Computing desk
< June 8	<< May \| June \| Jul >>	June 10 >

Welcome to the Wikipedia Computing Reference Desk Archives
The page you are currently viewing is an archive page. While you can leave answers for any questions shown below, please ask new questions on one of the current reference desk pages.

June 9

File hosting services

Hi. I've been trying to find a file hosting service that meets the following criteria. Can you please help me?

The service must have unlimited file storage.
The service must have no maximum file size.
The service must have "direct access" to downloads, with no CAPTCHA or wait time to download a file.
The service must not have a traffic or bandwidth limit.
The service's files may not expire (be deleted off of their servers) after a period of time. (However the service may delete the file based on inactivity since the file was last downloaded.)
And, most importantly, the service may not charge for the above services.

I would appreciate your help. Thanks! Samwb123^{T (R)-C-E} 00:58, 9 June 2010 (UTC)[reply]

Hi, after you have found a suitable file storage service, can you find a plane for me as well? I want it to be able to go to any place on earth in less than one hour, take up to 10000 passengers, use water as fuel, gold plated and encrusted with diamonds and most importantly, be free? thanks 203.167.190.79 (talk) 03:10, 9 June 2010 (UTC)[reply]

No need to be sarcastic! There IS a file hosting service that meets those criteria: it's called hosting your own server. :) Check out droopy, for instance. Sorry, Samwb123, nobody else will let you use unlimited amounts of their storage and bandwidth for FREE. Indeterminate (talk) 03:47, 9 June 2010 (UTC)[reply]

Honestly, that would have been similar to my response too. Hosting your own server doesn't meet half of the OP's requirements. Unless you ignore the "most important" one. The OP is going to have to be flexible on at least a few of the requirements before this becomes even remotely feasible. Vespine (talk) 05:49, 9 June 2010 (UTC)[reply]

Google documents? It allows users to store music, documents, pdf, etc, and share them. --Tyw7 (☎ Contact me! • Contributions) Changing the world one edit at a time! 04:32, 9 June 2010 (UTC)[reply]

Except for the maximum file size, mediafire meets all these requirements. If your file is larger than 200mb (or 2gb if you're on their premium service) then you can just rar them into smaller chunks then upload. 82.43.89.11 (talk) 10:15, 9 June 2010 (UTC)[reply]

The OP asks for "no maximum file size." The OP should rethink this limit, unless they are interested in designing the famous infinite data compression algorithm. Disks have finite size. File systems and operating systems have maximum addressable file offsets. Even if you could convince somebody to pay for the service, buy the disks, attach the storage with superfast network, when you hit files larger than, lets say, 10,000 petabytes, you are going to have some serious trouble using them. The same goes for "unlimited" bandwidth; unlimited storage time, and so on. Serious data archivers, like the Library of Congress, have spent considerable effort investigating the technical and economic feasibility of large, long-term storage: see, e.g., www.digitalpreservation.gov/ - but even they do not make outrageous claims about infinite storage and infinite bandwidth. I think what we can conclude from this and other observations is that the OP has not carefully thought out his/her requirements: they should reconsider exactly what they need, and bear in mind that outrageous requirements carry outrageous price-tags. Nimur (talk) 16:13, 9 June 2010 (UTC)[reply]

I agree with the last sentence in particular—you'd get more useful results on here if you actually told us what it was you were planning to do with it. It would be easier for us to actually suggest services that worked within the reasonable practical limits of whatever your project is, or to let you know why some of them really can't work out. What you're asking for does not exist in the terms in which you are asking for it, and the reasons for it not existing are fairly obvious (bandwidth and space costs somebody money, so any service that offered up literally unlimited space and bandwidth would either have to be run as a charitable institution or would run itself into the ground). There are services that can approximate some of the requirements within reasonable limits—i.e., upper caps on file sizes or total storage, or having ads before the link, etc. But you'll have to be more specific about what you are using it for, otherwise it is as ridiculous as the request made by 203.167. --Mr.98 (talk) 19:11, 9 June 2010 (UTC)[reply]

