Wikipedia:Reference desk/Archives/Computing/2011 March 13

= March 13 =

Google searches
I have a recurring problem that I do a Google search, find what looks to be the right article, from the preview on the Google page, then, when I pick on the article, I can't find that preview anywhere within it. So, what is happening ? Here are some possibilities:

1) The article has changed and no longer matches my search criteria.

2) Some type of error at Google, causing the articles to be mis-indexed. If this is the case, how do I keep from wasting my time with Google articles which are mis-indexed ?

3) The preview is contained within the large article, but the "page find" fails to locate it, for some reason.

Here's an example. In response to a request to prove that animals with higher metabolic rates also have a higher rate of cell division, I used the following search terms, together: "metabolic rate" "cell division". The top article sounded promising, with this preview: "Because metabolic rate and the rate of germ-cell division increase in smaller species, the association of fast rates of molecular evolution with small ...". There was no cached version, but I tried both the regular PDF and the Google Quick View, and couldn't find it under either. StuRat (talk) 00:03, 13 March 2011 (UTC)


 * If you scroll to the bottom of the SERP, you can click on "Give us feedback" and open this page.
 * —Wavelength (talk) 00:16, 13 March 2011 (UTC)


 * Thanks, I sent them my suggestion that they cache the OCR'd document. StuRat (talk) 13:10, 14 March 2011 (UTC)


 * You can find the sentence in the PDF at the bottom of page 4090. This PDF is not searchable, probably because it represents the pages as images rather than as arrays of text. Looie496 (talk) 01:49, 13 March 2011 (UTC)


 * Thanks. But how did Google manage to search the page if we can't ? StuRat (talk) 09:14, 13 March 2011 (UTC)


 * Might they run OCR software on the image? Is there any free software available to do this from a PDF image (or other image) without printing?    D b f i r s   11:05, 13 March 2011 (UTC)


 * Ah, yes, some details here, here, and here (for example). Sorry, I was too tired to search!   D b f i r s   07:40, 14 March 2011 (UTC)


 * Google runs their own OCR software on PDFs that they then use to index it. Obviously they can't replace the original file, though, which did not have OCR run on it. So when you open the original PDF, it will look just like a bunch of images. If you run OCR software on it, then it becomes searchable. --Mr.98 (talk) 11:49, 13 March 2011 (UTC)


 * Although it doesn't seem to be what's going on here, the ACM OCRs the old papers they publish and put the OCRed text invisibly into the resulting PDFs, so you can read the original text, but you can select (and search engines can see) some approximation to it. Paul (Stansifer) 18:03, 13 March 2011 (UTC)


 * Yes, Google needs to provide a cached version that's like that, rather than "toss out" the OCR'd version. That's a waste. StuRat (talk) 09:50, 14 March 2011 (UTC)


 * A waste of what? 40 seconds of processor time? (How fast we take things for granted! It used to be that if a PDF was uploaded but not OCR'd, you'd just never have a chance to search its contents via Google. Now you can do that, and it is judged insufficient, for Google does not then host a separate version of the PDF!) --Mr.98 (talk) 15:48, 14 March 2011 (UTC)


 * 40 seconds of processor time multiplied by all the scanned documents Google processes, and then discards, is a massive waste of computer resources, yes. And Google often hosts cached versions, so doing this in the case of OCR'd PDFs, where it's more useful, certainly isn't out of the question. StuRat (talk) 22:18, 14 March 2011 (UTC)


 * By the way that refers to germ cell division, which are not the same as ordinary cells. 92.24.186.239 (talk) 13:52, 13 March 2011 (UTC)

Thanks, everyone, I've marked this as resolved ! StuRat (talk) 13:11, 14 March 2011 (UTC)

Dell Latitude d620 CPU replacement
Resolved&#58; Magog the Ogre 2 (talk) 04:50, 14 March 2011 (UTC) The power socket on my laptop (see title for model) has gone bad. A friend and I tried to solder it off and replace it, but it proved undoable as the parts are glued on as well as soldered. As such, I want to buy a new motherboard. However, I can't tell if the board has the CPU still attached.

