w:ce:Гуьржи-Мохк. If you specifically ask google for enwiki (with Гуьржи-Мохк site:en.wikipedia.org as your search term) it will return a hit here. — xaosfluxTalk01:08, 4 May 2018 (UTC)
I've had occurrences of this in the past and it's almost always been some welcome message on my user talk on a foreign language wiki that I have never edited or am unlikely to edit. For some reason the notification doesn't always show but the counter does. Switching cross-wiki notifications off and on again in preferences normally sorts it out. Nthep (talk) 07:39, 4 May 2018 (UTC)
Prevent auto edit launch
During the Twinkle rollback bug issue (above) I reset all my settings. Now when I rollback and go to warn the User, the User TalkPage automatically launches in edit mode and not the static (read) mode I am used to. This also happens when I click on non-existant pages/redlinks. Any way to prevent this? Sorry if this is not the right forum to post such a question, let me know. Best, Classicwiki (talk) If you reply here, please ping me.11:25, 4 May 2018 (UTC)
How to deal with URLs that change as you click them
Note the line feed at the end (%0A) the tabs (%09) the quotes (%22) .. in the fragment portion (#) of the URI.. quite a specimen :) Maybe that's how WikiData is supposed to work. Anyway, both URLs above work for me regardless of "_" or "%20" maybe try clearing browser cache (or QuickStatements has since been fixed). -- GreenC18:35, 3 May 2018 (UTC)
User:GreenC thanks for looking into it. This URL is how one communicates with the tool that interacts with Wikidata. It has little to do with wikidata itself. Yes it is quite a specimen as it tries to encode in a URL wide range of commands and inputs. You said that both URLs above work for you. However I could not see in the history of Wikidata Sandbox (Q4115189) any evidence of your edits. You actually have to press "run" button (and possibly log in through oAuth) to test it. --Jarekt (talk) 19:34, 3 May 2018 (UTC)
Ahh, I didn't get that far as I don't have an account. The percent-encode for "_" is %5F does it work if substituted? -- GreenC21:11, 3 May 2018 (UTC)
Wikitext syntax highlighting, also known as CodeMirror, has been moved out of Beta Features and is available in the 2017 Wikitext Editor on all wikis. Syntax highlighting helps you see problems in your wikitext before previewing or publishing text. Please try out the tool if you did not do so while it was being developed, and feedback is welcome. - Keegan (WMF) (talk)
I've seen reports from Huggle, and just replicated with AWB. Unable to login via the API using passwords or botpasswords. Is anyone able to get in? api returning "bad password" type errors. — xaosfluxTalk03:16, 4 May 2018 (UTC)
FYI, after regeneration of botpassword, I was able to get an account on - yet another account that was not regenerated is still unable to logon with the existing botpassword. — xaosfluxTalk03:28, 4 May 2018 (UTC)
I would think that not as a technical issue but a security issue - you know, the login problems.--1233Talk03:31, 4 May 2018 (UTC)
This could relate to the reports above and elsewhere concerning the current account cracking attack. A threshold of attempts per account might be exceeded, or a dev might have imposed some throttle? Johnuniq (talk) 04:58, 4 May 2018 (UTC)
We haven't done anything in response to the mass-account-failed-logins that would affect botpassword logins by normal users. Possibly people have been changing their passwords (Always a good thing to do now and again) which might reset your botpassword tokens. BWolff (WMF) (talk) 20:37, 4 May 2018 (UTC)
No solid finds in the phab ticket, other then identifying that invalidation of botpasswords does not present notification (yet). Anyone with this issue should be able to resolve by regenerating your Special:BotPasswords by going there, clicking yout botpassword name, then checking reset password and saving. You will need to update your tools/bots with the new password. — xaosfluxTalk21:11, 4 May 2018 (UTC)
Test for the non-existence of a (possibly empty) third positional parameter?
Resolved
In a template, how do I test for the non-existence of a (possibly empty) third positional parameter?
In this case the template takes only a max of 2 parameters and third indicates a possible error even if empty. So I want to treat any of these as an error
{{Mytemplate|foo||bar}}
{{Mytemplate|foo||bar}}
{{Mytemplate|foo|bar|}}
... but be happy with
{{Mytemplate|foo|bar}}
{{Mytemplate|foo|}}
{{Mytemplate|foo}}
{{Mytemplate||bar}}
I thought I coud take some test which would always throw an error if {{{3}}} was not supplied and use that "error" as my sign of health.
But now I can't find any such test.
I think I have done this before, but my brain seems mushy and I can't recall how I did it.
With the example that I gave you, all of those testcases you mentioned in tour initial post, would pass. If you want to show content of third parameter (if it's not empty) (that you didn't mention in initial post), then you can use additionally: {{#if:{{{3|}}}|Here are contents of third parameter: {{{3|}}}}} What is your use-case? Some real-life problem/template would be good to show. --Edgars2007 (talk/contribs) 05:58, 5 May 2018 (UTC)
Thanks for trying, Edgars2007 ... but I have already demonstrated[25] how the test you propose gives a false positive.
If I understand the original question, Module:Check for unknown parameters may be what you want. You can put {{#invoke:Check for unknown parameters|check|unknown={{main other|[[Category:Pages using TKTKTK with unknown parameters|_VALUE_{{PAGENAME}}]]}}|preview=Page using [[Template:TKTKTK]] with unknown parameter "_VALUE_"|ignoreblank=n| 1 | 2 }} at the end of the template code to categorize any such articles. – Jonesey95 (talk) 14:21, 5 May 2018 (UTC)
What you wanted to hear about this topic on this page? From technical POV, it is possible. From non-technical POV (which of course would be better suited for other pages), I think it's OK. There are many examples on Wikipedia, person who is WWE and actor, e.g. --Edgars2007 (talk/contribs) 03:59, 5 May 2018 (UTC)
Linking articles in different languages with different names
I noticed that the English-language article on the Finnish architects Reima and Raili Pietilä was not linked to its respective articles in 10 other languages (incl. Finnish, Korean, etc.). However, in ALL the other languages the site is called just "Reima Pietilä". Reima and Raili Pietilä were an architect couple who nearly always worked together. But when I attempted to add the English-language version to the other language versions, it failed to so. I suppose I will have to change the name of the English-language site to be the same as the others in order for the language links to work? This bothers me, because in including both Reima and Raili Pietilä as an "architect couple" (cf. Charles and Ray Eames) the English-language page comes across as more enlightened.TTKK (talk) 15:52, 4 May 2018 (UTC)
Three things you could do: The first is to manually add old school local links manually to the article. See WP:ILL § Local links for more info and nb the spirit of the second and third points. However, this will allow linking only to the Reima or the Raili article. The second would be to merge wikidata:Q17496278 into wikidata:Q2582922, but then you would lose "duo"/"married couple" attributes in Wikidata. The third, as you mentioned, would be to rename/split this article like five other Wikipedias do. This would allow linking to both the Reima and the Raili articles.
I will conduct the first option on the English page now. Review and feel free to edit revert. I also look forward to hearing from others who might know more about this topic. — AjaxSmack00:30, 5 May 2018 (UTC)
No, the second option is bad, very bad. Those are different topics, so they shouldn't be merged. About problem in general. I would suggest creating module (that is, copying from other Wikipedia - don't remmeber which), which takes iws from Wikidata item, that you pass as parameter. Pro - then iws are updating, not as in old-iws mode. --Edgars2007 (talk/contribs) 04:06, 5 May 2018 (UTC)
Get some Lua guy, that writes you a module to do that (it shouldn't be very hard, but I'm not a Lua programmer). Or find the module on some other Wikipedia (don't remember which). --Edgars2007 (talk/contribs) 06:03, 5 May 2018 (UTC)
Thanks. It looks like this was copied from Swedish Wikipedia and can do it: Template:Interwiki extra. Not ideal as it cannot allow for links to more than one article (i.e., both Reima and Raili), but better than no IWs at all. — AjaxSmack05:11, 6 May 2018 (UTC)
Tools to comment/uncomment categories?
When you move something from article space to draft/user space, you want to disable the categories with ":" prefixes. And do the reverse in the other direction. Is there some nice bit of automation which makes this easy? -- RoySmith(talk)17:41, 5 May 2018 (UTC)
A category with sortkey like [[:Category:Words|X]] would display as X. I have made {{Draft categories}} which avoids this issue, permits lowercase c in "category:", imitates how categories are displayed, and actually adds the categories if the page is in mainspace. Example:
Maybe the imitation is too good and should give an indication that the page is not actually in the categories. PrimeHunter (talk) 23:00, 5 May 2018 (UTC)
I've added a sentence about the categories not yet implying, feel free to improve my wording, and more discussion about the template should probably go to the template's talk page. rchard2scout (talk) 12:12, 6 May 2018 (UTC)
Recent change breaks scroll in p-cactions menu
A few days ago I added #p-cactions {overflow:scroll} to my Modern.css, as there are times when the links overflow the screen and I'd like to access them. It worked just fine, up to and including this morning (7:30AM EDT). I came back on around 11:30 EDT, and now there is a white bar obscuring most of the p-cactions menu bar. What changed in the intervening four hours? Adding that css code is sufficient to add the bar, and it happens here in safemode as well as other WMF wikis. Theoretically it affects vector, although obviously the dropdown menu means it just affects the appearance. It happens regardless of being logged-in or logged-out, and in Firefox, Chrome, and Safari. I first thought it was changed in the new software, but Special:Version says mw:MediaWiki_1.32/wmf.2 was installed days ago. Can't for the life of me track down what changed, figured I'd check here before going to phabricator. ~ Amory(u • t • c)16:34, 6 May 2018 (UTC)
I guess it's the personal navigation links from userpage to log out you want full access to in Modern. They wrap in Vector. I don't know how to get scroll to work but can you live with a smaller font instead, e.g. 70%:
No, it's the #p-cactions menu (edit, history, delete, twinkle, etc.) It's the x-axis scroll bar blocking things; using overflow-y: scroll solves the issue. overflow: auto also works. I just can't figure out why it changed. ~ Amory(u • t • c)18:29, 6 May 2018 (UTC)
Page translation
Could someone please walk me through page translation? I made a small change to a template and now Wikimedia is trying to walk me through some byzantine translation process even though I didn't change anything translation related, and I even clicked "do not invalidate translations". Magog the Ogre (t • c) 18:44, 6 May 2018 (UTC)
In short, you edit the main template, and mark it for translation - you should be done now. Normally a script/bot updates the other languages. Is there a specific one not working? — xaosfluxTalk19:32, 6 May 2018 (UTC)
Gadget issue
Now, I don't know if it's just because I'm using the black background with green text or what, but I have "Add two new dropdown boxes below the edit summary box with some useful default summaries" selected in gadgets, and the text is very light, so it's hard to see against the white background the same goes with the upper bar at the top of the edit box. screenshot(talk page stalker)CrashUnderride22:58, 6 May 2018 (UTC)
Autorollbacker userright
What would you think of creating an autorollbacker user right that would automatically revert one's edits for a given time? Maybe it could be limited to rollbacking only non-autoconfirmed users and IPs to prevent abuse. If there is consensus to create this user right, I might start an RfC on it. Your suggestions would be welcome. Best, L293D (☎ • ✎)16:14, 6 May 2018 (UTC)
You want to create a userright that automatically reverts any edit a user with that right makes? Am I understanding you correctly? ~ Amory(u • t • c)16:36, 6 May 2018 (UTC)
No. The right I am proposing would enable a trusted user to automatically rollbacke persistent vandals, so that they don't have to refresh their contribs every minute and revert. (To save time to CVU users) L293D (☎ • ✎)16:38, 6 May 2018 (UTC)
@L293D: Re "server side" means this isn't something that can be done today, and either you would need to change mediawiki core, or write and have us adopt an extension for it; with blocking available I don't see either of these methods getting adopted. If you were thinking about something else on how the mechanics would work, please elaborate. — xaosfluxTalk16:51, 6 May 2018 (UTC)
@L293D: I think there are 2 better options: (1) become an admin, (2) revisit that "vandal fighter" type access with includes rollback, block, delete (but not undelete) - that gets opposed with "be an admin" arguments most of the time.. — xaosfluxTalk17:16, 6 May 2018 (UTC)
Well, for option 1, I have no chance of passing RfA anytime soon, but the "Vandal Fighter" right might be worth another RfC. If the proposal fails, then it would be nice if someone could write me a script that would rollback automatically, since I have almost no knowledge of Java. L293D (☎ • ✎)17:25, 6 May 2018 (UTC)
Please do not edit war. Not even with a troll. And certainly do not use an automaton to edit war. Report to WP:AIV or, for more urgent cases, also report at WP:ANI. Johnuniq (talk) 23:11, 6 May 2018 (UTC)
I said I would be used against vandals while they await blocking. But as you said, the tool might be abused to edit war, so this is why I suggested that the script would accept autorollbacking only users who are reported to AIV. L293D (☎ • ✎)13:45, 7 May 2018 (UTC)
Any client side scripts can be fully controlled by clients - this is just not technically feasible to work how you describe it, plus you shouldn't revert war with anyone. — xaosfluxTalk14:00, 7 May 2018 (UTC)
Latest tech news from the Wikimedia technical community. Please tell other users about these changes. Not all changes will affect you. Translations are available.
Recent changes
The Wikimedia Commons mobile app has a new version. It is now easier to find nearby places that need pictures. It helps you with direct uploads and title and category suggestions. The app only works on Android phones. [26]
Problems
The abuse filters had a problem with blocks where you had changed how long they last. It used the default length everywhere. This was in late April. Abuse filter users should make sure the right block length is used and change them if needed. This is only for filters where how long blocks last had been changed. [27]
Changes later this week
The advanced search function beta feature will be on all Wikimedia wikis. It makes it easier to use some of the special search functions that most editors don't know exist. [28]
The new version of MediaWiki will be on test wikis and MediaWiki.org from 8 May. It will be on non-Wikipedia wikis and some Wikipedias from 9 May. It will be on all wikis from 10 May (calendar).
