Wikipedia:Reference desk/Computing

Welcome to the computing section
of the Wikipedia reference desk.

skip to bottom

Select a section:

• Computing
• Entertainment
• Humanities
• Language
• Mathematics
• Science
• Miscellaneous
• Archives

Want a faster answer?

Main page: Help searching Wikipedia

   

How can I get my question answered?

• Select the section of the desk that best fits the general topic of your question (see the navigation column to the right).
• Post your question to only one section, providing a short header that gives the topic of your question.
• Type '~~~~' (that is, four tilde characters) at the end – this signs and dates your contribution so we know who wrote what and when.
• Don't post personal contact information – it will be removed. Any answers will be provided here.
• Please be as specific as possible, and include all relevant context – the usefulness of answers may depend on the context.
• Note:
  • We don't answer (and may remove) questions that require medical diagnosis or legal advice.
  • We don't answer requests for opinions, predictions or debate.
  • We don't do your homework for you, though we'll help you past the stuck point.
  • We don't conduct original research or provide a free source of ideas, but we'll help you find information you need.

Ready? Ask a new question!

How do I answer a question?

Main page: Wikipedia:Reference desk/Guidelines

• The best answers address the question directly, and back up facts with wikilinks and links to sources. Do not edit others' comments and do not give any medical or legal advice.

See also:

• Teahouse
• Help desk
• Village pump
• Help manual

February 18

Adding memory

We have several computers around here, i5s and i7s. The manual on most of them say that they can take four 4GB DIMMs. One of them is a HP 8200 Elite SFF i5-2400. The manual says that it can take four 4GB DIMMs but the Crucial and Kingston websites say that it can take four 8GB DIMMs. I don't think that any of the others are like that. Will a computer designed for 4GB DIMMs generally support 8GB DIMMs? Bubba73 ^{You talkin' to me?} 00:29, 18 February 2016 (UTC)

I suppose it will be a waste of resources, since your computer won't be able to address the bigger RAM. I'd stick with the official manual, and not follow Crucial or Kingston advise. It won't toast your motherboard though. Maybe the computer just does not boot. --Scicurious (talk) 00:48, 18 February 2016 (UTC)

I will respectfully disagree with the above. It's unlikely the system won't be able to "address the memory space" (assuming you are using a 64bit operating system, which anything using over 4GB total should be anyway). Motherboard manuals TYPICALLY will list the 'supported' memory, this means the manufacturer has performed stress tests and compatibility tests with THOSE specific memory chips. It does NOT mean that other memory will "not" work. In fact, if when the motherboard came out, 4GB was the biggest RAM chip you could get, it's quite likely a motherboard manufacturer will not bother to go back to test an old board just because 8GB sticks are now available (it's not likely to sell them any more old motherboards, if in fact they are even still for sale). CONVERSELY when a new memory chip comes out, it IS in the interest of the memory manufacturer to make sure that their chip works with as many mother boards as possible (that WILL sell them more memory chips) so they quite probably WILL test older motherboards with their new chips. SO, if kingston or cruicial say their chips will work with your motherboard, I would very much take their word for it. Vespine (talk) 03:19, 18 February 2016 (UTC)

I will respectfully disagree with your respectful disagreement. RAM limit (Wikipedia:WHAAOE) depends on other factors too. There is a CPU addressing limit. There is no reason to have a CPU able to address more memory for no reason. The 64bit OS is indifferent here. More lines are simply not wired to the memory controller.Scicurious (talk) 16:28, 18 February 2016 (UTC)