Assuming it's not attached, would it be reasonably possible to remove the CPU from the old motherboard and reattach it to the new one?

No need to advise me to ask the seller on if it's attached or not. I have already done so. As such I'd appreciate if you could handle the question above in lieu of going on an aside regarding contacting the seller. --Magog the Ogre (talk) 12:24, 13 March 2011 (UTC)


 * Looking at the same item on the non-mobile eBay page I can enlarge the photo and clearly see the CPU is not installed in the pink socket to the left of the fan. The CPU should be easy to remove from its socket in your broken laptop.  However, note that changing the motherboard on a laptop is a very tricky business - in my opinion, the service manual is an essential guide to dismantling and reassembling your laptop.  Astronaut (talk) 14:37, 13 March 2011 (UTC)

Well, in fact you're right (I couldn't see the page from my mobile device). Now I will have to remove the processor and put on a new one, which scares me. I may ask for advice shortly on that below. Magog the Ogre 2 (talk) 04:50, 14 March 2011 (UTC)

php
In an above thread I was recommended to use something other than Windows cmd.exe, but what would be a good alternative? The only other thing I have very basic knowledge of is php. Can php do everything cmd.exe can do? How would you run a program, like say wget --page-requisites --span-hosts --convert-links -P "%date%%time%" "http://en.wikipedia.org/wiki/Main_Page" in a php script? 82.43.92.41 (talk) 15:48, 13 March 2011 (UTC)
 * Have you tried Cygwin? --Magog the Ogre (talk) 16:07, 13 March 2011 (UTC)--Magog the Ogre (talk) 16:07, 13 March 2011 (UTC)


 * Because I don't really understand the overall goal, this seems a bit silly. I assume you have wget and you can run it from the command prompt. In PHP, you would run: passthru("wget --page-requisites --span-hosts --convert-links -P \"%date%%time%\" \"http://en.wikipedia.org/wiki/Main_Page\""); Then, at the command prompt, run: php my_script.php (or whatever you call the script). So, you end up running something at the command prompt unless this is embedded in a web page. Also note that I used passthru because I think you want to see the output. Using exec doesn't show all the output. -- k a i n a w &trade; 16:38, 13 March 2011 (UTC)
 * One possible way would be to write your own C program, which calls  with various arguments.   takes a string argument, and then uses that to run an external program, optionally giving it command-line arguments. This way, if you have the C coding skills, you can use any logic imaginable to get the arguments. But if you don't know how to program in C, this might not be of any use to you. Also, bear in mind that you can't use input/output redirection with , as its arguments aren't parsed in a shell but instead given to the program directly.  J I P  &#124; Talk 18:24, 13 March 2011 (UTC)
 * Actually,  does invoke the shell. -- BenRG (talk) 22:32, 13 March 2011 (UTC)


 * If you can make cmd.exe do what you want then you don't need to switch, but
 * Sooner or later you'll find something that it can't do, because it doesn't have a general set of facilities for programming, just a few special-case hacks. You got lucky this time.
 * It gives you lousy feedback when something goes wrong—for example, the bug in my code above (%i instead of %%i) causes a mysterious failure instead of a helpful message. For another example, my code will silently do something stupid and wrong if datetime.exe doesn't return what you expect. In the long run this will cost you a lot of time, unless you never make mistakes.
 * Data gets repeatedly encoded as strings and then re-parsed, which causes security/reliability problems that you probably won't notice in testing. I seem to recall there was a shell script bug in an early version of Mac OS X that would wipe the user's entire hard drive on an upgrade if some folder happened to have a space in its name.
 * For the most part these problems apply to other shells, including bash. It's just not a good programming environment and I don't think it's a good idea for anything other than one-liners or software that absolutely has to run on a stock OS install.
 * Python (to pick a random example) has much better error messages, and it has libraries that can replace the functionality of most programs that you would run from the command line, and they usually have more options and make it easier to manipulate the output data into the format you want. I think the up-front cost of learning Python programming will probably pay off in the long term.
 * All that said, you may want to continue to invoke wget in this case, because it's a complicated program and I wouldn't trust there to be a library that duplicates its functionality exactly. You can run it from Python like this (for example):