Actually, I just found something that meets all of those criteria (except the bandwidth one, but still, it has 2 GB of bandwidth, and that's a lot). See http://www.filehosting.org/. Samwb123^{T (R)-C-E} 01:27, 10 June 2010 (UTC)[reply]

I think we can assume the poster wouldn't charge themselves for use of a service then buying their own equipment does satisfy all the constraints. ;-) Dmcq (talk) 12:17, 10 June 2010 (UTC)[reply]

Indifference curves

I want to visualize indifference curves in 3D. So I need a graphing calculator that is capable of doing contour graphs. For example, let U(x,y) = X^0.45 * Y^0.55, I want to plot the indifference curves where U = 10 and 20. I don't want grids.

Is there any free software for this purpose? -- Toytoy (talk) 05:14, 9 June 2010 (UTC)[reply]

Sounds like something that can be done using GNU Octave. Titoxd^{(?!? - cool stuff)} 07:51, 9 June 2010 (UTC)[reply]

Here are some resources that I found while searching: an archived Math Desk discussion about Octave and economic indifference curves; the Octave Econometrics package (which does not have indifference curves, but may be useful anyway); this handy Octave plotting reference; and of course, because Octave provides a MATLAB-like interface for almost all common plot functions, you can read about MATLAB 3D line-plotting and 3D surface-plotting, and see how many of those commands work in Octave (most everything should be compatible or only slightly different). You can fall back to the built-in docs inside Octave, or check the Octave plotting documentation. Nimur (talk) 20:10, 9 June 2010 (UTC)[reply]

iSCSI SAN

I have a few questions in regard of iSCSI Storage Area Networking, and I want to learn everything that I can about iSCSI SANs

As iSCSI initiators are connected to iSCSI SANs over the internet, without any physical connection, is the limit on the number of hosts an iSCSI SAN have fixed to specific hosts, or is it a limit as to how many hosts can be connected any one given time?
How does the iSCSI SAN know which parts of its pool of its storage belongs to which hosts? Unlike a NAS, where this information is stored onboard the server, where is this information stored in regard of the iSCSI SAN? —Preceding unsigned comment added by Rocketshiporion (talk • contribs) 06:44, 9 June 2010 (UTC)[reply]

Reformatted and title added. --217.42.63.244 (talk) 07:20, 9 June 2010 (UTC)[reply]

The iSCSI system uses a pretty big ID (similar to a GUID) to track initiators so there is no practical limit on connections other than what the host is willing to allow (meaning, many more than you would actually *want*). This is usually licensed by the host vendor in a realistic way, and limited because each initiator must be assigned a storage partition or group in a 1:1 way (1:many and many:1 is possible but is basically an extension of 1:1). The host controller has a lot of intelligence (basically an embedded computer) to allow it to keep track of local disks and clients. However, the iSCSI system is very fragile; in my experience you are nuts to want to truly use it across the internet, especially with a lot of hosts. It is a very low level protocol so it won't stop you from doing foolish things like assigning two systems with non-concurrent filesystems to the same storage partition, and having them subsequently destroy each other. For more information on this it's useful to investigate the respective filesystems you plan on using; iscsi doesn't really care. --144.191.148.3 (talk) 16:18, 9 June 2010 (UTC)[reply]

I want to set up a Storage Area Network in Singapore, and have initiators from other parts of Asia, as well as Europe and the Americas, be able to connect to it. My SAN is to start off with a capacity of 2TB, and eventually scale to up to 100TB.

I expect to initially have around 100 initiators, and eventually expand to up to 5,000 initiators. I intended only to have unique partition(s) assigned to each initiator; i.e. either a 1:1 or 1:many relationship. As iSCSI does not appear to be a suitable protocol, do you know of any other protocol which can be used across the Internet for this purpose? In addition...

I wasn't even aware that many:1 relationship between initiators and targets is possible. How does such a relationship work; wouldn't the different initiators overwrite each other's data?
I am not familiar with what is a storage group; I'm only aware of storage partitions.
Does anyone know of any website where everything about Storage Area Networking is explained? I've tried the SNIA website, but it only extolls the benefits and gives a very basic overview of SANs.