I mostly agree with Vespine here. It's true there's no guarantee that a larger RAM size is going to work. Although this rarely has anything to do with lines not being wired to the memory controller, but instead what the memory controller supports, the CPU (seperate from any IMC), the actual DIMM design (for most forms of DDR sticks there can generally be multiple types of 8GB RAM sticks with different levels of compatibility) and sometimes also BIOS/EFI issues. Motherboard manuals are not always reliable on what the motherboard actually supports and in this case the memory controller is on the CPU which supports up to 32GB [1] although Intel hasn't really kept their sockets around for very long recently and I'm not sure if there was ever any real difference in IMC supported amount of RAM. (But it's only dual channel and AFAIK it's not really possible to support more than 2 sticks per channel on an LGA1155 platform which implies it must support some sort of 8GB sticks.) Anyway, if a major RAM seller is claiming that a major system does support their 8GB sticks, it likely does since most RAM sellers like these do provide some sort of money back guarantee so they tend to try and get that stuff right and it's likely someone has already tried it. If it's a fairly unknown system or a weird config, it's possible the info from the RAM seller could simply be wrong, but probably not here. Nil Einne (talk) 17:12, 18 February 2016 (UTC)

Mind you, while I have no idea what manual the OP is looking at, this manual for the same system explicitly mentions 8GB stick support [2] "The largest memory capacity possible is 32GB using four (4) 8GB DIMMs". It doesn't list 8GB sticks in any of the sample configs, probably because 8GB sticks suitable for a system like this (i.e. unbufferred) didn't really exist when it was written or if they did HP Compaq didn't sell them as standard components but it's obviously at least claiming theoretical support. Nil Einne (talk) 17:20, 18 February 2016 (UTC)

That is the document at which I was looking, but I missed that line. I was looking in the table on page 12 where it shows 16GB maximun. Bubba73 ^{You talkin' to me?} 20:48, 18 February 2016 (UTC)

Indeed, the whole issue is probably not a big deal. Crucial's forum already has a post about it: [3]. It is almost sure a case of a not updated manual (in OP's hand) vs. the updated DB of Crucial. That's also why Nil Einne found the figure of 8GB when he looked it up. — Preceding unsigned comment added by Llaanngg (talk • contribs) 18:44, 18 February 2016 (UTC)

How can I watch Hulu In Canada?

Please tell me. 208.181.190.136 (talk) 03:23, 18 February 2016 (UTC)

Use a VPN. --Scicurious (talk) 16:31, 18 February 2016 (UTC)

In other words, there is no non-technical way to do so. Hulu is not officially available in Canada. Mingmingla (talk) 01:53, 19 February 2016 (UTC)

Using a VPN is no longer a guarantee at all that a regional online streaming service will work, or will continue to work. Most such streaming services, including hulu and netfix, explicitly forbid using VPN services to bypass regional restrictions in their terms and conditions and DO enforce various levels of blocks and bans against users who they discover or detect using such services. Vespine (talk) 03:19, 19 February 2016 (UTC)

out of range

am trying to install Win 7 on a Pc its P4 but wen i press the boot frm cd room key, a message Resolution not supported comes out on another moniter a message Out of range pops up,wat do i do? — Preceding unsigned comment added by 197.79.9.249 (talk) 11:48, 18 February 2016 (UTC)

Sounds like the monitor doesn't support the screen resolution it uses for the install. If you have another monitor, you might try that. StuRat (talk) 18:27, 18 February 2016 (UTC)

I think so. Temporary use another monitor and change the screen resolution and refresh rate manually. --Hans Haase (有问题吗) 18:24, 22 February 2016 (UTC)

Macintosh LC

Why was the Macintosh LC a significant product of technology in 1990? Please write up to 3000 words explaining your answer and submit your paper by 03/03/16 to Mr Turnbull's office. --Finderoomertæs (talk) 17:39, 18 February 2016 (UTC)

Remember to provide references and sources. --Finderoomertæs (talk) 17:40, 18 February 2016 (UTC)

We can't tell if you're being facetious - but we will not do your homework. We will gladly volunteer our efforts to help you, though. Some of our reference desk researchers might even, you know, be good resources. Nimur (talk) 17:51, 18 February 2016 (UTC)

We have Macintosh LC, I think... --Stephan Schulz (talk) 22:44, 18 February 2016 (UTC)

I've finished and submitted the paper. The response I received from Turnbull's office was that they had no idea why I sent them a paper on the 1990's significance of the Macintosh LC but if I did it again Turnbull will ask Key to pull my pony tail. Not wanting that, I've deleted my paper. Nil Einne (talk) 18:05, 19 February 2016 (UTC)

I only read "I have a pony tail" - that's surprising new information! --Stephan Schulz (talk) 09:18, 20 February 2016 (UTC)

Our RD powers do enable us to identify the OP's Mr Turnbull - his office is number 3xxE. Not that we would reveal the value of "xx", of course. Tevildo (talk) 10:47, 20 February 2016 (UTC)

Why does the FBI need Apple's help?