 * Yes, the shell has a more succinct syntax for running programs, but I stand by what I said above. -- BenRG (talk) 22:32, 13 March 2011 (UTC)

What am I doing?
I'm using Firefox on OS X on a MacBook Pro. Every once in a while, I inadvertantly do something or another that moves a tab to a new window. I think it's a gesture, but I'm not sure! Any ideas? --jpgordon:==( o ) 21:17, 13 March 2011 (UTC)


 * I'm pretty sure you are dragging the tab down and dropping it on the page it's displaying. That closes the tab and opens it in a new window. 200.118.156.145 (talk) 23:00, 13 March 2011 (UTC)
 * INDEED! Thanks. Mystery solved. Instead of move-to-tab then click, I'm probably sometimes doing click-then-move. --jpgordon:==( o ) 23:11, 13 March 2011 (UTC)


 * You may want to increase your drag threshold in your mouse settings. I have to keep mine rather high because I regularly drag icons and tabs all over the place instead of clicking on them. By making it require an offset of about 30 pixels to register a drag event, it is harder to accidentally drag instead of click. -- k a i n a w &trade; 05:32, 14 March 2011 (UTC)


 * I must be dim about this; where is the drag threshold set in OSX? --jpgordon:==( o )
 * I don't think OS X has this option, at least that I've heard of. --Thekmc (talk) 20:06, 15 March 2011 (UTC)

Erasing data on old laptop
I'm donating an old laptop to charity and want to ensure that I leave no personal data on it, what is the best way to achieve this apart from the obvious of re installing the OS which I don't want to do. Mo ainm ~Talk  22:09, 13 March 2011 (UTC)
 * Maybe DBAN 82.43.92.41 (talk) 22:15, 13 March 2011 (UTC)
 * Offer to install and set up an operating system for the charity. Simply, reformat the hard drive and install whatever OS they prefer.  The installation process and running the system a few times will deter all but the determined snooper who has access to computer forensics tools.  Of couse, the truly paranoid will take a hammer to the hard disk after securely wiping it, but that leaves it useless to the next owner.  Astronaut (talk) 23:49, 13 March 2011 (UTC)
 * I don't wan't to re format the HD as I don't have copy of Windows to put onto it and the charity don't either and I also don't want to install Linux onto and then have the hassle of looking for drivers, so any other suggestions were I won't need to re install an OS onto it. Mo ainm  ~Talk  00:44, 14 March 2011 (UTC)
 * It's very difficult to ensure that there is no remaining private data without reinstalling the OS. You could try deleting any personal files you know about, then deleting your user account (this will require logging in to another user account that you don't normally use and that has administrative privileges), then running something like SDelete. However, your best bet would be to use Magical Jelly Bean Keyfinder to determine the Windows product key, find a Windows CD of the appropriate version, wipe the hard disk with DBAN, and reinstall Windows using the product key. It's the only way to be sure. -- BenRG (talk) 01:14, 14 March 2011 (UTC)
 * Reinstalling the OS is, as others have said, the only way to be sure but it also requires wiping the drive first. Reformatting doesn't, in many cases, overwrite data on the disc. Some NTFS types of formats do write zeroes, but I haven't ever tested/heard if that's a complete wipe. There are several commercial and free "wiping" programs that will help, but the least trouble, practical option would be to write zeroes over the entire disc (this is the standard linux/unix program to do this) and then reinstall the OS. Shadowjams (talk) 03:18, 14 March 2011 (UTC)
 * I've used Ben's suggestion of deleting accounts and files (from an admin account) in the past, then overwriting all disc space with a mixture of small and large dummy files (just by copying recursively). It doesn't guarantee that old files are irrecoverable, but it's a reasonable compromise.  SDelete will make a more thorough job of overwriting fragments.  It depends on how paranoid you are, and on how confidential the data was. It is unlikely that any subsequent user is going to use sophisticated data recovery tools, but you don't want them to be able to just "undelete" your personal files.    D b f i r s   07:56, 14 March 2011 (UTC)