Thanks in advance.116.197.207.98 (talk) 23:52, 12 June 2010 (UTC)[reply]

how to simulate a handoff scenario?

i was studying the different handoff techniques in wlan ieee 802.11x and was thinking about an idea of my own which i think might in its own way be able to reduce the handoff latencies in MAC layer scanning. But i need a simulation to find out if it would really work out. Should i use matlab for it? Does it provide built in functions for ieee 802.11 nerworks and specifically for handoff scenarios?? Or do i have to write codes for it? Can anybody please explain how such simulations work? An example or a link would be very helpful. Thanks. --scoobydoo (talk) 07:37, 9 June 2010 (UTC)[reply]

If you have consistent data to model handoff, I would *love* to see it. I have done work on 802.11 networks many times in the past, and handoff is such a crap-shoot of idiosyncrasies between each brand of access point and wireless client device that there's no good way to predict it; you just have to set it up, test it, and hope for the best when the users start to swell. --144.191.148.3 (talk) 16:25, 9 June 2010 (UTC)[reply]

So you are saying it is best to go for real time experiments with different brands of APs and MSs? Won't computer simulations work? Actually i was thinking of a handoff based on gps measurements of the postion of the MS and hexagonal cell structure of AP coverage areas. But i don't know whether it will work out. I thought maybe if it was possible to simulate handoffs through programming and stuff. Could not find anything in matlab communication toolbox or simulink blocksets... --scoobydoo (talk) 17:31, 9 June 2010 (UTC)[reply]

Not knowing that much about matlab, I would think you at least need to provide some constants like handoff probability related to proximity to adjacent stations, handoff speed for various signal strengths (and given certain negotiated speeds, vendors, assumptions for client activity, etc.) and other figures that can't be mathematically derived. Then, as your user count increases (as each user affects the others' ability to see base stations) you need to almost move to a finite element approach where you can take all these things into account for a snapshot model. It sounds super duper hard, but if you can pull it off you will probably have a model worth selling to Cisco or other wireless big-names since they are very interested in software approaches to optimizing wireless networks. There are only a few companies that have even tried the software optimization approach (the assumed goal of this exercise), and they are far from perfect. Personally (given that I have done this more than a few times) I would say real world testing (and ways for base stations and clients to react to real world indicators when negotiating handoff) will always trump computer models; there are simply too many variables like interference, objects blocking the signal, behavior of other stations during certain load conditions, etc. --144.191.148.3 (talk) 19:05, 9 June 2010 (UTC)[reply]

Time to copy files

Why does copying thousands of small files that total 100mb take longer than copying one 1gb file? This is on Windows 7, using the crappy default copy service. 82.43.89.11 (talk) 10:24, 9 June 2010 (UTC)[reply]

My guess is be that the default copy service is not optimised for bulk copying, and so does a target directory update after each individual file copy. The longer copy time would then be due to the additional overhead of thousands of directory updates. Gandalf61 (talk) 10:54, 9 June 2010 (UTC)[reply]

Is there any way to reduce this overhead? 82.43.89.11 (talk) 11:10, 9 June 2010 (UTC)[reply]

On this topic: make sure you have large areas of open space on your hard drive, otherwise the write head will be frantically scanning back and forth, trying to find space. As a further note, it will help massively if the data starts on a different hard drive to begin wtih (unless you're cutting and pasting, in which case you definitely want the same hard drive, if possible) Riffraffselbow (talk) 13:38, 9 June 2010 (UTC)[reply]