Being the FBI in possession of that iPhone, couldn't they just clone it bit by bit, literally, and test passwords on each clone? Obviously there is something here hindering this, but what is it? How can a storage medium block being copied? --Llaanngg (talk) 20:22, 18 February 2016 (UTC)

The Register has some well-informed speculation. [4] ---- LongHairedFop (talk) 20:27, 18 February 2016 (UTC)

That might be a good overview, and there's another one above in this same page. However, the question is more about storage hardware that does not let being copied 1-to-1. I wonder how this is possible. Couldn't they just disassemble the device and read the storage with another controller? --Llaanngg (talk) 20:32, 18 February 2016 (UTC)

iPhones use whole-disk encryption, which is protected by the user's passcode. Even if the flashdrive is copied to another iPhone, it is useless without the passcode. LongHairedFop (talk) 21:53, 18 February 2016 (UTC)

If you can copy the flashdrive to other iPhones, you could test more passwords on them (which are 4-6 numbers long). But there is obviously something that makes this impossible. --Llaanngg (talk) 21:56, 18 February 2016 (UTC)

In short, there is some piece of hardware that is designed to be difficult to clone, which holds a secret, and all other data is encrypted with a key derived from that secret. In most computers that hardware would be the TPM. In the iPhone 5c it's (a part of) the Apple A6 SoC. The A6 likely authenticates the first-stage boot code (which authenticates the next stage and so on), meaning there's no way to run unsigned password-forcing code. Even if it doesn't, it would probably be risky to inject a modified first-stage boot loader into the irreplaceable phone. -- BenRG (talk) 23:04, 18 February 2016 (UTC)

Llaanngg, the passcodes on iPhones can be longer than 4-6 characters. Even if the limit were 4 numbers, that's still 10,000 different passcodes that it could be. That's either a lot of iPhones that the FBI would have to buy (assuming that copying everything over to a different phone is even possible) or a smaller number with a lot of erasing of failed phones and reloading of the data. Dismas|^(talk) 00:00, 19 February 2016 (UTC)

The newer models have longer passwords, but in this case it is a iPhone 6, which has passwords which are 4-6 numerical characters long.

If it were possible to erase and reload the information to a new phone, the question would be how long would it take to accomplish this reset. You can try 10 passwords until the phone deletes the information. So, in worst case 1,000 load, erase and rinse cycles. But apparently some piece of hardware was made difficult to be cloned. Llaanngg (talk) 00:12, 19 February 2016 (UTC)

I wonder whether it might be worth saying that "reading" a memory chip is unlike reading a hard disk. You can't just point a read head at it and extract the bit values one by one. In silicon memory, you have to rely on the (usually) integrated memory controller to read and write values in and out of the memory. The memory controller has its own firmware which can be as "open" or "locked" as the designer wants. Clearly in the case of the iPhone, it's quite "locked" down. The difference is that Apple could us a special digital signature that only they have to load a special firmware during boot up before the phone normally lets anyone run any code. Vespine (talk) 00:19, 19 February 2016 (UTC)

Yes, this is surely worth (at least for me). Llaanngg (talk) 00:22, 19 February 2016 (UTC)

Just a nit but it's actually a iPhone 5C as pointed out in the first link in this section. But it could still have the latest iOS on it. Dismas|^(talk) 00:26, 19 February 2016 (UTC)

Yes, you are right. Anyway, it was not the last model. Llaanngg (talk) 00:34, 19 February 2016 (UTC)