Meetings
You can join the next meeting with the Editing team. During the meeting, you can tell developers which bugs you think are the most important. The meeting will be on 8 May at 18:30 (UTC). See how to join.
You can join the technical advice meeting on IRC. During the meeting, volunteer developers can ask for advice. The meeting will be on 9 May at 15:00 (UTC). See how to join.
Future changes
All wikis with fewer than 100 high-priority linter errors in the main namespace will switch to use the Remex parsing library. This is to replace Tidy. It will happen on 16 May. Other wikis will switch soon when they have fixed the errors that must be fixed. Tidy will be removed on all wikis before July 2018. [29][30]
@Headbomb: I'm fairly certain that search will not return redirects when using a keyword like incategory (this search doesn't work). The documentation at Mediawiki wiki indicates that this is definitely true for some parameters, but doesn't say so about incategory. I did use WP:PETSCAN to get all the pages and then exported it to CSV (I couldn't get petscan's title regex to work), where I filtered the page title column in LibreOffice Calc. That might be your best shot. --Izno (talk) 05:20, 5 May 2018 (UTC)
@Izno: The idea would be to have something that can search for this live on the wiki, that I can put in a search box on Category:Redirects from ISO 4, similar to (but not quite the same as) the search box in Category:Infobox templates. The CSV file would be mostly pointless here. I tried insource:"R from ISO 4" or hastemplate: "R from ISO 4" to find the templates used to populate the category instead, but that ignores redirects too. Headbomb {t · c · p · b}05:27, 5 May 2018 (UTC)
Ah, then you might be SOL. You can try filing a phab task to get the attention of the search guys to see if they know a way to do that. --Izno (talk) 05:39, 5 May 2018 (UTC)
Is there a way to display a random picture of the day along with the corresponding caption?
Also, how can I set the {{rand}} template to restrict the numbers it picks. For instance 5-6? Also how can I set it to generate a leading 0 before any number smaller than 10 i.e. 01 not 1. --Tyw7 (☎ Contact me! • Contributions) 10:42, 8 May 2018 (UTC)
From May 8, AdvancedSearch will be available as a beta feature in your wiki. The feature enhances the search page through an advanced parameters form and aims to make existing search options more visible and accessible for everyone. AdvancedSearch is a project by WMDE Technical Wishes. Everyone is invited to test the feature and we hope that it will serve you well in your work!
Looks nifty! I suppose I'll find out tomorrow, but will this affect how search results are displayed in any way, or will it leave those unaltered? ~ Amory(u • t • c)15:44, 7 May 2018 (UTC)
Thanks :-) - no this will not affect how search results are displayed. Basically, AdvancedSearch serves as an interface for some of the special search options that are already provided in CirrusSearch, but are difficult to find or to remember - especially when you want to combine several of them. E.g. you would get the same results if you manually type intitle:foo into the search field or use advancedSearch for that. The main advantage of AdvancedSearch is: The existing options become way more visible, users don't have to know the syntax behind each search field, different parameters can be combined easily, and you can decide if you want to learn more about the search syntax (or just use it). - In case you test it in the next days - we're really interested in hearing how it works for everyone :-) Birgit Müller (WMDE) (talk) 14:24, 8 May 2018 (UTC)
Update on page issues on mobile web
Hi everyone. The Readers web team has recently begun working on exposing issue templates on the mobile website. This follows a proposal from 2016. Currently, details about issues with page content are generally hidden on the mobile website. This leaves readers unaware of the reliability of the pages they are reading. The goal of this project is to improve awareness of particular issues within an article on the mobile web. We will do this by changing the visual styling of page issues.
One way to fix this is to remove all redirects from Wikipedia:Good articles/all. Then when I create the mismatch list they will not appear. Is there an automated way to fix redirects at Wikipedia:Good articles/all. I know WP:NOTBROKEN, but in this case it kind of is. Email me if you are worried about WP:Beans. Failing that do you know an easy way (i.e. a tool or AWB function - not something that requires a bot) to remove the redirect from the bottom list if it leads to an article in the top list. The article from the top list should also be removed at the same time. I would rather not use a bot as it means I can't keep it updated (and we have enough trouble with AWOL bots at GA at the moment). AIRcorn(talk)19:21, 8 May 2018 (UTC)
@Aircorn: There is at least one gadget, and the possibility exists in your CSS, to display redirects in a different style than the default. That would get you most of the way to knowing which are the redirects, and removing or editing them as appropriate. --Izno (talk) 20:35, 8 May 2018 (UTC)
Thanks Izno. I have it enabled so the redirects show light green, which is how I came up with the 400 number. It does help, but there are a few issues. Not all redirects match up and some are genuine mismatches that need to be removed. This usually happens when an editor merges or redirects a Good Article to another article. The other problem is that for me it doesn't highlight the target article. I need to identify both so I can remove them from both lists. AIRcorn(talk)22:47, 8 May 2018 (UTC)
@Aircorn: I use the CSS to check for redirects in navboxes--the way I know what the actual target article is is with either POPUPs (a second gadget) or Page Previews, which I think are in preferences under the Reading tab now (I use the Page Previews). --Izno (talk) 00:53, 9 May 2018 (UTC)
Bidirectional isolation bdi
I wrote a module at Commons that has worked well for 8 months. Now the links that it generates are broken. Like the template that it replaced, the module outputs links for navigation at Commons and it uses <bdi> (Bidirectional Isolation element) to wrap the label of each link so the browser can decide whether right-to-left text is needed. The first line that follows is an example of the output wikitext. The second line is the same with bdi removed.
The first line used to give a link just like the second line. Any idea what has changed that means it now doesn't? I merely copied the replaced template (which had performance problems) and don't know if bdi is necessary. Should bdi be used? The issue was raised at Commons here. Johnuniq (talk) 23:11, 8 May 2018 (UTC)
A note re debugging: I tried a template with the problem at c:Special:ExpandTemplates and it worked correctly, whereas previewing the same template in a sandbox had broken links. I guess ExpandTemplates does not use Tidy and knowing that might be helpful if debugging stuff like this. I worked around the problem at Commons by temporarily removing bdi from the output. Johnuniq (talk) 03:38, 9 May 2018 (UTC)
Has there been a change in the way navboxes behave? I'm sure that previously if I viewed a navbox on a page which is linked in the navbox, the link to the host article was disabled and displayed in bold but now that behaviour doesn't happen and a clickable link remains. Nthep (talk) 14:47, 9 May 2018 (UTC)
Definitely not a redirect and I realise now it might be a skin issue, I'm using Modern and a check shows no issue on Vector or Monobook. Nthep (talk) 14:59, 9 May 2018 (UTC)
It is a skin problem due to a change in the core parser. The task DJ links below applies to the same issue in Timeless; there should be a separate patch (and/or task submitted) for Modern. --Izno (talk) 16:33, 9 May 2018 (UTC)
Template:Str find/Module:String allows you to search for a string in a string. Is it possible to have them search for multiple strings in a string? Or to code something in LUA that would have an equivalent functionality?
So how would I structure the module invocation for that? {{#invoke:String|find|SOURCESTRING|REGEX}}? Because I don't see how a regex or (|) would play nice here. Headbomb {t · c · p · b}18:44, 9 May 2018 (UTC)
All of the Module:String functions will use Lua patterns but Lua patterns don't support the alternation (|) operator. In Lua itself you would 'or' together the several string library functions to get the equivalent:
(string.match (source_variable, '(Medicine)') or string.match (source_variable, '(Oncology)'))
It's hard to help without knowing what you are trying to do. I guess some of your pipes should say {{!}} instead. That's a way for some features to display or return a pipe without the pipe being interpreted as part of the syntax for the feature itself. See mw:Help:Magic words#Other. PrimeHunter (talk) 22:20, 8 May 2018 (UTC)
The goal is to use parameters (here "fixed", "TeraScale" and "GCN") to control the table layout. More precisely when the template is called like this {{User:Wikiinger/sandbox/AMD GPU features|fixed=0|TeraScale=0|GCN=1}} only the rows belonging to the GCN instructions set (the 5 most-right) and the left-most row should be displayed.
However it is failing much more early... From my understanding {{{GCN|1}}} should default the GCN parameter to 1. Next {{#ifeq: {{{GCN}}} | 1 |! [[AMD RX Vega series|Vega]]|}}, should "print" ! [[AMD RX Vega series|Vega]] if GCN=1. But one (or both) of these statements seem to fail...Wikiinger (talk) 16:21, 9 May 2018 (UTC)
Both your statements are correct. I said I guess some of your pipes should say {{!}} instead. I took a guess at which ones.[31] Does that help? It sounds like you are confusing rows and columns. Rows are horizontal and columns are vertical. There is no "left-most row". If you want to hide a whole column then you have to hide each cell in the column. PrimeHunter (talk) 20:39, 9 May 2018 (UTC)
Something to note: 'ifeq:' is for comparing string functions, and might get confused with numeric comparisons. Try a simple "if:" instead. E.g.: {{#ifeq: {{{GCN|0}}} | {{!}} Jan 2012}}. Note that logical value of both a supplied string value and a digit "1" are "true". ♦ J. Johnson (JJ) (talk) 21:19, 9 May 2018 (UTC)
I think you mean {{#if: {{{GCN|}}} | {{!}} Jan 2012}}. #if: is true for any non-empty string including the string "0", so {{#if: {{{GCN|0}}} | {{!}} Jan 2012}} returns {{!}} Jan 2012 if GCN is not set. PrimeHunter (talk) 21:36, 9 May 2018 (UTC)
As I sit here reading a notification that there has been a failed attempt to log in to my account, I find myself wondering how it is the site I spend the most time on is the only site I spent much time on that doesn't have multi-factor authorization. I'm quite sure this has come up before, but apparently I don't know a good search term for the VP archives. Now, I have a pretty strong password, but it's unsettling nonetheless, as it most certainly was not me who tried to log in. — Rhododendritestalk \\ 13:34, 3 May 2018 (UTC)
Thanks. As I'm not an admin, it doesn't look like there's anything I can do at this point. Reading over the meta page, however, I wonder what reason there is for this to be seemingly more involved than other systems. Why not just use the texted code like everyone else? I'm sure that would require a little investment from WMF, but this sure seems like a good use of its cash... — Rhododendritestalk \\ 13:50, 3 May 2018 (UTC)
Texted codes are not all that secure, so the privacy implications of storing people's phone numbers and the expense of an SMS gateway don't seem to be worth it. Others, such as Google and NIST, are moving away from SMS-based 2FA. Anomie⚔12:37, 4 May 2018 (UTC)
Also, though it's not something typically offered with this sort of mechanism, it might also be useful to determine the IP that tried to log in. In this case, the notification came shortly after my first edit of the day, this revert. With an unregistered user, there's not a whole lot of point except as an additional reason to block, but it makes me think that registered users might do it, too, in which case that information would be important if confirmed via checkuser. Getting into a whole can of worms there, I know. — Rhododendritestalk \\ 13:56, 3 May 2018 (UTC)
@Rhododendrites: if you really really really want it, you can request tester access at meta SRGP - there is no good way to manage these and you could lose access to your account forever. Some recovery options are being discussed on some of the tasks I linked to in the phab workboard. As far as the logon notifications, much discussion is in place regarding improving these and including more information that may be useful (see phab:T174388). — xaosfluxTalk14:24, 3 May 2018 (UTC)
Rhododendrites, I just also got one of those (for the first time in 11 years). Does WP record the IP of these attempts? If so, is this information available for discussion at a venue like this? Why should anyone try to do this when it's pretty obvious I'm logged in and editing? Perhaps it's some kind of stupid bot?| Thanks. Martinevans123 (talk) 14:51, 3 May 2018 (UTC)
This type of notification has only been available for about a year. Right now it is informational only, bots try to hack accounts all the time. — xaosfluxTalk15:09, 3 May 2018 (UTC)
Do they try to hack my account all the time? Or have they tried all the time in the last year? Or are you saying that bots won't trigger this warning? Thanks. Martinevans123 (talk) 15:23, 3 May 2018 (UTC)
Note, the bots I refer to are not "wikipedia bots" - these are malicious parties on the internet; and they try to hack lots of accounts all the time. I got a failed logon notification myself today. For the most part, you don't need to do anything on these - if you find the notifications annoying, you can turn them off in preferences. — xaosfluxTalk15:34, 3 May 2018 (UTC)
Understood. From your last reply it seems like you are assuming the notifications that all three of us got were triggered my malicious bots. I'm quite happy that I don't need to do anything (although if I start to get one every day, I might do something anyway). I'm just wondering how often other folks get these notifications, whether IP's are recorded and if anyone does anything as a result. Sorry is this has careered off topic. Thanks. Martinevans123 (talk) 15:45, 3 May 2018 (UTC)
A lot of other people have received this notification, it doesn't seem to stop, even on frwiki we receive a notification from enwiki. See [32]. Lofhi (talk) 16:09, 3 May 2018 (UTC)
That's a very interesting graph! That looks quite cataclysmic. Can the WMF issue any information on where these attempts are being made? Martinevans123 (talk) 19:11, 3 May 2018 (UTC)
Unless any of the attempts are succeeding, I'm guessing someone's re-using the list that got Jimbo and some other admins hacked last year. If you've changed your password in the last few months then you should be safe. Ian.thomson (talk) 19:19, 3 May 2018 (UTC)
Wait, I see that there was a new data dump in February that seems to have built off of the one from the Anti Public Combo List and other breaches. I haven't grabbed it yet, and hopefully I'm right that if no accounts have been compromised so far, then it's probably older passwords. It'd be a good idea to check haveibeenpwned.com. Ian.thomson (talk) 20:02, 3 May 2018 (UTC)
Yep, I got one (my first) about a half-hour ago, too. Does the Foundation take note of bursts of failed login attempts? TenOfAllTrades(talk) 16:31, 3 May 2018 (UTC)
Similarly to TenOfAllTrades, I've just got my first such notification ever where it hasn't been me on a different device. Is this sort of rate common or is something going on? Red Fiona (talk) 17:37, 3 May 2018 (UTC)
Question: if I simply mistype my password and fail to log-in, will I get another one of these messages? Or does it look at the device used as well? Thanks. Martinevans123 (talk) 10:54, 4 May 2018 (UTC)
Thank you for that clarification. I suppose some users may now be tempted to try this on their own accounts just to test the system. Is there any reason they should not? Martinevans123 (talk) 13:11, 4 May 2018 (UTC)
password resets failing
This seems to be some sort of mass password reset troll. At threads at both WP:AN and WP:TEAHOUSE users are reporting that after receiving the notification they are unable to actually reset their password. One user report trying 17 times and failing. Software glitch? Beeblebrox (talk) 18:39, 3 May 2018 (UTC)
@Beeblebrox: They may be getting the message "A password reset email has already been sent, within the last 24 hours. To prevent abuse, only one password reset email will be sent per 24 hours." If so, this is expected as it has a throttle on it, there is some discussion of changing this at phab:T181070. — xaosfluxTalk18:53, 3 May 2018 (UTC)
I can verify that I have not received that message as I don't have email. The only messages I get when I try are the standard "Incorrect username or password entered. Please try again." except once when it told me to wait for five minutes. -- Millionsandbillions (talk) 19:08, 3 May 2018 (UTC)
I really hope everyone changed their email password at least once (in addition to their WP password) after that incident where Jimbo got hacked. Ian.thomson (talk) 19:21, 3 May 2018 (UTC)
Just changing a password makes no difference, however many times you do it. You need to change it to something stronger? 86.187.160.63 (talk) 07:50, 4 May 2018 (UTC)
If whoever is doing this is using a list of passwords from some other sites (which is generally how accounts get compromised here), then yes, changing the password is going to help. If they're going after as many accounts as possible, and most of us are only getting one failed login attempt, then it's save to assume they're not taking the time to brute force every possible combination (which is where password strength matters). Ian.thomson (talk) 14:48, 4 May 2018 (UTC)
Illicit signin
I just got a message saying someone has tried to sign in to my acc from an unrecognized device. So I changed my password. Thing is - NOBODY can access my stuff - except maybe someone who is on my network. I share my hub with some neighbours; it would be immensely helpful to me if sysops could tell me if the interloper was "coming from inside the house". Sysops know my address - was the interloper from the same address? MarkDask19:27, 3 May 2018 (UTC)
I got the same notification today, which was first time ever. So either someone is trying all usernames with "123456", "qwerty", "password", and other common passwords ow someone is testing new notification features. --Jarekt (talk) 19:39, 3 May 2018 (UTC)
@GregorB: I guess they don't like me as I haven't received any such notifications so far today, but yes, it appears to be a widespread and planned attack, not just one of those every now and then things. Amaury (talk | contribs) 21:02, 3 May 2018 (UTC)
Thanks all. I think two factor authorization on here would be a step too far. That said - I dont like being told someone might be trying to hack my space so I've dissed all my hangers-on. (Sigh) - I've taken back control of my hub - (screaming in the background). MarkDask
There's a thread on this happening at WP:AN I got a notice of a failed attempt to hack my account. Others are getting it also. I already have the 2-factor enabled, but I also changed my password after the failed attempt.— Maile (talk) 00:01, 4 May 2018 (UTC)
From what I’ve been told, there were 60,000-70,000 failed logins per hour for several hours. So, tens of thousands of users got these messages, but they do not represent an actual security breach, just some troll with a little bit of programming knowledge that set up a trollbot.