Riffraffselbow, write heads don't scan back and forth looking for free space. Free space is found by searching the volume bitmap, which is easily cached in RAM (the bitmap for a 1 TB volume would typically be 30 megabytes), so this search takes no time at all by disk-writing standards. If the free space on the destination drive is in small fragments then the write will take longer because the disk head will have to seek between available regions, but it's seeking to a precomputed track, not searching for a free one. Gandalf61, metadata updates are not under application control. The OS always caches them and sends a bunch to the disk at once. Pre-creating zillions of files could easily make things worse, since the metadata for each file would probably be written to disk before the file was written, and would then have to be updated later. If you create, write, and close each file in a short time, there will probably be just one metadata write per file. Mr.98, it's hard for me to believe that Firewire latency would noticeably affect the copy time. Firewire hard drives work at the disk-block level, not the file level, so there isn't a per-file wait time. It might be different when copying to a network share, though. I don't know how SMB works, but there could be one or more network round-trip waits per file. -- BenRG (talk) 23:16, 9 June 2010 (UTC)[reply]

My understanding is that it's kind of the equivalent of sending one large package through the postal service and sending 1000 small ones. The overall data volume/weight might be the same, but the postal service is going to have a much easier time processing the one big one (look at address, note proper postage, put in right bin) than the 1000 small ones (where each one has a different address and postage). Sure, you might need a bigger fellow to carry that one big one, but you only have to process one package in the end. I find this is especially so when using external harddrives with Firewire connections, where the speed of the transfer of the file data is very fast, but the speed of opening up a new file for writing, and then closing it again, is very slow when multiplied by a thousand. --Mr.98 (talk) 11:59, 9 June 2010 (UTC)[reply]

There's two reasons. Firstly, and probably chiefly, just because those thousands of files are all in the same folder there's no guarantee that they're stored on contiguous clusters on the disk; indeed, it's very likely that they're not, and often that they're distributed (seemingly at random) across the entire disk surface. When the copy program opens each file, the disk head has to seek (move across the disk surface) before it can read the next block - this seek time on a modern hard disk is something around 7ms. Then it has to wait for the data it wants to spin around - this rotational delay averages at around 3ms. So, on average, every time the next file (strictly the next cluster of data) isn't stored adjacent to the last, the disk has to wait for 10ms, which means it's not reading any data during that time. If the files (strictly, clusters) are distributed fairly randomly across the disk, this delay will dominate the time actually spent reading data, and performance will be very slow. OS designers know this, of course, and built layers of caching (and often readahead) to help minimise this, but if the distribution is really random, caching doesn't really help at all. Strictly this can be a problem even for that one 1gb file too, as there's no guarantee that its clusters are adjacent to one another either - but the filesystem takes a very strong hint and tries its best (bar the disk being very fragmented already) to keep things either in one or only a few contiguous runs. Secondly there's the problem of clustering overhead. If you create a file, it takes up a whole number of clusters on the disk; even a 1 byte file takes up a whole cluster. Cluster size on NTFS is 4kb by default. In practice the block device layer of the OS, on which the filesystem is built, deals in whole clusters, so it will read and write the whole cluster when you read that file. If the files each take a small fraction of the cluster size, most of that read and write is overhead. Large files make full use of the clusters, so they have minimal cluster overhead. -- Finlay McWalter • Talk 15:19, 9 June 2010 (UTC)[reply]

I would say that with SMB (especially in windows) the problem of per-file overhead related to the directory information and file handling information discussed here has more to do with it if you files are 100kb or larger, if not the slowdown will be the disk. Say we are going with the 100mb of 1000 files figure, that's 100kb per file. A modern disk can still read 100kb files at 10 MB/sec or better, the really egregious slowdowns don't happen until the files are around 5kb in size, when they are scattered all over the disk. Want to combat this? Archive your files; in a simple way like the tar format or in a compressed way like the zip format. If you really do have 100,000 1KB files (100mb worth), the time it takes to zip on one end and unzip on the other end (since disks are pretty fast these days) will pale in comparison to the time it takes SMB to wade through that many files. SMB is going to have more overhead than your disk in almost any practical case where multiple files are involved, I have observed this many times. --144.191.148.3 (talk) 16:37, 9 June 2010 (UTC)[reply]