 * Don't even think about trying to accomplish this by deleting file. You have to completely wipe the HD and reinstall the OS, and even that might not be enough.  The only way to REALLY be sure is to buy and install a new hard drive, and slag the old one. 75.57.242.120 (talk) 19:24, 14 March 2011 (UTC)
 * Wiping the hard disk is enough. There's not a shred of evidence that anyone in the world, even the NSA, has the ability to recover data from an overwritten hard drive. I'd rather avoid the spread of false rumors that encourage people to scrap perfectly good electronic equipment. -- BenRG (talk) 21:25, 14 March 2011 (UTC)
 * I second Ben's comment on this point. Being concerned about someone using an electron microscope to recover data is overkill. However, recovering deleted files, or disc areas that have not been overwritten is a trivial process by someone who has the software. That doesn't require sophisticated data recovery tools... only some simple open source software. Recovering after the actual sectors have been overwritten (we're talking about a normal HD; flash and solid state drives are different) is nearly impossible. Shadowjams (talk) 00:36, 15 March 2011 (UTC)
 * There are several issues at play. One is the possibility of actual data recovery from wiped disk sectors, which I agree is very difficult and unlikely unless you've somehow attracted the attention of (say) the FBI.  Another is that wiping a disk is not always as easy as it sounds.  Sure, you can run some "file wipe" utility that says "ok, wiped", but who knows what it really does?  I was amazed to find that the Windows "format" command (at least at the time) did a "quick format" where it just reinitialized some file-system tables in a few sectors of the disk, instead of wiping the whole drive.  Second, even if the software does the right thing and tells the disk to wipe certain sectors, who knows what the disk will actually do?  Disks do internal hard-drive defect management which means they will sometimes copy pieces of data in multiple physical locations for reliability, and the rewrite command may not get all of them.  This appears especially significant with solid state drives, that sometimes have "secure erase" commands that are implemented incorrectly.  The article I linked says overwriting the entire drive twice is usually enough. 75.57.242.120 (talk) 02:53, 16 March 2011 (UTC)


 * A long time later but I forgot to mention this at the time. It's not really surprising Windows format by default does a quick format. I'm actually surprised if anyone technically competent wouldn't realise. If you are using a large HD (say 250GB+) this is very obvious since a full format will take at least 30 minutes or longer whereas when you format with Windows it takes only a few minutes at most. It's also mentioned in numerous places it's quick format. You can do a full format, you just have to specify. Note that in modern contexts when people want to remove data, they usually call it erase or wipe. The primary reason for a full format is not to make sure you remove existing data (for that you should really zero fill the HDD with manufacturer tools or use Derik's boot and nuke or whatever) but to make sure there are no problems with the disc.
 * You're right that reassigned sectors are something that can't really be dealt with normally (although unless your hard drive has enough problems it is RMAable if under warranty the number is usually fairly small) but that's acknowledged in most discussions of erasing HDDs. There's also the Device configuration overlay and Host protected area although the number of people likely to have any info they don't want people recovering in them is small (and most of them should hopefully be aware they stored info there). Secure Erase is supposed to allow you overwrite reassigned sectors and also kill anything hidden by the DCO or HPA. but although implemented in the HDD firmware some BIOS prevent the command being sent (so malware can't do it) . Having said that I don't know how common this is, while I can't remember for sure, I may have carried out a secure erase once when a HDD suddenly developed a user password which after trying numerous things I decided wasn't removable without killing the data. (I can't remember exactly what I did but I know I had to remove the data and from some sources secure erase should kill any password.)
 * You're wrong about the FBI. As Shadowjams and BenRG have said, there is no evidence it's possible even for the NSA. For starters, if the FBI can do it, how come no known professional data recovery companies can, is there really insufficient commercial interest? Remember these companies can charge thousands of dollars for clean room recoveries, given the possibility of large lawsuits etc you would think there would be a market for at least one company to recover data from single pass overwritten HDDs if the FBI can. More importantly the FBI primarily operate in the public eye with the intention of bringing cases to court. One would presume the FBI would have to offer some testimony of what they did to recover data. Yet AFAIK no one has ever shown one court case where data was recovered from a parts of a drive overwritten with a single pass.
 * It's slightly plausible the NSA may be able to since they by nature tend to keep their methods a secret. (Although as someone pointed out recently, this also means they're more likely to be willing to beat you with the HDD until you tell them what was on it.) But even Peter Gutmann (computer scientist) who is sometimes considered one of the ones who started the paranoia doesn't seem to think it that likely with modern HDDs (see the updates at the end).  is also worth a read and suggests even in old HDs it wasn't possible (although the statement on marketing is misleading, AFAIK Guttman has never endorsed or received anything from any of the tolls using the technique he proposed and he later said they were nonsense anyway).
 * is widely quoted but when re-reading Guttman I was reminded of why I've avoided it. The author doesn't seem to understand a magnetic force microscope is not an electron microscope. The paper makes the same mistake. I don't think I can (and haven't tried) to understand the statistics which may be sound. I admit I always find it hard to get past when people make such basics mistakes, particularly since what they were using is very important and from looking at the paper I think they must have used an MFM since they seem to have measured the magnetic flux density. There are also others problems Guttman suggested, some which are beyond my level of understanding without further research but are apparently more fundamental. Then again Guttman claimed intelligence agencies can do it although as I've highlighted there's no evidence any have. There is also Brian Wahoff  who's company developed a mobile drive shredder (although that doesn't guarantee he's knows much about recovery) but also believes 1 pass is enough.
 * P.S. Just to be clearly I'm solely talking about magnetic HDDs. As for SSDs, no idea. They're still too expensive for me so not really of great interest.
 * P.P.S. There was the Great Zero Challenge which offered a prize if anyone could recover the data and also the reason I was searching for this only to find I never responded but as I later found out, the prize was only $40 along with some silly rules (e.g. need to recover the file names, only a professional company or government could dissamble the HDD which doesn't make much sense since I don't know of anyone who suggests it could be done without and disassembly understandbly voids the deposit so it's not there problem if someone random idiot disassembles then says oops sorry can't do it) meaning it was more just a bit of fun.
 * Nil Einne (talk) 17:02, 17 June 2011 (UTC)