Good morning. I wonder if you could help me out please? Approximately 9 hours ago (9pm UTC, May 3rd) an attempt on this Wikipedia was made to login to my account. The wrong password was used, thankfully, and I got flagged. Does Wikipedia record logins and where they originate from please, and if so, is there anything you can do? Obviously you can't tell me who it was, but I'm curious as to why someone would attempt to access my account. Thanks. Dane|Geld05:49, 4 May 2018 (UTC)
I and my fellow mates on Marathi wikipedia too have got this notification from English Wikipedia. I think we must also have a two factor authentication like gmail on wikipedia. --✝iѵɛɳ२२४०†ลℓк †๏ мэ13:37, 4 May 2018 (UTC)
Detecting suspicious failed logins
Is there anything in the software that would notice if one IP address were making a lot of failed login attempts to different accounts? (I got an email that here had been an attempt to log into my account from a new computer and to make sure I had a strong password. A lot of people have commented on getting this message. If it's part of a lot of attempts has noticing it been automated?) RJFJR (talk) 14:12, 4 May 2018 (UTC)
RJFJR Moving this thread up to appropriate section. See all above, been happenig to everybody.Two sections up, Diannaa has said the attack is still happening, and included this link: this graph— Maile (talk) 14:33, 4 May 2018 (UTC)
My understanding is that technically this isn't possible. It's not possible to see the burglar until he gets through the front door with his counterfeit key? Am surprised though. Martinevans123 (talk) 08:28, 5 May 2018 (UTC)
I changed my password, and now I can't log in
Resolved
Hey, i received the same message two days ago and changed my password. The next day i was logged out and my new and old password don't work. I requested a new password to my actual mail adress but that was not the one, which was listed under "preferences" and of course none came. The one adress listed is dead. Any chance to get my old account (Kante4) back? Kante44 (talk) 08:52, 5 May 2018 (UTC)
Is there a log of my login attempts, a log I can see? Or can someone please check that log for me? I was notified that about 5 days ago a login attempt failed. Then I was notified by an email stamped Sunday for 8:01p that someone, probably I, had "recently" logged in. I did log in Saturday night from a network I haven't used in a while and logged off some time around 2 a.m. Sunday, but not at the time the email was sent. But the email says "recently", so maybe you meant my Saturday night login. My Special:Contributions end with a contribution I made; no one contributed under my username since then. I since changed my already-secure password to a still-more secure one. Is there a way to find out the time of the last login under my username last Sunday (if any) (not a login for Monday or since)? Thank you. Nick Levinson (talk) 21:11, 8 May 2018 (UTC) (Correction: Forgot the tildes.)
This should not be considered evidence of malware in the user's computing environment.
Just a quick heads-up that Technology's Performance team is planning to start a little on-page microsurvey later this month. It won't be running on enwiki, but somehow, all roads lead to VPT when people are surprised by what's on their screens... Feel free to ping me if someone brings questions to you about this. Whatamidoing (WMF) (talk) 21:44, 10 May 2018 (UTC)
I'm adding searchable text so people can find this. "Did this page load fast enough?" "For more information, see this survey's privacy statement." PrimeHunter (talk) 22:07, 10 May 2018 (UTC)
Control level of section auto collapse in mobile
In mobile views, only the first level of headings is auto-collapsed, when it's expanded, all subsections are already expanded.
Visit WP:ITNC on a mobile device (browser, not app, I don't use the app so I don't know) and you'll see how crazy huge it gets. I'm wondering if this can be controlled somehow on a per page basis. I'd be thrilled if all the date specific subsections under "Suggestions" were collapsed.
--LaserLegs (talk) 12:15, 11 May 2018 (UTC)
I can confirm it. mw:Page Previews (previously called HoverCards and not the same feature as Wikipedia:Tools/Navigation popups) omits displaying an image for lots of articles which have an apparently suited image, for example most clubs below the top-5 at 2016–17 Primeira Liga#League table. I don't know its algorithm. It doesn't have a ban on non-free images. It's possible it will only show images designated as the page image by mw:Extension:PageImages. Page information for S.L. Benfica shows the logo is the page image so this doesn't explain why it's omitted. Limited testing indicates that Page Previews may prefer larger images and svg images. PrimeHunter (talk) 23:32, 6 May 2018 (UTC)
@SLBedit, PrimeHunter, and Izno: Per English Wikipedia community request, PagePreviews won't ever show an image categorised as non-free. The images you highlight are all added to said category through {{Non-free logo}}; if you think the community erred in this request, I'd recommend an RfC, but I imagine fellow volunteers will disagree. :-) Jdforrester (WMF) (talk) 19:53, 7 May 2018 (UTC)
@Jdforrester (WMF), SLBedit, PrimeHunter, and Izno: the information about PagePreviews excluding non-free images was mistaken. The timeline is a bit complicated. Initially the function PageImage would return any image, free or non-free. That resulted in non-free images appearing in inappropriate places. So PageImages was modified to exclude non-free images. Later I suggested to the PagePreview manager that non-free was OK in previews - previews appropriately tie the images to significant related educational text. I opened the RFC Non-free_content/Hovercards, with consensus and an update to Non-free_content#Exemptions policy. Phab T131105 enhanced the PageImage function to default to free images only, and to include non-free images on request. Finally, per Phab T147317, PagePreviews were updated to include non-free images. Alsee (talk) 23:16, 11 May 2018 (UTC)
Over the last few days, I’ve noticed two things on my Special:EditWatchlist page. First, the checkboxes have changed from the little grayish 3D-looking buttons to blank white squares. At the same time, I’ve found that clicking on the links in the table of contents has no effect anymore; I end up having to manually scroll down the page.
I don’t care about the checkbox appearance, but I wish the links in the TOC would work. [running Mobile Safari in iOS 11.3.1 on iPad; viewed as desktop version] NotARabbit (talk) 02:55, 12 May 2018 (UTC)
Something's wrong with 'unsortable' column headers
Hi,
Since maybe 24 or 48 hours, the headers of the columns with the 'unsortable' attribute are sized just like a sortable column, without the sorting button.
A better example is Opinion polling for the next Dutch general election#Seats. The table has 19 columns, of which 16 are unsortable. Now look at the headers. On each of these 16 headers, the text has moved to the left (when it should be centered), and there is a large unfilled space on the right, probably meant for the button, despite it doesn't have to be there. As a result, the table is twice as high as before because the rows' height has doubled (except, I suppose, on very large screens).
It does not seem to be browser-related, but maybe it is Firefox 60's fault.
This is a deliberate change, to avoid the browser having to re measure everything when the sorting system is applied. I've created phab:T194451 to improve the case of unsortable columns. —TheDJ (talk • contribs) 07:32, 11 May 2018 (UTC)
The columns are also definitively not the correct width, either; now they're about 67px when they're supposed to be just 40px wide. Mélencron (talk) 14:13, 11 May 2018 (UTC)
Very true. So we'll have to manually reduce the width of every unsortable column until we find something correct? I don't know the extent of the attribute's use on Wikipedia as a whole, but there are certainly hundreds of polling pages affected. On the Dutch page, we must change from "width:40px;" to "width:22px;" or smaller, so that it fits (at least on my screen). In fact, removing the width property altogether is the best solution. But we can kiss goodbye to columns of equal width, when there are plenty like on the Dutch page. Kahlores (talk) 15:14, 11 May 2018 (UTC)
I think it's a bug/unintended behavior – as far as I can tell the new column widths seem to be dependent on the width of the text within the columns, which is... annoying, to say the least. I hope this can be resolved soon – a temporary "solution" would just be to drop the sortability of the tables entirely to restore the normal appearance. Mélencron (talk) 15:26, 11 May 2018 (UTC)
Sorry. It's a side effect of a deliberate change. It will be fixed for unsortable columns. I don't think anything changed for other columns. However.... I would say that if you need to tweak widths of columns, then maybe you are trying to squeeze too much data into a table. Wikipedia is not a spreadsheet. It's not a database. Good luck to the reader with a table like that on a mobile device... Good prose, good graphs etc are way more useful than big tables of election results. —TheDJ (talk • contribs) 15:38, 11 May 2018 (UTC)
It's never a good idea to force table column widths, particularly if you're doing it to make it look nice and neat on your own screen. You have no idea how wide anybody else's device is, particularly in this era of ultra-wide desktop monitors coexisting with android phones held vertically. The only sensible thing to do is omit all width specifiers, and let the browser work it out based upon the column content. --Redrose64 🌹 (talk) 07:11, 12 May 2018 (UTC)
Hidden text
Let's say I want to add something that makes editing articles easier to an infobox, but only display it to editors that choose to see this, rather than to everyone. How would I do that (with CSS or something).