In NTFS, the content of small files is stored in the metadata record itself, and doesn't use a cluster. Explorer will report these files as using 4K (or whatever the cluster size is), but it's wrong. Also, although disk space is allocated in cluster units, NT will only read or write the sectors that it actually needs, as far as I know. Since NT's caching is tied to x86 paging, which uses 4K pages, you will usually end up with reads of 4K or larger anyway; but at the end of a file, where NT knows some of the sectors in the cluster are meaningless, it won't read or write those sectors as far as I know. -- BenRG (talk) 23:16, 9 June 2010 (UTC)[reply]

Getting video from my camera NOT using Firewire

Hi, I have a mini-dv video camera that has Firewire as its main output. I just replaced my MacBook (old one was damaged beyond repair) and the new model does not have a Firewire port. Is there any other way I can get video from my camera to my laptop? Cheers, Joe^Talk_Work 11:58, 9 June 2010 (UTC)[reply]

Since you didn't give us the model number of your camera, or anything else that can help us help you, you can answer this better then us. Does the camera have any port other then Firewire, like USB for example? If not then obviously you'll need to add a Firewire port to your laptop in some way (USB to Firewire if that exists (a quick search suggest they may but wasn't able to find anything for sale that was clearly what was wanted), PCMCIA, ExpressCard, which options are available to you will depend on your laptop for starters), or use a different computer or device or buy a new camera. It occurs to me since Firewire allows devices to connect to each other, you may be able to connect the camera to an external hard drive with Firewire and transfer directly and then presuming the hard disk has Esata and/or USB which your laptop also has you can then transfer it to the laptop but I don't know if that is generally possible and in any case, it will likely depend again on you camera which as I've said we don't know (and a good way to find out whether it can may be to read the manual). If the camera has a removable hard disk or other form of storage, it may be possible to buy something which can connect that to your laptop in some way, but again it will depend on your camera. Nil Einne (talk) 12:12, 9 June 2010 (UTC)[reply]

Firewire to USB adapters and cables certainly do exist. They're quite handy. --LarryMac | Talk 12:38, 9 June 2010 (UTC)[reply]

Some cameras have a memory card, or the option to add one and the software to move files from hard drive to card. I assume that yours doesn't because this would solve your problem easily. Dbfirs 14:43, 9 June 2010 (UTC)[reply]

Sorry I didn't give much information before. The camera is a Canon MV890. It has only Firewire as its output to PC, with a separate output to TV. I think I will get a Firewire to USB adapter. Thanks for the replies, Joe^Talk_Work 16:55, 9 June 2010 (UTC)[reply]

Yes, unfortunately, it has no USB or memory card. Your only other option would be to use a digital recorder to record from the DV output to a DVD or hard drive that you could read on your PC. Dbfirs 17:28, 9 June 2010 (UTC)[reply]

Program to Find Primes in C++

I have created a program in C++ for finding prime numbers... This is a very simple program and so, I think you all can understand the gears and wheels of it... I was able to find the first 1000 primes in a little less than 1 second... Please tell me if there is a better way to find prime numbers... Just tell the way and I will program it myself... Do people find larger and larger primes only this way??

#include <cstdio>
#include <cstdlib>
#include <iostream>
using namespace std;
int main(int nNumberofArgs, char* pszArgs[])
{
    unsigned numberOfNumbers;
    cout << "Hey Donkey!! Type the number of prime numbers to be printed and press Enter key : ";
    cin >> numberOfNumbers;
    unsigned subjectNumber=2;
    unsigned printedNumbers=1;
    unsigned tester=1;
    unsigned hit=0;
    while (printedNumbers<=numberOfNumbers)
    {
          hit=0;
          tester=1;
          while (tester<=subjectNumber)
          {
                if (subjectNumber%tester==0)
                {
                                            hit++;
                }
                tester++;
          }
          if (hit<=2)
          {
                     cout << subjectNumber << "               " << printedNumbers << "\n";
                     printedNumbers++;
          }
          subjectNumber++;
    }
    system ("PAUSE");
}

harish (talk) 12:59, 9 June 2010 (UTC)[reply]