If you have the additional storage space, you can probably copy the existing OS files, partition layout, and MBR from the disk (while the OS isn't running) to this additional storage space, then zero the drive (as others have said, once is enough [the non-"quick" NTFS format does this, I think(?)]), then copy the files back to the freshly formatted disk. Personally, assuming the charity has no techie of their own, I would just zero the drive and install Linux on it — I'd be quite surprised if you have any hassle looking for drivers, especially for older hardware; many distros auto-detect and install necessary drivers during the install process. If this organization is smart at all, they won't want to have your personal data just as much as you don't want them to. ¦ Reisio (talk) 01:15, 15 March 2011 (UTC)


 * I suppose you have to weigh up your own paranoia against the likelyhood of a charity having a computer criminal who is keen and motivated enough to scan old, donated equipment. If you really don't want to take extreme steps like reinstalling the OS (never mind replacing the disk), I suggest you delete all your files and delete all the temporary stuff like the browser cache.  Additionally, you can defragment the disk and add some large files to effectively overwrite the empty areas of the disk with randomness - large media files like music, photos, etc. might be suitable for this job (remember to delete the files again).  Astronaut (talk) 06:28, 15 March 2011 (UTC)


 * You can wipe the free space on your drive with various tools. Eraser for Windows is a highly rated freeware. ---— Gadget850 (Ed)  talk 11:51, 15 March 2011 (UTC)


 * I would use SDelete over Eraser because I see no evidence that the authors of Eraser understand what they're doing. SDelete's home page goes into detail about the subtleties of erasing data on NTFS volumes, but I see nothing like that at the Eraser web site. In the Eraser forums there's a recent thread from which I gather that (a) after years of development, Eraser still didn't support compressed files as of mid-2010; (b) Eraser's lead developer (Joel) doesn't know that decompressing a file moves it, so decompressing and shredding a file is equivalent to deleting it without shredding. In particular, he hasn't even read SDelete's home page, which mentions that. I think you are probably fine using Eraser to wipe free space, but I just generally lack confidence in the product. -- BenRG (talk) 21:01, 15 March 2011 (UTC)


 * I don't know the equivalent procedure on a Windows box, but on a Mac I would create a new administrator account, delete all other accounts, use a tool to delate all logs and caches, and then use a tool to securely erase free disk space. This will leave some semi-personal data deep down in the system area, and if whoever you give the computer to has a lot of free time on their hands they may go digging around and find it, but there shouldn't be anything down there that could be used for (say) identity theft.