In general editor notes are done with HTML comments, only visible in the source. Anything else would be a hack to "hide" from everyone (like with a display=none directive) except for people that knew to make their own directive to override it. I don't think this is a good idea for articles - what are your use cases? — xaosfluxTalk20:05, 11 May 2018 (UTC)
@Xaosflux: Basically, in {{infobox journal}}, I'd want a sort of "look for sources" series of links that cross checks the ISSN/publication title with several databases/National Libraries. This would allow us to fill other infobox parameters more easily and expand the articles. It's not something that should be displayed to readers, since it'd be an editor help, not reader help. The display=none thing sounds like exactly what I need, I just don't know how to set it up so I can override it in my .css/.js settings. Hidding the text is easy <span style="display:none;">...</span>. The problem is making it display. Headbomb {t · c · p · b}20:13, 11 May 2018 (UTC)
You can make your own, pointed to the same class and include a !important; designation to override with a new setting. Making a change to all infoboxes just for YOU is a horrible idea though! — xaosfluxTalk20:54, 11 May 2018 (UTC)
I tried that, but it seemed to just not work. And it wouldn't be just for me, but rather for a handful of editors that heavily deal with that infobox. It's the difference between being able to improve 100s articles per day because you've got everything you need at the end of a click, or 5 articles per day because you have to do a dozen queries by copy pasting from the infobox into a variety of databases. Headbomb {t · c · p · b}21:03, 11 May 2018 (UTC)
@Headbomb: I think project-wise a better solution would be to just add a unique class to the box, then use user javascript to add the special links. — xaosfluxTalk00:02, 12 May 2018 (UTC)
I agree that it is coding, but you might be able to puzzle it out. The help text at CS1 errors might help you figure it how the CS1 templates do it. – Jonesey95 (talk) 03:22, 12 May 2018 (UTC)
We already have cases of links that are normally hidden but which may be revealed with some user CSS. For instance, any WikiProject banner template that is built around {{WPBannerMeta}} has three "v-t-e" links which are normally hidden; these may be revealed with the CSS rule
.wpb.navbar{display:inline!important;}
Similarly, stub templates built around {{asbox}} also have three "v-t-e" links which are normally hidden, and these may be revealed with a very similar CSS rule
.stub.navbar{display:inline!important;}
These use the !important annotation (which is normally a cop-out since it's usually better to increase the specificity of the selector) because the display:none; declaration which we are overriding is part of the element's style="..." attribute. --Redrose64 🌹 (talk) 19:41, 12 May 2018 (UTC)
In general, we tend to avoid this technique in article space. As noted, it is used in some areas outside of article space and it is also in use on elements of transitory nature (notices). But making it part of the article content is general frowned upon. Especially, when it is used in the top of the article, where search engines etc tend to pick it up and possibly accidentally include it in their previews. Other transformed views could also be affected, think for instance of a machine learning system being fed this accidentally. This is also my problem with the current implementation of {{short description}}. xaoxflux suggestion of using a class and gadget seems like an interesting approach. —TheDJ (talk • contribs) 20:10, 12 May 2018 (UTC)
I don't see how FOUC is an issue in my two examples. Both templates use style="display:none;" so unless the browser ignores style attributes when first displaying the page (and I don't know of any that do), the first that the user sees of the v-t-e links is when the style sheets are processed. They then appear, and do not disappear again. --Redrose64 🌹 (talk) 20:20, 12 May 2018 (UTC)
can I check my contributions across all Wikimedia sites for hacked login?
There was a login possibly not by me (after which I changed my password). I checked contributions under my username at en-Wiki and there were none not by me, so that's good, but can I check for contributions across all Wikipedia projects and sites and all languages under my username within a time period of a few days between two legitimate logins? That's too many sites to check by hand, because I think there are hundreds of sites (including languages). Nick Levinson (talk) 02:44, 11 May 2018 (UTC)
It was an actual login, preceded by a few days by a failed one. The actual login may have been by me, but since the email did not say when it was (only that it was "recently") and I did not log in at the date-time of the email's time stamp, I decided to be cautious (to my annoyance) by changing my password.
The slow method is good, although it can't be used too soon after the period in question (someone has to wait, and maybe make a later edit to be sure of completeness) (whether the 20-edit limit matters I don't know). The generic URL for it is https://tools.wmflabs.org/guc/. The other (faster) page does not show edits within a time period, so it does not work for this purpose.
Hi, I returned after a very long hiatus. I have completely forgotten how I had edited the Quick links on the left Margin of this page. (The list of links below the Wikipedia logo on the left). I had placed a few customized links here and cant seem to remember how I had done it. or may be the newer Wikimedia software has made it harder to do so. In any case I thought may be someone here knows how to edit it. --DBigXray19:26, 12 May 2018 (UTC)
Can someone take a look in the BASIC-PLUS article? About half way through there is the text code|MAT A=ZER, which is being rendered as {{{1}}} on-screen. None of the other code templates is having a problem, why this one? Maury Markowitz (talk) 10:44, 13 May 2018 (UTC)
It is the equals sign, Maury Markowitz; {{code|MAT A=ZER}} breaks as mediawiki thinks there's an argument |MAT A= being passed with the value ZER instead of the whole "MAT A = ZER" as the first argument. I've fixed it as in hereGalobtter (pingó mió) 10:56, 13 May 2018 (UTC)
Adding insource:File insource:/File:.*File:/ to a search will find pages where the wiki source says "File:" at least twice. That misses cases with images added in other ways, and the search is slow and may time out if there are many results on the part before the insource code. PrimeHunter (talk) 14:59, 13 May 2018 (UTC)
Is it just my computer or is there a bug on Wikipedia. I've been noticing black squares appearing now & then on articles during both editing & viewing. GoodDay (talk) 00:54, 27 April 2018 (UTC)
GoodDay, next time you see one like that, please link the article here, along with where the black squares are appearing. By any chance, have you been reading any articles that deal with foreign people or places that use a non-European language? Mathglot (talk) 01:09, 27 April 2018 (UTC)
They appear randomly & disappear randomly. Sometimes flickering into different places on the same article, each time I press a key. GoodDay (talk) 01:41, 27 April 2018 (UTC)
GoodDay, please try to get a screenshot of that, if possible. In the meantime, what's your browser and OS? Maybe people with a similar setup could try watching out for it. Whatamidoing (WMF) (talk) 02:13, 27 April 2018 (UTC)
I don't understand those terminologies. If anyone else has noticed these things, then I'll know it's Wikipedia related, not just my computer. GoodDay (talk) 02:15, 27 April 2018 (UTC)
This is also happening to me on bn.wiki (see right). Browser google chrome, latest version, windows 8. This appear randomly. --আফতাব (talk) 20:33, 27 April 2018 (UTC)
WMDE’s Technical Wishes team conceptualized an idea how the problem could be solved: A generic solution that can be used for any refinement, such as pages, chapters, verses etc., and that could be used as a voluntary option, not forcing the users who don’t want to change their working mode to use it.
In order to find out if we can start working on this solution, we’re inviting editors from all wikis to have a look at it and tell us what they thinkin a feedback round from May 14th to May 27th. Thanks to everyone who participates and helps spreading the word!
Possible spambots targeting Wikimedia mailing lists
Hello,
A large number of @AOL.com email addresses just signed up to our usergroup mailing list (seemingly also successfully passing the "confirm email" requirement as well). Just curious if any other mailing list admin faced this?
Suspecting this to be some sort of spambot targeting Wikimedia mailing list(s), I have removed all of them, and changed mailing list rules to require admin approval. I've also posted a notice on our usergroup. About 6 addresses managed to successfully subscribe (before I removed them), while another 30+ was stopped thanks to the rules update.
All of them signed up with only a single name (first name), such as Nicolas, Alfie, Homer, Cody, Hort, etc.
For those running mailing lists with open memberships, please remember to exercise caution. As it could be some sort of a targeted spam/phishing attempt. Whoever is in a mailing list, often has access to participating members' emails, as well as discussion histories, real name and username link references, etc.
I've got 3 subscription requests from yahoo emails today, all with similar names. @JJMC89, on a separate note, would you by any chance know how to check why I cannot post anything to Wikimedia-l? I've tried emailing 3 separate times, but just got the moderator alerts only, and nothing thereafter. I could see others posting though. Cheers, Rehman04:25, 11 May 2018 (UTC)
The spam subscriptions are surging from both AOL and Yahoo. I've got 294 emails banned from our mailing list (and counting). If you want the list (to mass block in any of your mailing lists), let me know. I'll be happy to provide. Cheers, Rehman09:25, 14 May 2018 (UTC)
if login doubtful or fails, send email at that moment, not later
If a login fails or is spurious (e.g., Wikimedia is notifying us in case it was not by us) and Wikimedia sends an email to the user represented by that login, the user should know immediately and, if the user does not see the email immediately, at least the user should see a date-time stamp that is within a minute of when the login was attempted. There seems to be no other way to find out the date and time of the attempt, e.g., no log to check later. Please send the email right away, even if other kinds of emails go out once a day. Nick Levinson (talk) 02:49, 11 May 2018 (UTC)
What made me think that is that the email said the login event was "recently" and the email was timestamped for when I did not log in, but I had logged in within the preceding 24 hours, so, since I'm set to get emails daily, it could have been my own login and therefore nothing to worry about. But since the email was vague, I replaced the password, which means, since it's new, it's harder to remember. I don't want to have to replace it every time because of a vague recentness of a login.
I do have emails set to daily, which is more efficient for my work. However, a security email that is sent in the interest of both the user and Wikimedia probably should be allowed to override, so, if that's the problem, I can propose adding an override. Do you think there could be a problem with that?
I use the blackscreen gadget with monobook on Edge, Win 10. TOday I have noticed a problem when trying to view Notices (eg thanks, reverts of my edits). The problem is that the text displays in black on a black background. It is not possible to read black text on a black background. Help would be appreciated! I would upload a screenshot but it appears to be illegal and contrary to Wikipedia's terms and conditions to upload a screenshot of Wikipedia. DuncanHill (talk) 00:08, 12 May 2018 (UTC)
@DuncanHill: screens shots of wikipedia placed here for technical support are generally OK, mostly so long as you are not also including non-free media in the screenshot. — xaosfluxTalk01:36, 12 May 2018 (UTC)
See, now I just got a notice, but I can't tell from whom or for what. I think it might be for Xaosflux above. I did try uploading a picture, but the instructions made no sense whatsoever for what I needed to do. DuncanHill (talk) 02:23, 12 May 2018 (UTC)
Apart from actually getting the screenshot and saving it as a file on your computer/smartphone/device/etc., it's really no different from uploading a photo or other image to Commons. --Redrose64 🌹 (talk) 20:10, 12 May 2018 (UTC)
DuncanHill, feel free to drag-and-drop your screenshots into a comment on that Phab task, if the screenshots in that task don't show your problem clearly. You can also send me an e-mail message, and I can forward them along to the team, so they can determine whether the problem is in their code or in the gadget. Whatamidoing (WMF) (talk) 18:25, 14 May 2018 (UTC)
Simple way to extract DEFAULTSORT?
Is there an easy way to extract the DEFAULTSORT value from an article (using AWB or other methods)? As {{Sortname}} is deprecated, I was hoping to quickly get the DEFAULTSORT value from biographies to use for table sorting. S.A. Julio (talk) 09:31, 14 May 2018 (UTC)
@S.A. Julio: can you elaborate a bit. Are you looking if there is a declared defaultsort on the page (and/or possibly in a template transcluded), or just what the current sort value is? — xaosfluxTalk13:24, 14 May 2018 (UTC)
@Xaosflux: I was hoping to be able to quickly retrieve this information through a template or using something like AWB, is this possible? There will be over 700 players listed at 2018 FIFA World Cup squads, and the template used on the page, {{nat fs g player}}, has the parameter |sortname=. I was hoping to see if there was a quick method to add the sort values to the template? S.A. Julio (talk) 13:49, 14 May 2018 (UTC)
The DEFAULTSORT cannot be pulled by wiki code unless you read the whole page source with transclusion and try to parse it. I don't recommend that. Instead you need some script to collect the data and save it. I don't know a current script for it. "Page information" under "Tools" in the left pane shows the DEFAULTSORT value near the top. PrimeHunter (talk) 18:59, 14 May 2018 (UTC)
Latest tech news from the Wikimedia technical community. Please tell other users about these changes. Not all changes will affect you. Translations are available.
The new Advanced Search interface is now available as a Beta Feature on all wikis. This makes it easier to learn about and to use many of the powerful options in our search. Feedback is appreciated. [33]
Changes later this week
The new version of MediaWiki will be on test wikis and MediaWiki.org from 15 May. It will be on non-Wikipedia wikis and some Wikipedias from 16 May. It will be on all wikis from 17 May (calendar).
Meetings
You can join the technical advice meeting on IRC. During the meeting, volunteer developers can ask for advice. The meeting will be on 16 May at 15:00 (UTC). See how to join.
Future changes
In the mobile view, warnings for when something is wrong with a page are not as clear as they should be. The developers are working on this. You can give feedback and suggestions.