The basic way to find primes is the Sieve of Eratosthenes; it's a lot faster than what you're doing, because it stores all the primes it's previously discovered, and only tests candidates by dividing by these. Implementing that in C++ would generally mean you'd keep a store of all the primes you'd found and your inner loop would use these, rather than tester. (Incidentally your code does lots of pointless work too, because it doesn't terminate the inner loop when it sees a hit). Beyond the Sieve of Eratosthenes (which is simple to understand and implement, but not the fastest possible) there are things like the Sieve of Atkin, and many things listed at the Primality test article. Note that for some applications of prime numbers, people don't actually generate numbers that are definitely prime, but ones that are probably prime. -- Finlay McWalter • Talk 13:19, 9 June 2010 (UTC)[reply]

Harish, please do not cross-post the same question on several Reference desks.—Emil J. 13:30, 9 June 2010 (UTC)[reply]

Reading your code, there are a number of optimizations that are "low hanging fruit" that you might want to experiment with, rather than switching algorithms to the Sieve or something else; and I like working on prime number detector optimizations because it's high school math and is easy to think of small improvements and learn about optimization. (The Sieve of Eratosthenes is going to be faster than this method, ultimately; but it's not suitable for some applications — the Sieve has to start at 2, wheras harish's method can start calculating at any arbitrary positive integer; and harish's method just works, and is good for learning.) Some things I would consider if I were you:

Right now you're dividing the potential prime number by every single number lower than it, but you know that if you get a single "hit" (where the mod yields a 0) then it's already not prime, and you don't need to test that number anymore. Avoiding all those extra divisions would be useful.
Similarly, you know you don't have to do any more testing after you reach half the value of the number. (11, for example, isn't going to be evenly divisible by any number greater than 5.5.)
You also know up front that no even number is going to be prime, so you could avoid even testing any of these by starting at 3 and then incrementing by 2 each time.
You could do a little of what the Sieve of Eratosthenes does by keeping an array around of the primes you've already discovered, and only bother to divide each potential prime by the numbers in the array. This would speed up the evaluation by never dividing anything by 9 or 10, for example.

Comet Tuttle (talk) 16:34, 9 June 2010 (UTC)[reply]

Read The Art of Computer Programming Volume 2. Zoonoses (talk) 12:35, 10 June 2010 (UTC)[reply]

Not just half but stop at the square root of the number. That should cut down the number by alot. Also I tried this once in Java, screwing around like you, and I'm not sure about how much faster using a list is. Not sure if it's because I used a parameterized ArrayList but it actually ran slower when I included it. 66.133.196.152 (talk) 09:14, 11 June 2010 (UTC)[reply]

UTF-8 and HTML

Hi, I've read UTF-8, Character encodings in HTML and Unicode and HTML, but I'm still kind of confused. When I save an HTML document as UTF-8 and include <meta http-equiv="Content-Type" content="text/html; charset=UTF-8"> in its source, do I still have to use &somecodehere; for special characters like German umlauts (ü = ü, etc)? -- 109.193.27.65 (talk) 12:59, 9 June 2010 (UTC)[reply]

No, the utf encoding should take care of it; the HTML4 standard says "As some character encodings cannot directly represent all characters an author may want to include in a document, HTML offers other mechanisms, called character references, for referring to any character." which I take to mean that, if the character encoding does do what you want, it's your choice as the page author. But there's always the worry of old browsers and wonky search engines that don't understand utf properly ... -- Finlay McWalter • Talk 13:09, 9 June 2010 (UTC)[reply]

Naturally, that is assuming you really do represent the Ü correctly in UTF8; that means you've verified that a text-editor with which you edit the html file honours the UTF encoding properly, and any database that you store the character data in (e.g. for a blog posting) also honours the encoding correctly. -- Finlay McWalter • Talk 13:39, 9 June 2010 (UTC)[reply]

Google Chat

How many MBs does one hour of the following consume???