 * You may be over-worrying, however. Most corporate charities automatically do clean installs on donated machines - partly for legal reasons (charities don't want to get caught using software they have no rights to) and partly for safety and security reasons (it wipes out any virus, trojans, spyware, and whatnot before the machine comes online behind the charity's firewall).  I suggest you do whatever reasonable job leaves you feeling comfortable, and don't worry about the rest. -- Ludwigs 2  01:12, 16 March 2011 (UTC)

I would recommend installing Piriform Recuva. Delete all the files you don't want anyone to recover, then scan the drive and check the check the checkboxes next to all the files/folders you just deleted. Right-select and click Secure overwrite checked. There are four overwrite options (which can be selected by clicking the Options button and going to the General tab); Note that you will need administrator privileges to perform these actions. Rocketshiporion ♫ 01:03, 18 March 2011 (UTC)
 * Simple Overwrite (Single-Pass)
 * DOD 5520.22-M (Three Passes)
 * NSA - Seven Passes
 * Gutmann - 35 Passes



Facebook and Mediaplex
In the last half hour, I have used my Firefox bookmark and the link from Google to try and log into Facebook. Instead, in each case, after a half dozen tries, I get Mediaplex at [http://www.mediaplex.com/}. Anybody else have either the same problem or an explanation? Bielle (talk) 22:21, 13 March 2011 (UTC)
 * My first thought was that it may have a virus or something. Try running Malwarebytes, then proceed to try it again. General Rommel (talk) 00:18, 14 March 2011 (UTC)
 * What is Malwarebytes and where do I find it? Thanks Bielle (talk) 00:40, 14 March 2011 (UTC)
 * Malwarebytes is a malware scan and removal tool that has a good enough reputation. It can be downloaded from here.  Astronaut (talk) 07:51, 14 March 2011 (UTC)
 * Malwarebytes found nothing, but the redirect remains. Any other ideas, or do I just have to trash my 3-month-old computer? Bielle (talk) 18:35, 14 March 2011 (UTC)
 * Maybe something has tampered with your DNS settings. Run ipconfig /all from the command prompt and look for the "DNS Servers" - two IP addresses should be listed.  Visit the website of the Regional Internet registry which covers your location (eg: RIPE for Europe, ARIN for North America, etc), and look for their Whois or database search tool.  Enter each IP address and check that it is registered to your ISP (or their parent company, or their parent company's parent).  As a further check, you can ask your ISP what the DNS settings should be.  Astronaut (talk) 06:07, 15 March 2011 (UTC)

Can cookies be blocked by name?
I use the current version of FF on WinXP. Cookies can be blocked by site, and (with varying degrees of success) by the session/permanent attribute. Can a script or some other addon block selected cookies by name, independent of which site is trying to set them?

My thought is, I'd like to block all the __utm* cookies; I could then set "allow for session" as a default for most others. I've looked briefly at GreaseMonkey, but am not sure if what I'm trying to do really is a scripting problem or not.

Any suggestions? DaHorsesMouth (talk) 22:54, 13 March 2011 (UTC)


 * If "FF" means Firefox, then click on Tools/Add-ons/Get Add-ons. You should be able to find many add-ons which can block cookies, some of which should be able to block them by name. I cannot be bothered to find out which ones can specifically do it, but I have BetterPrivacy, CookieCuller, and Ghostery installed. I also use Ccleaner which offers the choice of deleting named cookies. Not sure if SpywareBlaster can do the same. Edit: try https://addons.mozilla.org/en-US/firefox/search/?q=cookie 92.15.11.100 (talk) 11:40, 14 March 2011 (UTC)
 * You're probably better off blocking google analytics altogether, than still accessing it but merely droppig its cookies. I use AdBlock for that. 75.57.242.120 (talk) 04:32, 16 March 2011 (UTC)