The developers are working on making the Wikipedia Android app available in more languages. You can give feedback, suggestions and help test it. Read more on mediawiki.org[34]
An attempt to use the search box on WP:MOS gives the following error:
A warning has occured while searching: The namespace found in the search term prefix: is not compatible with the namespaces requested. You can fix it by prefixing your query with all:
It's Thursday again. New MediaWiki version. Per phab:T193392, an inputbox now requires a namespaces parameter to match a namespace in the prefix parameter. This case was fixed by [35] but there must be many other inputboxes needing similar fixes. PrimeHunter (talk) 22:22, 10 May 2018 (UTC)
It's up to the person implementing the change to do it, not us. And although it's hard to follow the poor English in phab:T193392, it says it should emit a deprecation warning, not error the search off. I suggest pulling the change. Hawkeye7(discuss)22:35, 10 May 2018 (UTC)
A search of insource:inputbox insource:prefix in all namespaces gives 1964 results. It looks like most of them need a fix if the feature is not changed. I guess a bot could be coded to fix most cases but a warning in advance would have been nice. Maybe this should be pulled for now. PrimeHunter (talk) 22:43, 10 May 2018 (UTC)
The warning is made by MediaWiki:Cirrussearch-keyword-prefix-ns-mismatch which currently says: "The namespace found in the search term prefix: is not compatible with the namespaces requested. You can fix it by prefixing your query with all:." See e.g. italics prefix:Wikipedia:. The warning will probably make little sense to most users who just entered a term in an inputbox and didn't write prefix:. I suggest we change the message to say they can click "Everything" without getting technical. Maybe: 'Click "Everything" above to perform this search. A software issue currently requires this step.' PrimeHunter (talk) 23:06, 10 May 2018 (UTC)
Looks like the "fix" was not backward compatible. This is not called a "fix", this is called, "breaking working functionality". The burden is on those making a change to ensure current functionality continues to work transparently, or in the rare case, to provide an automatic migration as rapidly as possible, with notification to affected users of a window of functionality downtime well in advance of the migration. D- . Mathglot (talk) 01:40, 11 May 2018 (UTC)
...but there must be many other inputboxes needing similar fixes. — Indeed; like every page using {{Talk header}} to search Talk page archives. About 390,000+ just for that one template, according to Template:Talk header, and there are other headers that use it. @PrimeHunter:. Mathglot (talk) 01:52, 11 May 2018 (UTC) Update: Oh, I see what you mean, you're counting inputboxes, and I was counting affected pages, but that's how I see it as a user, not a fixer, and that's how many will see it. Happy if there's < 2000 pages to fix, though. Mathglot (talk) 02:11, 11 May 2018 (UTC)
26000 uses of {{Talk header}} have an optional search parameter which generates a search box with {{Search box}}. I have fixed it there [36]. There are still other talk page templates which generate search boxes in other ways, e.g. {{Round in circles}}. Is there an expected timeframe for deploying the patch to revert this change? A template search only finds around 60 affected templates. We could quickly fix those. PrimeHunter (talk) 10:48, 11 May 2018 (UTC)
Nvm that nvm, doesn't work. I could be missing a parameter but it doesn't even seem like inputbox allows fixing this; which just shows terrible foresight in making this change. I see JJMC89 has tried something but It doesn't seem to fix it - the namespaces option seems to show buttons to specify namespace, not for specifying namespace in search automatically..Galobtter (pingó mió) 05:40, 11 May 2018 (UTC)
Please excuse me for the annoyance this has caused, I prepared a patch to revert this change. I'll make sure that the prefix feature offered by InputBox is not affected again. DCausse (WMF) (talk) 09:49, 11 May 2018 (UTC)
The fix has been deployed and the prefix keyword should work as before. The intent was not to deprecate the prefix feature as a whole but some unique behaviors it supported (allowing to override the namespaces selected). The main reason was to address some of the technical debt by trying to reduce the moving parts. Please see my comment here. The approach I've taken was clearly a mistake and I will find other solutions that do not break the search boxes that use prefix. DCausse (WMF) (talk) 14:43, 14 May 2018 (UTC)
A warning has occured while searching: The namespace found in the search term prefix: is not compatible with the namespaces requested. You can fix it by prefixing your query with all:.
Wondering if someone could provide a line of jquery that would move "[rollback]" (class "mw-rollback-link") to before the user name, on each line (ideally in whatever page context it occurs in). I'm not sure about that location yet, but I think if given the form of the jquery I could adjust it myself, as I understand basic DOM stuff. Thank you! Outriggr (talk) 01:35, 14 May 2018 (UTC)
Outriggr, there's a project going on to make a fundamental change to rollback: users who opt in will be given a confirmation screen. I don't know how much change that will entail, but it's a big enough deal that any changes related to rollback links may end up being useless rather soon. Nyttend (talk) 02:21, 14 May 2018 (UTC)
I wouldn't go that far. There was a proposal for this, and feedback was requested. The feedback (I participated) has been largely negative, so I can't imagine this being... ahem, rolled out any time soon. There are certainly no solid plans for anything, so there's no need to make decisions with that in mind. ~ Amory(u • t • c)14:34, 14 May 2018 (UTC)
@Outriggr: For a quick and dirty option (i.e., there's probably a more efficient way):
should do it on diffs, while using history-user instead of mw-userlink should do it for history pages. You may need to futz to figure out brackets, etc. ~ Amory(u • t • c)14:30, 14 May 2018 (UTC)
Thanks very much Amorymeltzer. This works on my watchlist, except that because it grabs the link (which is what I said, I guess :-) and not the whole span, there are spacing problems and the square brackets remain at the end of the line. As usual, my own adjustment, to get the rollback span, does not work:
What gadgets, etc. do you have installed? Those tweaks work fine on my watchlist. The group-them-all-together feature of watchlists does weird things to links, as does the mark-as-read feature, maybe that's affecting? ~ Amory(u • t • c)01:14, 15 May 2018 (UTC)
Meanwhile I have it working (on the watchlist) by putting back another level of "parentElement". I think there may be too many factors that affect the layout of these pages... But you'd think I could get it to work for my own. (What I have now, with two ParentElements, is working everywhere but user contribs, where it groups ALL the rollback links in the same place (I've now chosen mw-changeslist-separator), so it's still not fully working. If I can figure out how to fix or turn off the JS for contribs, I will use it, and if not I'll forget it. Thanks for your help! Outriggr (talk) 01:29, 15 May 2018 (UTC)
Could somebody pls have a look. We are in a project everybody can edit, but I am afraid this goes beyond my technical abilities. The error notice is at the bottom of that page.--Ymblanter (talk) 06:53, 15 May 2018 (UTC)
We'd need a free/libre 3D rendering to use, but I love the idea. I'm sure we could use HTML5 to do something like this. Richard061218:17, 14 May 2018 (UTC)
@Richard0612 and Xaosflux: & others - BRIEF Followup - seems the images in the NYT example may have been from (or, at least, based upon) NASA images, usually in the Public Domain (PD) AFAIF - more specifically, image credits seem to be given as => NASA; Jet Propulsion Laboratory; Southwest Research Institute. Maps by NASA’s Cassini mission (Jupiter), the U.S. Geological Survey (Io) and Björn Jónsson based on data from the Voyager and Galileo missions (Europa, Ganymede, Callisto) - which, being US govt, may all be in the PD I would think - these (and related images) may be available for Wikipedia to use - if one knows exactly where to find the best images to use - and if one knows exactly what coding container/? to use - iac - hope this helps in some way - Enjoy! :) Drbogdan (talk) 20:20, 14 May 2018 (UTC)
No special code is needed to display the 3D images that Xaosflux linked - they're just stereolithography files that MediaWiki handles on its own. We might want to develop some form of embedded viewer if we find enough images to use them in many articles. Richard061220:25, 14 May 2018 (UTC)
Try changing var showUserGroupSettings to window.showUserGroupSettings in your global JS. Changing various scripts to not be run in the global context has been happening for a while, perhaps such a change recently happened for GlobalCssJs. Anomie⚔11:50, 15 May 2018 (UTC)
What happens if you did not fill in the title on Chrome 66.
For some apparent reason, the Special:AbuseLog, or edit filter log, now requires you to fill out the title of the page before searching. Otherwise it won't work. It happened to me while using Google Chrome on Windows 10. I did manage to bypass this by using Chrome's "Inspect Element" function and removing the "required" tag, so it couldn't have been just because of what browser/OS I'm using to edit Wikipedia. Posting an image soon. Done. theinstantmatrix (talk) 14:47, 15 May 2018 (UTC)
There were some blockers on early deployment that apparently may not get resolved. The fix will come Thursday with the normal weekly deployment. — MusikAnimaltalk22:46, 15 May 2018 (UTC)
I am using the gallery template rather than the gallery tag as I want italics in the gallery heading (& the tag doesn't seem to be able to do that). The drawback is that the template doesn't seem to offer a way to set the vertical alignment of the images - they all end up jammed up the top of each image frame :(
Is there a way to do this?
Sample gallery/code is as follows ...
The radical right has broken into the WikiMedia software and case access and alter your watchlist. Walk away now, don’t look back. —SmokeyJoe (talk) 16:42, 15 May 2018 (UTC)
It was created a few hours ago while I was out either driving or shopping. There's no reason I would even know about it. I've got 2FA, no one has hacked me. Never mind, software does have strange glitches from time to time. Doug Wellertalk18:01, 15 May 2018 (UTC)
Category lists articles that shouldn't be there; articles have no links to category
I found a category that has a bunch of articles that
are badly sorted, and
shouldn't be in the category in the first place.
They relate to the topic, but at best, they'd belong in parent categories. But when I go to the articles to edit them, I don't find any category links to that category.
I do understand that if an article uses a template that has a category link, the article can appear in that category, whether it should or not. These articles do have some templates in common. But I'm getting a little lost trying to figure out if there's a problem category link in one of those templates.
Suggestions? I'm an extreme fixer of categories here, but this problem has me stumped. In case it's of any help, the category in question is flatfish. Uporządnicki (talk) 17:40, 16 May 2018 (UTC)
Oops! Sorry! (What a clumsy mistake!) The category is Scophthalmus a genus of flatfish. Yes, one of the articles listed in the category is Flatfish. If anything, the Scophthalmus category should be in a category for flatfish, not the other way. And the problematic entries in that category are all listed under M. P.S. The Scophthalmus category is actually a sub-subcategory of Pleuronectiformes, which is flatfish. Uporządnicki (talk) 18:31, 16 May 2018 (UTC)
Yes, it was a missing colon (the error has been there for six years); and the intended link being piped to display "More..." explains why pages were all listed under M. --Redrose64 🌹 (talk) 19:17, 16 May 2018 (UTC)
User:Sj/Update Congress, a user sandbox page, is appearing in several mainspace categories. User sandbox pages aren't allowed to do this, so the categories have to be removed, but the problem is that there are no category declarations directly on the page at all — the categories are there because the page is transcluding a mainspace article rather than copying its text, such that my first attempt to remove the categories actually had the unintended effect of stripping them from the article rather than the sandbox page. So I'm stuck: is there any other way to get it out of the mainspace categories? Thanks. Bearcat (talk) 23:03, 16 May 2018 (UTC)
It looks like this editor wants to edit sections of the mainspace page and have easy access to the reference links they added to the top. I don't think there's any policy against transcluding articles into userspace pages. The closest I can think of is WP:FAKEARTICLE, but that doesn't really apply here. --Ahecht (TALK PAGE) 00:07, 17 May 2018 (UTC)
I can't think of a good reason why it would be the case, but if a mainspace page is frequently transcluded for some reason, you can also wrap the category calls in <noinclude></noinclude> tags. --Ahecht (TALK PAGE) 00:07, 17 May 2018 (UTC)
I'm just wondering, would a template like {{Draft categories}} work in a case like this? If the categories on the mainspace article are placed in the template, they should appear as normal in the article but as links in the transclusion, right? Or is the order of template expansion not suitable for this? rchard2scout (talk) 07:49, 17 May 2018 (UTC)
@Hawkeye7: Sounds like the value you pass is User:Hawkeye7, which is not a username, but a page. You need to pass it the value Hawkeye7. Not sure what tool you are using exactly, so if you could link to that, that would be very helpful. —TheDJ (talk • contribs) 08:07, 18 May 2018 (UTC)
It's the Perl MediaWiki::Bot module. The API has been working until now. I've had to shut down the MilHistBot monthly membership update run. I didn't have time to investigate further this morning. I had considered the same thing as you. The call in question simply says:
So the module is prepending the 'User:'. It hasn't changed, so a bug has been introduced into the Mediawiki software itself. There should, therefore, be an entry in phabricator. Hawkeye7(discuss)08:19, 18 May 2018 (UTC)
Our instincts were correct. That is indeed the error, and it means that someone put a major change in without testing it. The Mediawiki code needs to be reverted asap. Where's DCausse (WMF)? Hawkeye7(discuss)08:35, 18 May 2018 (UTC)
Ticket filed. This was undocumented behavior of the API, that apparently the volunteer maintained perl library depends upon. BTW. —TheDJ (talk • contribs) 09:34, 18 May 2018 (UTC)
Hi, I'm trying to modify the wikitable, here, but nothing worked so far, even with the !important declaration. I tried bypassing the cache, but no results after that either. So what I'm doing wrong? Thanks! --SE(talk)18:13, 18 May 2018 (UTC)
@Dsm.wiki: "Warning: There is no skin "Common". Custom .css, .json, and .js pages use a lowercase title, e.g. User:Foo/vector.css as opposed to User:Foo/Vector.css." ;) I moved the page for you. —TheDJ (talk • contribs) 18:23, 18 May 2018 (UTC)
Latest tech news from the Wikimedia technical community. Please tell other users about these changes. Not all changes will affect you. Translations are available.
Changes later this week
The new version of MediaWiki will be on test wikis and MediaWiki.org from 22 May. It will be on non-Wikipedia wikis and some Wikipedias from 23 May. It will be on all wikis from 24 May (calendar).
Meetings
You can join the technical advice meeting on IRC. During the meeting, volunteer developers can ask for advice. The meeting will be on 23 May at 15:00 (UTC). See how to join.
Future changes
It could become easier to reference different pages of a book in an article. You can give feedback. The last day for feedback is 27 May.
Let us say an image is 1000x500 pixels in dimensions. Is there a way to only show a 500x500 pixel portion of the image without uploading a new image? Thanks. SharkD Talk 08:37, 21 May 2018 (UTC)
Not without a lot of CSS trickery which could make the wiki markup harder for other users to read and edit. What exactly are you trying to achieve? Richard061208:50, 21 May 2018 (UTC)
Thanks for the template! Actually it is for other wikis like Wikia and Gamepedia. I will ask, maybe they can implement the template. SharkD Talk 23:10, 21 May 2018 (UTC)
Always say if a question is not about the English Wikipedia. Those wikis aren't even Wikimedia wikis. Special:ExpandTemplates can show the code produced by a template but a wiki may dislike if complicated code like that is saved for a specific image. PrimeHunter (talk) 00:09, 22 May 2018 (UTC)
Russian, Japanese, and French Wikipedias will be read-only for a few minutes next Tuesday
This project will not affect the English Wikipedia, but I'm posting this just in case you get questions here:
Remember the server switches in 2016 and 2017, when nobody could edit for half an hour while Ops did some complex maintenance work? Think of this project as a one-time, miniature version for just three Wikipedias. I understand that they're putting in new hardware for ruwiki, jawiki, and frwiki. To avoid database problems, they're going to first turn off editing, turn off the old hardware, turn on the new hardware, and then turn editing back on.
If all goes well, the read-only database lock could last as little as five minutes (maybe as long as 15 minutes). The estimated time of the outage is Tuesday, May 29th, somewhere between 10:00 and 12:00 UTC (late morning in Europe; very early morning in the Americas).