1. One hour of pure typing chat using GTalk. 2.One hour of voice chat using GTalk. 3.One hour of video chat using GTalk. —Preceding unsigned comment added by 117.193.155.79 (talk) 15:47, 9 June 2010 (UTC)[reply]

This question is mostly unanswerable since it depends on what happens during that one hour for each. In particular if the two parties type at a constant rate of 120 words per minute the data usage is going to be significantly different from if the two parties are typing at an average of 5 words per minute (whether very slow typists or more likely they aren't constantly typing but reading and replying and perhaps doing other things in between). The data usage will still be small but there could easily be an order of magnitude difference. I don't know if Gtalk varies the voice codec but even if they don't many modern voice codecs have silence detection and other things which means the rate will generally vary too. Video is probably the worse. I'm pretty sure Gtalk as with many video conferencing utilities varies the quality automatically based on several things including available bandwidth, potentially computer speed and camera resolution+frame rate. If you both have symmetrical 100mbit connections with 720P video cameras and very fast computers you're likely to have a far higher bandwidth and therefore data usage then if you both have 256k/128k connections with a typical VGA camera on a netbook. P.S. In relative terms, the voice will always be a lot more then the text and the video quite a big higher then the voice Nil Einne (talk) 17:05, 9 June 2010 (UTC)[reply]

Surfing with Python

How do you say in Python:

-go to url x and open it into a new Firefox tab
-push that javascript button on the page x.
-fill the field on page x and push the ok button

--Quest09 (talk) 17:28, 9 June 2010 (UTC)[reply]

I think what you're looking for is automation. Take a look at this thread. Indeterminate (talk) 17:51, 9 June 2010 (UTC)[reply]

If you instead want a Python program that does the same thing instead of a Python program that takes control of Firefox in order to do what you need to do, you need to know more about what is actually happening behind the scenes. How are the fields sent to the server? (POST (HTTP)? GET (HTTP)? AJAX? JSON?) What format is it in? What other parameter (user-agent? HTTP Cookies? Referrer (HTTP)? etc.) does it pass? After you've figured all that out you can write a simple python script that does exactly what Firefox does behind the scenes using the httplib and urllib in Python. --antilived^{T | C | G} 05:37, 10 June 2010 (UTC)[reply]

alternatively, from the perspective of an end-user with limited programming experience, something like autohotkey might be more useful. Just record a macro that fills in each field/clicks the buttons, selecting them via tab. Then loop said macro. Riffraffselbow (talk) 07:01, 10 June 2010 (UTC)[reply]

Follow-up: and what is the equivalent to autohotkey for Linux?--Quest09 (talk) 15:57, 10 June 2010 (UTC)[reply]

Searching for autohotkey linux in Google produces plenty of promising results which, as I do not use Linux yet, I do not have the motivation to explore. 92.15.30.42 (talk) 11:37, 12 June 2010 (UTC)[reply]

email info

please help me to rename my secondary email in simple terms, as i am pretty new at computers, windows live help, said go to your account summary page with your windows live id, and i don't know how to do that? —Preceding unsigned comment added by Saltyzoom (talk • contribs) 18:51, 9 June 2010 (UTC)[reply]

Your question is unclear. You don't say what it is that needs you to rename your secondary email, or which company or organization you have this account with. Astronaut (talk) 01:15, 11 June 2010 (UTC)[reply]

Internet

I'm sure this has been asked before, in fact I'm certain of it because I remember seeing a thread here, but I can't find it in the archives. Anyway. Is there such a program for windows that can monitor every webpage the computer visits and save all the pages, images etc. Sort of like a web spider except it only saves what your browse. Sort of like building an offline internet of every page you've ever visited. 82.43.89.11 (talk) 21:18, 9 June 2010 (UTC)[reply]

Maybe Wikipedia:Reference_desk/Archives/Computing/2009_June_14#Search_whole_site and Wikipedia:Reference_desk/Archives/Computing/2007_May_16#saving_web-pages_in_one_file are the questions you remembered seeing here? -- 109.193.27.65 (talk) 22:09, 9 June 2010 (UTC)[reply]