To expand on what I believe BenRG was saying, each phone has a hardware module that stores a unique key. The disk is then encrypted using a combination of that unique key and the user's password. It shouldn't be hard to copy the physical contents of the iPhone storage, but trying to decrypt those contents without the corresponding hardware module is much harder than trying all possible 4 or 6 digit passcodes because you also have to consider all possible versions of the unknown hardware key. I don't know how long the unique hardware key on an iPhone is but the computer equivalent TPM has a built-in 2048-bit (256 bytes) hardware key. Trying to decrypt the storage without the hardware key would be far harder (virtually impossible) so one pretty much has to use the same phone to decrypt its contents. Dragons flight (talk) 18:15, 19 February 2016 (UTC)

February 19

Installing utorrent on linux

How do I install utorrent on Linux? — Preceding unsigned comment added by Me has a question (talk • contribs) 23:21, 19 February 2016 (UTC)

I have some good advice: don't use uTorrent. Get another torrent client. Here is info on how to install the torrent client Deluge on Linux (specifically Debian or Ubuntu distro's). Deluge is a free, open source software. The Quixotic Potato (talk) 00:20, 20 February 2016 (UTC)

Which Linux distribution are you using? The package manager will let you install additional applications; there should be several torrent clients available. 62.56.67.192 (talk) 12:06, 20 February 2016 (UTC)

Some Linux distributions include Transmission (BitTorrent client). Use instead, if possible. Else get μTorrent from its webpage. As its lincense includes Adware, it will not be listed in repositories. You need to know the consequences of installing it on Your computer. --Hans Haase (有问题吗) 18:23, 22 February 2016 (UTC)

February 20

/\ in proof of Ardens Theorem means?

Could anyone tell what does the symbol '/\' mean in the proof of Arden's Theorm given here(P.18).

NB:'/\' doesn't mean the regular expression written as '^'.Here '/\' is an element of R as given in this book(P.127).

JUSTIN JOHNS (talk) 09:25, 20 February 2016 (UTC)

It stands for the empty word, more often written as ε. --Stephan Schulz (talk) 13:08, 20 February 2016 (UTC)

Yeah thanks for that.JUSTIN JOHNS (talk) 06:32, 22 February 2016 (UTC)

Windows 10 on a slower PC

I have one computer running Win7 on an AMD dual-core 64-bit processor rated at 1.6 GHz, 4 GB of RAM. It's used only for email and basic browsing, and its speed is on the slow side of borderline. Generally speaking, would you expect Win10 to be a little faster, roughly the same, or a little slower? If the latter, I'll leave it on 7 for the remainder of its life.
The computer clearly meets the minimum system requirements for running 10, but "run" is not the same as "run well", and I'm looking for "run well". It also meets the requirements for Win7, but as I said I'm not entirely happy with the performance. I don't want to make it worse. ―Mandruss ☎ 15:43, 20 February 2016 (UTC)

My experience with several old PCs (Intel 3.X Mhz dual-core, 8-16GB RAM, 64-bit Win7 vs. 64-bit Win10) is that Win10 was just a bit faster, but they were close. This review found the same thing. I also found that, on both Win7 and Win10, going from 4GB to to 8GB made a big difference, but going to 16GB didn't do much. Also, replacing the rotating disk with a SSD made a big difference.

You can get Dell Optiplex 780, Dual core@3.33GHz, 8GB RAM, 500GB HDD, Win7 Pro 64-Bit for well under $200.[5] You have until 29 July 2016 to load and install the free Win7 to Win10 upgrade. --Guy Macon (talk) 18:04, 20 February 2016 (UTC)

If you want to make the PC faster, it is probably more efficient to clean it up a bit (and maybe even buy some RAM, old RAM is very cheap if you go to a second-hand store). Autoruns and Defraggler may be useful. The Quixotic Potato (talk) 01:28, 21 February 2016 (UTC)

February 21

Name generators

Are there any reliable secondary sources out there that discuss name generators (such as http://FakeNameGenerator.com) in any kind of detail? Thanks. —67.14.236.50 (talk) 07:59, 21 February 2016 (UTC)

Markov text generators might get you started. -- Finlay McWalterᚠTalk 13:08, 21 February 2016 (UTC)

What naming scheme(s) does Wikipedia use? Is it structured, flat (unstructured), or, attribute based?