The usual processes apply: Probably nobody will post anything here, but if you happen to get questions, then you can tell people that it's expected and should be over in less than 15 minutes. If it goes on longer than expected or if something weird happens, then please report urgent problems via IRC in the #wikimedia-techconnect channel. You can also ping User:Trizek (WMF) on wiki, as he is much more likely to be awake at that hour than I am. Whatamidoing (WMF) (talk) 18:52, 22 May 2018 (UTC)
I've used it several time today and it has worked normally. One advantage that reflinks has over reFill is that it can format PDF's. OTOH, neither of them is foolproof and there are some refs that just have to be formatted by hand. MarnetteD|Talk23:24, 23 May 2018 (UTC)
Is there a way of having a collapsable list similar to how the CategoryTree extension displays
Hi
I want to create a simple collapsable list of regions with countries in them for a competition I'm setting up. Is there a way I can produce something I can edit manually and specify the links that looks a little or a lot to how CategoryTree works?
You specifically say "collapsible list" so maybe you already know {{Collapsible list}}. It has similar functionality but does not look similar. Example with nested lists:
Specifying an image width as a percentage of the screen rather than a number of px
Hi all
I'm putting together a competition and would like to add a banner to the top of the page, can someone tell me if there is a way of specifying that an image stretch the whole width of the page (perhaps with a maximum number of pixels for very high res screens). I want to avoid the page looking good on desktop and then breaking on mobile phones.
If you're doing this on meta wiki you can specify a background image for a div(using TemplateStyles) and then specify the height and width...I don't know of a way to do it on enwiki before June 2018(When TemplateStyles goes live here) thought — FR+14:43, 24 May 2018 (UTC)
Thanks very much @FR30799386:, do you know if this available on Commons? I'm looking at the Phabricator board for template styles and I can't see either en.wiki or Commons on there.... Do you know where I can find an example I can copy from to work on Meta? Or where any instructions are? I feel like this is a bit:
Sorry! This site is experiencing technical difficulties.
Try waiting a few minutes and reloading.
(Cannot access the database: Cannot access the database: No working replica DB server: Unknown error (10.64.16.84:3318))
When I click on prev, diffs and history,
Wasn't sure if those behind the scenes knew so thought I'd give a heads up, Thanks, –Davey2010Talk19:38, 24 May 2018 (UTC)
I can't login, or rather I can but then get the error message above. Somehow I got logged out, maybe fat thumbs on my iPad. Is 2FA affected by this? — Preceding unsigned comment added by 86.11.114.177 (talk) 19:55, 24 May 2018 (UTC)
My question is whether the information which I had typed and was saving when the message popped up is irretrievably lost? SusunW (talk) 20:13, 24 May 2018 (UTC)
I've been copying edits before I try to save them, just in case I have to repaste and try again. Despite that, sometimes I've had to do the repaste thing, but I have also had some edits where I got the error message but they somehow went through anyway. Bearcat (talk) 20:16, 24 May 2018 (UTC)
@SusunW: (e/c) Unless you went back in your history when you received the error (in a modern browser that keeps state) and then copy-pasted what you typed, then likely yes it is gone. —TheDJ (talk • contribs) 20:18, 24 May 2018 (UTC)
@Bearcat: thanks for that. Animalparty I was able to hit my back key three times and eventually get to the information that I was trying to add. 2,547 bites would've been a lot to recreate. I will start making copies for a bit to prevent problems. SusunW (talk) 20:21, 24 May 2018 (UTC)
Finally. I tried to start a thread here and couldn't. I tried to reply when I discovered this one and couldn't. Tried again a few minutes later and couldn't.— Vchimpanzee • talk • contributions • 20:25, 24 May 2018 (UTC)
Just FYI, this hotfix has resolved the issue temporarily while the devs figure out a permanent fix for what appears to be Wikidata object disambiguation DoS-ing the database servers. Richard061221:06, 24 May 2018 (UTC)
And as an FYI to that FYI, note that this hotfix has disabled some Wikidata functionality:
Searching among drafts among a particular category
Is there a way in which I can search for India related drafts?(Best if anyone can do it without fancy tools/SQL/regex or whatever.i.e something that I can remember the next time I want to do the same thing) — FR+13:58, 24 May 2018 (UTC)
Drafts should not have categories or WikiProject tags so it may be hard to do better than searching the draft namespace for the string India or Indian: [42]. If you want pending submissions then add incategory:"Pending AfC submissions": [43]PrimeHunter (talk) 14:36, 24 May 2018 (UTC)
I've just noticed that whenever I hover my mouse over a link to get the popup on a Wikipedia page, Adblock Plus's counter on the status bar shows it blocked something. The filter lists I'm using are the latest versions of Adblock Warning Removal List, EasyPrivacy, Fanboy's Social Blocking List and EasyList. The counter gets incremented but there's seemingly no difference in the content display whether Adblock Plus is disabled or enabled. I don't know how to find out which element is Adblock Plus blocking. What is going on? 93.136.122.81 (talk) 19:37, 24 May 2018 (UTC)
You can check what adblock is blocking in your own browser. For example in Firefox, click the Tools menu, Web Developer, Toggle Tools and then Adblock plus. Then reload the page. Entries with an red font are the entries that Adblock is blocking on that page.--Snaevar (talk) 10:35, 25 May 2018 (UTC)
I have previously logged into phab using my mediawiki account. Now, when I open https://phabricator.wikimedia.org/ and click the login button, I'm offered buttons to 'Log in or register' with Mediawiki or 'Developer log in' with a Wikitech Account (LDAP). Clicking the first of these takes me to the mediawiki.org OAuth page, where I click the 'Allow' button. This redirects me to Auth > Mediawiki > Register (https://phabricator.wikimedia.org/auth/register/<key redacted>/) where it asks me to fill in a username and email. If I try to do so, it says that a user with this username and email is already registered (which is not unexpected). There appears to be no way to log in.
WP:VPT would be better for reporting this. I don't know, but a first check would be to ensure that nothing is blocking cookies or scripts. A solution might involve partially or fully clearing your browser history. Johnuniq (talk) 07:13, 22 May 2018 (UTC)
Alternatively, try logging in using incognito/private browsing mode - that'll allow you to test with a fresh slate without having to clear your entire history. Richard061209:26, 22 May 2018 (UTC)
...but when I got there, I noticed that someone else has reported the same thing. I have an LDAP account after all, presumably because I'm registered on ToolForge or whatever it's called now - "developer account" sounds like it's for MediaWiki developers. GoldenRing (talk) 12:10, 25 May 2018 (UTC)
I seem to have stopped receiving email notifications
The last email notifications I have from a Wikimedia project were in March. I just did a search in my inbox (which seems otherwise fine), and just checked my preferences (still checked for talk page messages and a few other things). Can't think of anything that I would've changed. Has this happened to anyone else? — Rhododendritestalk \\ 12:44, 25 May 2018 (UTC)
Well, I didn't receive it but Facepalm somehow I thought my email searches included the spam folder. Indeed it's in there. Must've misclicked at some point, thanks. :) — Rhododendritestalk \\ 13:59, 25 May 2018 (UTC)
Why are these inline CSS and not in the MediaWiki/Common.css for the vevent declarations? Just trying to learn some CSS for my own stylesheets. --Chelston-temp-1 (talk) 23:07, 24 May 2018 (UTC)
@Chelston-temp-1: a very good question. The reason is because "MediaWiki:Common.css" is loaded on EVERY page, for EVERY user. We have a lot of CSS that is specific to a small subset of pages and moving all of that into a central page would be highly inefficient. So instead, we only have the most common/widely spread css into this central file and statements that are much rarer remain as style attributes.
However, very soon, we will have TemplateStyles, specifically to deal with this problem. It will then be possible for templates to include style sheets specific to that template, that will then be included only on the pages that are using those style statements. —TheDJ (talk • contribs) 09:57, 25 May 2018 (UTC)
@Chelston-temp-1: In HTML, classes have several purposes, and construction of selectors in style sheets is just one of them. Another use is in web scraping: external software might use a class in order to pick out certain data from a web page, and I am pretty certain that this is the purpose of the vevent class.
As regards where we also use that class, have a look at e.g. Wikipedia:UK Wikipedians' notice board, the section headed "Meetups". The list in that section is a table, and every entry for a single meetup has the vevent class. If the styling text-align:center;background:inherit were to be added to the style sheet for that class, this table would have centred text and a pink background, which we probably do not wish. --Redrose64 🌹 (talk) 11:50, 25 May 2018 (UTC)
@Redrose64: I was asking as I was using the page as something to learn from for styling HTML tables (Wikipedia's table design and Arial font is simple, but it's a good one for basics). I used view source in Firefox 56 but could not find the class anywhere in the CSS files linked to, and not declared as an inline stylesheet. This is about technical stuff, so I guess it's kosher to ask about webdesign here. --Chelston-temp-1 (talk) 12:05, 25 May 2018 (UTC)
Taken to its most basic level, a class is merely a way of giving a shared name to one or more elements in order to indicate that they have something in common. There is no requirement that a class should be mentioned in CSS files. We might write a web page thus:
<!DOCTYPE html><html><head><title>Example</title><style>body{color:black;background:white;}.apple{color:red;background:green;}</style></head><body><h1>Example</h1><p>Apples and <spanclass="banana">bananas</span> are kinds of fruit.</p></body></html>
and this is perfectly valid HTML even though (i) one of the two rules in the <style>...</style> element has a selector that does not match anything in the HTML document; and (ii) one of the HTML elements has a class attribute specifying a class that is not mentioned anywhere else. --Redrose64 🌹 (talk) 22:01, 25 May 2018 (UTC)
@J. E. C. E.: Interlanguage links and categories on pages used for transclusion should always be inside <noinclude>...</noinclude> to prevent them from applying to pages with a transclusion.[44]PrimeHunter (talk) 21:13, 25 May 2018 (UTC)
N.B. I'm writing in haste and a bit distracted, so apologies in advance if I've simply overlooked something.
I recently spotted this edit, which seems to be appropriate to revert based upon e.g. the comment at the start of the relevant section. (Please don't derail this with a discussion of the merits of reverting/retaining that edit, though; this thread is not about that.)
I therefore clicked the "undo" link, but instead of opening a revert edit, it took me to this URL, which says in red writing at the top of the page, "The edit appears to have already been undone. You may have attempted to undo a page move, protection action or import action; these cannot be undone this way. Any autoconfirmed user can move the page back to its previous location, and any administrator can modify or remove protection."
However, AFAICT, none of those possibilities apply. WikiBlame shows the insertion of the offending text here but does not show it to have been removed (though note that the latter search seems to result in some kind of infinite loop, which might be a clue to what's causing the problem I'm reporting?). Nor does a glance at the article show it to have been removed.
Nor was I attempting to undo a page move, protection action or import action.
So, why the page with the red writing instead of a page lining up a revert commit?
@Zazpot: The text that was inserted in that edit was removed in the following edit (and re-inserted in a different place in the article). That removal is why the "undo" feature is saying that the first edit has already been undone. Anomie⚔16:04, 26 May 2018 (UTC)
Thanks :) Just shows how distracted I was. The WikiBlame recursion still seems odd, but I guess that's a separate issue. Zazpot (talk) 17:57, 26 May 2018 (UTC)
Help compiling a list of editors by experience level
I am a little out of my depth on this one, so would appreciate a bit of help.
The background for this is to help identify a pool of editors from which might be found candidates to invite to join New Page Patrol and/or Articles for Creation.
I would like to compile a list of editors (listed by edit count) who:
Are not members of the new page reviewer or administrator user groups,
Have >3000 main space edits,
Have >500 edits in the last 365 days,
Have been registered longer than 365 days,
Have a clean block log for at least 365 days,
Have performed at least 5 page moves,
Have created at least 2 non-stub articles in main space,
Have had no created articles in mainspace deleted in the last 365 days.
While I would prefer to be able to run it as an SQL query through Quarry, Xaosflux mentioned that it might have to come from a dump, because with the many criteria its run time might be too long for Quarry. Xaosflux mentioned that NKohli (WMF) might be able to help there.
First though I need to have a query to run, and I am sad to report that my coding experience is limited to copy pasting and stitching bits of other people's queries together into bizarre Frankenstein's monster queries. When it comes to making something like this from scratch I need some help (I need someone to write it for me). If anyone can help, I'd appreciate it. Cheers, — Insertcleverphrasehere(or here)03:05, 25 May 2018 (UTC)
I am well aware of all the lists. I previously used the top Wikipedian's list for my last round of invites, however, manually sorting editors on basic criteria such as activity level was very time consuming. I am hoping to narrow the field beforehand this time (the process will still involve manual checks on some stuff, but if I can run a query for editors meeting ALL of the above, then most of the work will be done before I start. Edit counter tools are great for analysing individual users, and the edit count lists are great for specific things, but I am looking for a complex combination of criteria that can only be put together via a custom query. — Insertcleverphrasehere(or here)03:31, 25 May 2018 (UTC)
If there's a remotely efficient way to query any of that besides user groups and registry time without a specific user to check, its beyond my ability. (I've got a query that's the polar opposite of "remotely efficient" running for everything up to and including the block log check running; I'll check in on it in the morning, and be very unsurprised if it's still running.) You might get more luck, or at least a different audience, at Wikipedia:Request a query. —Cryptic04:37, 25 May 2018 (UTC)
I wonder if it would work better to run a couple of smaller queries, and then combine the lists (people with the correct number of edits, minus people with the userright, etc.). Whatamidoing (WMF) (talk) 17:45, 25 May 2018 (UTC)
To some extent, but some of the individual conditions take too long to be practical by themselves. Counting either main space edits or edits in the past year takes about four seconds each per user, and there's almost 9500 of them to check after querying everything that can be found relatively quickly (quarry:query/27196). —Cryptic19:55, 25 May 2018 (UTC)
Thanks for your help Cryptic. It seems like your level 2 query finished, which is pretty awesome in and of itself. Doesn't look like the Level three one is going to finish unfortunately. Any idea what wording for the last two criteria would look like (at least 2 creations, and no deletions of created articles)? While we might not be able to run it in quarry, we may be able to run it on a database dump, as suggested by Xaosflux. If it takes a while to run that won't be a big deal. — Insertcleverphrasehere(or here)21:05, 25 May 2018 (UTC)
I restarted the third query fairly recently after I noticed a missed optimization in the first that cut its runtime from about three minutes to thirty seconds, so I haven't given up on it yet.Non-stub article creations in mainspace would look something like (completely untested)
AND(SELECTCOUNT(*)FROMrevision_userindexJOINpageonpage_id=rev_pageWHERErev_user=user_idANDrev_parent_id=0ANDpage_namespace=0ANDpage_len>=2000-- length in bytesLIMIT2)>=2;
That's going to incorrectly count, for example, redirects that the user created that were turned into articles by someone else, and incorrectly miss the reverse case. Nothing we can do about that, since neither quarry nor the labs replicas have article text. Looking for categories with "stub" in their name (à la the latter part of quarry:query/18928) might be more accurate than page_len, but it's more effort to write and I'd guess slower.I haven't thought about the query against deleted revisions yet. I hadn't even been aware it was accessible. ... it probably shouldn't be. —Cryptic21:22, 25 May 2018 (UTC)
@Cryptic: Hmm... you can view what articles they created that have been deleted in Xtools, in the page creation log, So i guess I thought it would be query-able, but perhaps its not worth it anyway. I suspect a lot of false positives might be returned (i.e. redirects deleted), which will exclude people that we don't want excluded. Shelve that criteria for now, as it might be more trouble than it is worth. Something that might help with optimisation though; I already manually combed through the top 1000 editors (everyone with >64,000 edits) so we can exclude them from the search if it makes things faster. — Insertcleverphrasehere(or here)22:53, 25 May 2018 (UTC)
@Cryptic:Also, if mainspace edits are too resource intensive to search for, I suggest raising the global edit count from 3000 to 5000, and this will also help a lot with optimisation (the curve gets very long the further you go out when it comes to edit count, as it is a power law curve). I decided to try running this query HERE. — Insertcleverphrasehere(or here)22:58, 25 May 2018 (UTC)
Thanks very much Cryptic, I ended up finding a workaround, posting the queries in tiers i.e. 64000-60000 edit count, then 60000-50000 etc. in 9 total queries that would each run for just under 30 min. Then I compiled the results which can be found here: User:Insertcleverphrasehere/NPR invite list. All in all this set of criteria works for what I need it for, so thanks for helping me out. Cheers, — Insertcleverphrasehere(or here)22:21, 26 May 2018 (UTC)
Help please! I am going to be unable to sign this posting as when I press the normal four tildes sign on my computer, I get four vertical lines instead of what I always used to get - it like so ||||. Does any one know what has gone wrong on my computer and how this can be rectified? If this is of any assistance, it might be worth my saying that the @ and the " keys on my computer seemed to have reversed functions. Many thanks for in advance for any co-operation. |||| — Preceding unsigned comment added by Vorbee (talk • contribs) 18:54, 24 May 2018 (UTC)
It sounds like you changed your keyboard layout in your operating system, check for controls there. Also, in the edit tools above this editing box there should be a sign button above and/or below the box. — xaosfluxTalk18:59, 24 May 2018 (UTC)
Vorbee, it sounds like you might have enabled the IPA keyboard. Do you get other oddities, when you type other (non-basic) characters? If so, then look around in the corner of the editing windows for a little keyboard-style pop-up button. Click that, and switch back to something that isn't IPA. Whatamidoing (WMF) (talk) 20:53, 24 May 2018 (UTC)
I tried switching to "International Phonetic Alphabet" (IPA), and it didn't give me a vertical bar in place of a tilde. --Pipetricker (talk) 23:03, 24 May 2018 (UTC)
Vorbee, above the wikitext editing area, do you have a toolbar looking like any of these?:
If you have this toolbar, clicking on the button marked by the arrow will give you four tildes.
If you have this toolbar, clicking on the button marked by the red square will give you four tildes.
Or, if below your wikitext editing area you have something looking like the below, then you should be able to find a button with four tildes in the upper row:
––
Questions to find out where the problem is: Do you use Windows or Mac, and which version? Which web browser (and version) do you use?
If you write text on your computer, but not in Wikipedia, such as A) on a Google search page, or B) not in your web browser, but for example in Microsoft Word or similar, do you have this problem if you try typing a tilde?
Many thanks for your help. I have now found a way to get four tildes - if press on "Edit this page" and then go to symbols, I find the tilde symbol and can then click on that. Thank you a lot for every one's help, it is appreciated. Vorbee (talk) 08:15, 25 May 2018 (UTC)
@Vorbee: I have a desktop PC with a UK keyboard, and my Windows is set up for UK keyboard layout. Occasionally, the browser forgets this and defaults to the US keyboard layout; as you note, the tilde key generates a pipe (|) instead. Other keys move around too: the at-sign (@) and double quote (") exchange places, and the tilde moves to the key to the left of that for the figure one (1 !) - this is the one marked with three symbols (` ¦ ¬), and when shifted, produces the tilde instead of the ¬ character. --Redrose64 🌹 (talk) 09:50, 25 May 2018 (UTC)
This fits with Vorbee's symptoms. Redrose64: I assume Opera doesn't have its own setting for keyboard layout, so after it has shifted to US layout, does it go back to following your OS setting after restart of the browser? --Pipetricker (talk) 17:11, 26 May 2018 (UTC)
Yes, closing the browser and restarting it (no need to reboot the PC) fixes the keyboard map. Both Opera and Firefox exhibit this kb remapping problem, but it happens more often with Firefox; and it can be triggered by visiting certain external sites (such as YouTube) which execute JavaScript. --Redrose64 🌹 (talk) 19:29, 26 May 2018 (UTC)
Vorbee, have you restarted your web browser or computer? Is the keyboard back to normal? If not, before restarting anything, please try pressing ⇧ Shift at the same time as the key to the left of the 1 key, to see if that produces a tilde. Then close and restart your web browser and try the normal tilde key. --Pipetricker (talk) 09:38, 27 May 2018 (UTC)
Thank you. I have tried the shift key and the key to left of 1, and am getting tildes when I do this. Many thanks,Vorbee (talk) 16:42, 27 May 2018 (UTC)
@L3X1: sorry, not plausible - the EF can not calculate edits over time, only atomic values like total edits. It could report every edit where: (User is registered) AND (User is not confirmed) and (user total edit count is over 50). You can discuss further at Wikipedia:Edit filter/Requested. — xaosfluxTalk01:07, 28 May 2018 (UTC)
Missing interlanguage links in sidebar
A reader asserts that the Portuguese version of many articles is not shown in the list of languages for many articles.
I see a link to the Portuguese version so I'm not quite sure how to help them. Asked for a screenshot but they didn't know what that meant.
I have a vague recollection that the displayed interlanguage links can be configured and perhaps they have it configured wrong, but this is just a guess.
I'm attempting to import the template {{efn}} to wikiversity, but I'm getting the error:
Import failed: A database query error has occurred. Did you forget to run your application's database schema updater after upgrading? Query: INSERT IGNORE INTO `page` (page_namespace,page_title,page_restrictions,page_is_redirect,page_is_new,page_random,page_touched,page_latest,page_len) VALUES ('10','Efn-lr','','0','1','0.873740892682','20180527065343','0','0') Function: WikiPage::insertOn Error: 1205 Lock wait timeout exceeded; try restarting transaction (10.64.0.205)
@John of Reading: That's almost solved it - thank you. Any ideas on what last steps should happen to finally render it correctly? Oddly it all renders correctly except for one of the inline superscripts is "[lower-roman 1]" in stead of "[i]". Any ideas? Thanks again, T.Shafee(Evo&Evo)talk23:15, 27 May 2018 (UTC)
Latest tech news from the Wikimedia technical community. Please tell other users about these changes. Not all changes will affect you. Translations are available.
Recent changes
You can now use global preferences on most wikis. This means you can set preferences for all wikis at the same time. Before this you had to change them on each individual wiki. Global preferences will come to the Wikipedias later this week. [45][46]
It is now easier for blocked mobile users to see why they were blocked. [47]
There is now a checkbox on Special:ListUsers to let you see only users in temporary user groups. [48]
Some rare invisible Unicode characters have recently been banned from page titles. This includes soft hyphens (U+00AD) and left-to-right (U+2066) and right-to-left (U+2067) isolate markers. Existing pages with these characters will soon be moved by a script. [49]
There's a new Wikimedia Foundation team to support the Wikimedia technical communities. It's called the Technical Engagement team. Most of the team members did similar work in other teams before this. [50]
Problems
Some translatable pages are showing old translations instead of latest ones. The cause of this issue has been fixed. We will update all pages automatically to show the latest translations. [51]
Changes later this week
There will be a new special page named PasswordPolicies. This page gives information about the password rules for each user group on that wiki. [52]
A new way to see moved paragraphs in diffs is coming to most wikis. This is to make it easier to find the moved paragraphs and the changes in them. [53]
The new version of MediaWiki will be on test wikis and MediaWiki.org from 29 May. It will be on non-Wikipedia wikis and some Wikipedias from 30 May. It will be on all wikis from 31 May (calendar).
Wikis can enable Citoid to provide automatic reference look-up in the visual editor and the 2017 wikitext editor. This is complex. The tool will now disable itself if the configuration isn't correct. It has warned about this in the JavaScript console since February. Check that your wiki is configured correctly. You can ask for help if you need it. [54]
Meetings
You can join the next meeting with the Editing team. During the meeting, you can tell developers which bugs you think are the most important. The meeting will be on 29 May at 18:30 (UTC). See how to join.
You can join the technical advice meeting on IRC. During the meeting, volunteer developers can ask for advice. The meeting will be on 30 May at 15:00 (UTC). See how to join.
Future changes
Content Translation drafts which have not been updated in over a year will be removed. This allows other users to translate those articles. [55]
A survey is collecting information on what users think about how Wikimedia wiki pages are loaded. This information could be used in future development. [56]
Some wikis will switch to use the Remex parsing library. This is to replace Tidy. It will happen on 30 May and 13 June. Wikis with fewer than 100 linter issues in the main namespace in all high-priority linter categories will switch. This includes Wikidata. Tidy will probably be removed on all wikis in the first week of July. [57][58]
This, You can now use global preferences on most wikis. This means you can set preferences for all wikis at the same time. Before this you had to change them on each individual wiki. Global preferences will come to the Wikipedias later this week., looks like it came from the Simple English wikipedia. Natureium (talk) 16:02, 29 May 2018 (UTC)
I think it's because English is the lingua franca of tech news and this notice gets sent to many wikis where users don't have a native level understanding. Killiondude (talk) 16:17, 29 May 2018 (UTC)
Indeed. Tech News is intentionally written in Simple English. Translations do exist, but occasionally it is not translated into certain major languages (for example, there never seems to be a German translation, and last week Spanish was missed). — This, that and the other (talk)03:08, 30 May 2018 (UTC)
Simple English, particularly avoiding complex grammar, may also make it easier for translators. BTW, if anyone who knows German or other languages wants to make sure a translation gets written in time, m:Tech/News states that each week's issue is available for translation on the Thursday ("end of day UTC") before it goes out. Anomie⚔12:35, 30 May 2018 (UTC)
Hello technical wizards. Any ideas on how to force display of a reference list before the reference is mentioned in the text? Don't worry, it's not for article mainspace! The code below renders ok in preview but gives an gives an error when the page is saved:
I believe that is officially unsupported with cite.php. You could probably fake it, though, by using an internal anchor link, if you didn't need automagic numbering:
[[#note1|1.]] First note.
Words.<sup>[[#note|1]]</sup>
The new improvement to <ref>...</ref> proposed at meta:WMDE Technical Wishes/Book referencing/Call for feedback (May 2018) would obviate the need for the {{Rp}} template, as well as provide various other enhancements. The discussion is presently swamped by people who just don't like fully-inline citations and only want to use {{sfn}} and page-bottom referencing, but this is a false dichotomy. The discussion isn't about which citation style is better (the answer to that is "it depends on the article"); the question is whether this feature would be good to have for referencing that is fully inline, and the answer is clearly "yes". I would be delighted if my old {{Rp}} template was finally superseded by an actual (and more tidy) feature of MediaWiki itself. — SMcCandlish☏¢ 😼 22:04, 1 June 2018 (UTC)
Can anyone tell me why the "Watch this page" and "This is a Minor edit" tickboxes in the edit view no longer work in Safari. When I click on them they fill in (blue) but aren't ticked. So, for instance, when I finish an edit and click the "Watch this page" box, the page is not added to my watchlist (evidenced by the star in the top right not being filled in). Shhhnotsoloud (talk) 09:33, 1 June 2018 (UTC)
its a Safari bug, there is a ticket in phabricator and a bugreport on webkit/safari too. Ppl are looking vot solutions. Luckily, they still work, its just the checkmark that is missing —TheDJ (talk • contribs) 10:56, 1 June 2018 (UTC)
@Ahecht: Was hoping for something that'd still be sortable at the end of the day. And something that would inherit all the properties from the top level, rather than have to specify every width in sub-levels to ensure consistency. Not sure if that's doable. Headbomb {t · c · p · b}09:52, 25 May 2018 (UTC)
@Headbomb: It should be technically possible with JavaScript, but I cannot read or write JS. I think I might have asked about something like this before, possibly to TheDJ about {{Routemap}}, but the solution in that template, for now, is also to nest tables. How would sorting a collapsible table work? If you were to press "collapse" after reordering the rows, what would happen? Jc86035's alternate account (talk) 09:35, 2 June 2018 (UTC)
It is bad practice (inaccessible) to have intermediary column-spanning items in a table as this moves tables away from data tables toward layout tables. If you want separate tables, make separate tables. If you want one table, these should be as row-spanning cells, presumably in the first column. --Izno (talk) 19:49, 2 June 2018 (UTC)
Search Index
Hi all,
So I made a page about 2 months ago. The page was reviewed by a patroller on May 29, as you can see here. [[59]] For some reason it still doesn't appear on Google searches. Does anybody know why? I'm a bit concerned because other pages I created that were reviewed didn't have this long of a lag time before appearing. Thanks in advance. Wikiman5676 (talk) 20:03, 2 June 2018 (UTC)
Dharma Bum Temple is the third Google hit for me on Dharma Bum Temple. I'm in Denmark. Google's cache says their current snapshot is from 2 Jun 2018 20:12:20 GMT. That's nine minutes after your post and may have been triggered by an edit [60] at 20:08. I don't know whether they had it indexed before. PrimeHunter (talk) 20:46, 2 June 2018 (UTC)
Wow so all i had to do was edit it once after it was reviewed? haha thats so funny. I'll remember this for future pages. Thank you for pointing that out, i appreciate it. Wikiman5676 (talk) 21:02, 2 June 2018 (UTC)
webrecorder.io for |archiveurl=
Archive.org, webcitation.org and archive.is are great at recording and playing back static web pages. But dynamic content pages often don't do well, such as Google Books, flash, YouTube, Facebook, Twitter etc...
Funded by the Andrew W. Mellon Foundation, Webrecorder is targeted towards archiving social media, video content, and other dynamic content, rather than static webpages. Webrecorder is an attempt to place web archiving tools in the hands of individual users and communities. It uses a "symmetrical web archiving" approach, meaning the same software is used to record and play back the website. While other web archiving tools run a web crawler to capture sites, Webrecorder takes a different method, actually recording a user browsing the site to capture its interactive features.
The quality and fidelity of webrecorder.io archives on replay are unmatched. It also allows for archiving multiple pages in a single archive. The downside is need to register for an account and a slight learning curve. The upside is the archives are fully open-source meaning the WARC files can be downloaded. If there are sites you want to archive but have not been successful this is a solution. -- GreenC03:47, 27 May 2018 (UTC)
@GreenC: Should there be a guideline for how to use it in Wikipedia, like the ones for the Internet Archive, archive.is and webcitation.org? It seems to be more complicated than the other web archives, especially since an account is required to archive pages. Jc86035's alternate account (talk) 16:00, 27 May 2018 (UTC)
Yeah that would be good. There are over 100 thousand YouTube links on enwiki. A sample video:
This is typical. Almost none of the YouTube links are backed up. They last 5-10 years. -- GreenC15:57, 29 May 2018 (UTC)
I do think that source archival is something that needs to be looked at more closely by the community. Webrecorder could be the best option for interactive sites, which seems to be on the main problems with our current archiving. The other main problem is when a site has died and it has not been archived at all, regardless of interactive or not interactive elements. Emir of Wikipedia (talk) 19:05, 2 June 2018 (UTC)
Another problem: when a page is archived, the underlying HTML code is from that era. As time goes on, the ability to replay that HTML degrades because standards change, browsers change, security changes. So pages from the 1990s don't display well on browsers from 2018. It gets worse with time. Webrecorder uses a virtual machine to emulate old browsers when recording/replaying so it is possible to replay a page as it was intended to be seen no matter the age. It's slick technology and the only archive capable of it. -- GreenC20:24, 2 June 2018 (UTC)
I came across several hundred talk pages that use the User:ClueBot III/ArchiveThis without the required "archiveprefix" parameter. According to the documentation this is supposed be okay, but yet the archive boxes on these talk pages are filled with links to unrelated pages, and the actual archives are nowhere to be found.
I made a list of all the affected talk pages User:Bradv/ArchiveFix. I'm not sure if my proposed fix actually works, as I tried one and it appeared to have no effect.
Please note this is also happening to user talk pages, although I have not attempted to index those. Here's a diff from a few minutes ago where ClueBot wiped out the archive index and replaced it with bogus data. Bradv04:34, 2 June 2018 (UTC)
Issue with editing - End of paragraph and "return"
Hi,
Been noticing the last couple of days that editing does not leave a clear space between paragraphs. In some cases, it also appears to ignore a para break/return and concatenate sections. Most frustrating. Regards, Cinderella157 (talk) 23:53, 28 May 2018 (UTC)
Please give an example and name you browser and your skin at Special:Preferences#mw-prefsection-rendering. Are you referring to the edit area in your browser while editing a page, or to a rendered saved page? If it's the edit area then which editor do you use? Does it happen if you are logged out? PrimeHunter (talk) 00:11, 29 May 2018 (UTC)
See:
1. Would remove “criminal” before Commissar order because it sounds as if that is Wikipedia’s voice, and also duplicates the end of that sentence re contravening laws of war. (btw would be good to clarify what we mean by “laws of war’ – ie what convention or international agreement set these laws in this instance). Alternatively, we would need a specific source for “criminal” as it has a particular meaning beyond violation of a convention – criminal under what national or international law?
4. Do we know what happened to his family after the war?
5. Overall, this is a comprehensive summary of Hoepner’s military career, with some interesting additions and quotes on his views and interpretations of different stages of activity. It would have been good to have a bit more about the nam as an individual (ie not just a soldier), and some further background on his early life. But we’re obviously limited by the soruces we have, which seem fairly comprehensively reviewed.
I am using WikiEd in vector. Does this answer all of your questions? Regards, Cinderella157 (talk) 00:37, 29 May 2018 (UTC)
See this also, posted by another ed indicating the same problem.
First: Purity of sources. I'm not convinced that looking for sources that deal specifically and only with a given event is helpful in contextualizing the horrors of WWII and German operations in Central/Eastern Europe, particularly in the General Government.
Second: Context is a way of understand the purification of Eastern Europe. Instead, he preferred to rely on one source to document the article, and to limit the article to its barest bones. There is no context of the operation, no sense of how and why it happened in the broader Nazi operation. It sounds to me more like a book report than an article.
Third: Generally, I see this dispute as, first, a sourcing dispute, complicated by personal styles of communication. We have, obviously, a conflict between the kinds of secondary sources acceptable. Second, we also have a dispute over the level of detail appropriate to an article.
I'm very glad this has been reported, something indeed has happened in the past few days that makes editing paragraphs with Firefox (and the Monobook skin, if that's relevant) a sorrow: it had worked perfectly for many years before that. An instance happened with this edit; it seems one needs to insert two line-breaks to create, and sometimes to maintain, a paragraph break, and trouble can happen both at the editing point and at the next paragraph break which sometimes also disappears. It's well wrong. Chiswick Chap (talk) 12:02, 30 May 2018 (UTC)
FYI, the maintainer has been so sporadically present and there are currently so many issue open with WikEd, that I consider it unmaintained, which means it will slowely degrade further as time moves on. If you are truly interested in continueing to use it, u should see about learning the code and making your own changes. Im happy to deploy them for you, but i dont have the time to maintain this tool myself. —TheDJ (talk • contribs) 13:37, 30 May 2018 (UTC)
WikEd's code hasn't changed, so there must have been an external change that broke it. It's the only editor we have, so it has to work. Hawkeye7(discuss)20:28, 30 May 2018 (UTC)
@Hawkeye7: “WikEd's code hasn't changed, so there must have been an external change that broke it.” Correct, a common misconception is that this means the problem is not in WikEd. The eco system changes continuously. Code that isnt maintained is on a slow path towards death. Its the only certainty there is. “It's the only editor we have, so it has to work.” Nah, we have CodeEditor, 2003 editor, 2008 editor, 2010 editor, 2017 editor, VisualEditor and the syntaxhighlight modes for WE2017 and WE2010 (available from Beta) (oh and floweditor, wikidateditor, pagesource editor [wikisource] etc etc etc) —TheDJ (talk • contribs) 22:17, 30 May 2018 (UTC)
I'm having the same problem with wikEd under Firefox. WikEd handles regular expressions, which is why it's especially useful to me. Which replacement editor does the same? My attempts to find out on my own indicates that only a subset of the above, if any, do. Dhtwiki (talk) 22:40, 30 May 2018 (UTC)
The default source editor and toolbar has regex. Click "Advanced" above the edit box and then a magnifying glass icon to the right. PrimeHunter (talk) 22:50, 30 May 2018 (UTC)
@Whatamidoing (WMF), I have used WikiEd mainly for editing, mainly because of its undo function but also, because it has a wider variety/easier access to markup code than the basic editor. What is most likely to be similar in functionallity and a basic point in the right direction please. Regards, Cinderella157 (talk) 02:16, 31 May 2018 (UTC)
I really don't know what to recommend, because personal preferences are personal. But looking at your recent contributions to articles, you seem to make a lot of small edits (instead of one big one) to individual sections, some of the articles use the sfn template family (which the visual editor doesn't support), and you don't seem to be using much markup. With that in mind, you might consider the default (2010) wikitext editor as the next one to try, and seeing whether any extra wikitext markup codes you want are present below the editing window. It should be faster to load than WikEd. You might also want to try the syntax highlighting beta feature for it. Whatamidoing (WMF) (talk) 22:43, 31 May 2018 (UTC)
I've been trying out the default (2010) editor, which I haven't used for years. (I'm using it now.) There is not much evidence that it has been maintained since then. It has two kinds of syntax highlighting; the default one works, although the colour scheme takes a lot of getting used to. The regexp replace works, in the Perl dialect. The find does not work; it does not reset the cursor at the find position, and has no reverse find function. It has the special characters characters menu to add the abominable en dash and em dash characters that the MOS requires, but does not render them so you can detect them. It's by no means unusable, but it isn't very good either. Hawkeye7(discuss)02:34, 2 June 2018 (UTC)
There also appears to be a bit of a problem with the generic editor. I am having to add two returns (ie a blank line) to get text to separate otherwise, with one return, text just runs on as if it is part of the same para. Regards, Cinderella157 (talk) 06:31, 31 May 2018 (UTC)
I don't see a problem here. For as long as I've been around (nine years), it's always been the case that a single newline will run text together into a single paragraph, and two newlines (one blank line) will start a new paragraph. See Help:Wikitext#Line breaks. BTW you should not leave blank lines between colon-indented replies, see WP:INDENTGAP. --Redrose64 🌹 (talk) 10:00, 31 May 2018 (UTC)
I concur with Cinderella157 and Hawkeye7, at least in regard to my use of the classic text editor in Firefox/Windows 7 (desktop): line spacing started misbehaving a couple of weeks ago,.
That is:
(i) Until a couple of weeks ago, return + blank line would always create a new paragraph, but I now have to insert a hard line break (e.g. <br>), to ensure that I don't get one giant paragraph. Even double blank lines vanish in some cases!
(ii) If I use indentation, as I am in this post (i.e. :::), I get an extra, unwanted blank line!
These bugs do not seem to occur with the only other platform that I use to edit articles – classic text editor in Chrome/Android (phone, which is less user-friendly than the desktop, obviously).
Anti-vandal bots do not perform precise, exact work like most other bots do. They act more like humans, with most edits correct and good, but a small percentage of mistakes. Bot edits show up as (unflagged) human edits so they can be reviewed for possible mistakes if necessary, like other human edits.
And yet I still don't want to see them when I filter my watchlist. ¯\_(ツ)_/¯ What then was the point of all the effort spent on improving watchlist filtering? — Scott•talk11:24, 1 June 2018 (UTC)
Respectfully, that's both a non-solution and is terrible. Firstly, it doesn't appear (I haven't tried it) to provide a visibility toggle, but a permanent hiding, which makes it technically inferior to the watch list's built-in functionality. It's a kludge. Secondly, it relies on the user being a technical type, along the lines of you and I, who's capable and willing to implement it on our own behalf. That's exactly the kind of hostility to general users that the gradual overhauling of MediaWiki's interface is supposed to be doing away with. I'm perfectly capable of installing yet another technical doodad myself, but I'm not raising this purely on my own behalf. The lack of bot flag on these bots affects anyone who uses their watch list, and it's not proportionate or contemporary to impose this addition technical burden on them to cope with a deliberate exclusion of the bots from an improved part of MediaWiki's capabilities. — Scott•talk15:21, 3 June 2018 (UTC)
So far the community has deemed ClueBot edits to be something that should not use the botflag to be hidden from editors, per the rationale you've pasted above. WP:CCC, of course, and you're welcomed to create an RFC on the issue, but if you want a solution to your problem, you have one. Headbomb {t · c · p · b}22:40, 3 June 2018 (UTC)
Introducing Toolhub
What does your participation on the Wikimedia projects look like? Do you edit articles? Upload files? Patrol vandalism? Translate articles? Translate interface messages? Do you organize people, online or offline? Do you train new editors, or new trainers? Do you write code?
There are many different ways to contribute to Wikimedia – more than you would expect just from reading Wikipedia articles. Over the past several years, volunteers have developed technical tools that help Wikimedians improve content, patrol vandalism, and perform many other tasks. They make it possible to do what the wiki software alone cannot accomplish. Without these tools, many of our projects would slow down to a crawl.
I am very happy to announce a new project called Toolhub which seeks to create a searchable index of these tools in all languages. We are building this tool catalog based on what our communities need. If you would like to help, please take a look at m:Toolhub and review the question at the top of the page. You can also leave feedback in any language on the talkpage. You can also email me private feedback. Harej (WMF) (talk) 23:23, 2 June 2018 (UTC)