I'm learning distributed systems and one of the topic is naming. So I'm wondering what naming scheme(s) does Wikipedia use? Is it structured, flat (unstructured), or, attribute based? Do other web services also use the same naming scheme? I can't seem to find information on structured, flat, or, attribute-based naming scheme on Wikipedia. Could you please point me into the direction of finding the information? Thank you so much. — Preceding unsigned comment added by 116.87.54.73 (talk) 09:44, 21 February 2016 (UTC)

There is no "scheme". The relevant policy is at WP:TITLE. -- Roger (Dodger67) (talk) 11:00, 21 February 2016 (UTC)

@Dodger67: I think he’s asking about technical behind-the-scenes details, not about how we choose the human-friendly article titles. I’m not sure if the question is even applicable to Wikipedia (I don’t know much about distributed systems), but I know our revisions are identified by a simple incrementing number, currently in the hundreds of millions. So, I guess that would make it flat. But for naming the articles (for public use; I don’t know if they have a different identifier behind the scenes), we have namespaces like “WIkipedia:”, “Help:”, “Template:”, and a “talk” namespace for each of them. We have e.g. an article named Wikipedia and a project page named Wikipedia:Wikipedia. I’m not sure if that makes it structured. Hope that helps, though I don’t know if I gave you any new information. —67.14.236.50 (talk) 03:50, 22 February 2016 (UTC)

In the early days, Wikipedia articles had names like "Namibia/Economy" and "Namibia/History" and the software understood that those were child articles of "Namibia" (which is structured naming), but now those articles are called "Economy of Namibia" and "History of Namibia" and there is no well defined hierarchy (which is flat naming). If you include pages in other namespaces like Talk: and Wikipedia:, the hierarchical names are still used (on this page, for example), and the namespace can also be seen as the top level of the hierarchy.

I've never heard the terms "flat naming" and "structured naming" outside of university courses, but that may be because I don't work in distributed systems. Wikipedia should probably cover them regardless. Someone just needs to write the article. -- BenRG (talk) 03:45, 22 February 2016 (UTC)

Now the names to refer to article sections would be "Namibia#Economy" and "Namibia#History". StuRat (talk) 04:01, 22 February 2016 (UTC)

Wait a minute. What is being named in distributed systems? If the question pertains to the files on the project’s web servers or the way they’re networked, then none of us are qualified to answer it unless we’re familiar with Wikimedia’s backend, which has nothing to do with how articles are named. —67.14.236.50 (talk) 04:03, 22 February 2016 (UTC)

Thank you all for your replies. I'm still trying to understand how naming works in distributed systems, so forgive me if my questions are not appropriate. I guess my next question would be why Wikipedia uses colon (:) in its URL, example would be https://en.wikipedia.org/wiki/Wikipedia:Reference_desk. I understand this is to divide into different categories but then why not just use slash (/) instead like https://en.wikipedia.org/wiki/Wikipedia/Reference_desk or https://en.wikipedia.org/wiki/Wikipedia/Talk_page_guidelines. Thanks. — Preceding unsigned comment added by 203.217.187.24 (talk) 12:30, 23 February 2016 (UTC)

Colon is for Wikipedia:Namespace. Many of the namespaces have special software features associated with them. Slash is for Wikipedia:Subpages. PrimeHunter (talk) 12:43, 23 February 2016 (UTC)

.MACOSX directory in zip file

I got some zipped lectures, and there were a __MACOSX directory with files like ._.DS_Store, and ._<name of the lecture>.mp4 inside. What is this?--Llaanngg (talk) 13:57, 21 February 2016 (UTC)

It means the zip file was made on a Macintosh computer. You can safely ignore them. See superuser.com and Resource fork and .DS_Store for more info. The Quixotic Potato (talk) 14:06, 21 February 2016 (UTC)

Regexp's literal strings

Following regexp works for me:

 rename 's/(Chapt).(.*)/$2.$1er/' *

but in emacs, when I try to extract a pattern, and include the found group () into the target string, I have to do:

  \(Chapt\).\(.*$\) -> \2.\1er

and not:

  (Chapt).(.*$) -> \2.\1er

Why is it different? And why isn't \(\) in emacs reserved for the case where you actually want to reuse the found pattern? --Scicurious (talk) 23:10, 21 February 2016 (UTC)

Emacs uses an older syntax for regular expressions (a combination of the POSIX basic and extended syntaxes) than Perl (the modern standard). The biggest change (for the common subset of functionality) is that you must escape (){} exactly when you must not for Perl. No regular expression system uses \(\) to turn on (or off) capturing: you turn off capturing (where it's supported) with (?:…). In Emacs, you escape those parentheses with backslashes (just like for a capturing group). --Tardis (talk) 01:09, 22 February 2016 (UTC)

Calling Perl regexes "the modern standard" is a bit misleading. For one, they aren't standardized. Neither is PCRE, which means things that claim to support "PCREs" often have subtle incompatibilities with each other. It's a bit of a mess (insert xkcd comic about standards here). Yes, most of the time if you stick to the simpler PCRE features things will be portable, but don't assume it. Look at the documentation for whatever you're using. --71.119.131.184 (talk) 05:17, 22 February 2016 (UTC)

(EC)What you have encountered are some of the differences between the many flavors of regular expressions. As regular expression implementations have added features over the years, there have also been corresponding variations in the supported syntax, some of which are not compatible with one another. One early distinction was between the UNIX Basic Regular Expression (BRE) and Extended Regular Expression (ERE) syntax. (See Regular expression - POSIX basic and extended.) Emacs implements BREs (or some variant thereof), where only a few characters have special meaning when not escaped.

For more information and a comparison of syntax, I suggest you look at [6]. Set the drop-down flavor selectors to "POSIX BRE" "POSIX ERE" as a start, or select any other flavors that my be of interest. Syntax feature categories (Characters, basic features, quantifiers, ...) can be selected at the left. -- Tom N talk/contrib 01:49, 22 February 2016 (UTC)

This page is also a good reference on regex syntax. Also note that "rename" is not a standardized Unix command, although a lot of people think it is. There are multiple programs called "rename" floating around out there, with incompatible syntax. See here for some more portable ways to rename files. --71.119.131.184 (talk) 05:17, 22 February 2016 (UTC)

February 22

Thomas edison

hello i need a reply as soon as possible for this, it is for a school assignmetn and iv beensearching for 1 hour now for the answer so i decided to make a post the answer i need help with is (Famous inventor) Thomas Edison. How is this persons name now used to describe or measure something in electronics.? — Preceding unsigned comment added by Buck Dunford (talk • contribs) 03:21, 22 February 2016 (UTC)

Many light bulbs have an Edison base (that's the wide screw threads at the bottom). Note that while our article features incandescent light bulb pics, many newer CFL and LED lights continue to use the same base. StuRat (talk) 03:28, 22 February 2016 (UTC)

See Thermionic emission also. Robert McClenon (talk) 03:46, 22 February 2016 (UTC)

Do you refer to the Edison effect ? StuRat (talk) 04:02, 22 February 2016 (UTC)

Programmable game controller

Years ago I had a Microsoft SideWinder Strategic Commander game controller. Does anyone know if there's anything similar on the market nowadays? I can't find anything obvious out there, everything seems to be in programmable keyboards nowadays. I need ot to be Windows 7/10 compatible. -mattbuck (Talk) 15:12, 22 February 2016 (UTC)

The Steam controller is very configurable (in reality the configuration is its driver, but that was probably the case for the sidewinder too), where one can map different buttons to different actions and map the "joysticks", touchpad, triggers etc. to different actions as well. An example config is here. I don't know, however, if it's possible to configure one button to generate a sequence of keystrokes. I should say I've personally not used a Steam controller, but every review I've seen of them has been positive. -- Finlay McWalterᚠTalk 15:54, 22 February 2016 (UTC)

On looking further, it seems the Steam Controller does not have macro (keysequence) support), but that people have had success using a custom key on the controller to trigger a macro defined in AutoHotkey. -- Finlay McWalterᚠTalk 16:02, 22 February 2016 (UTC)

Transition from http to https

So I've just switched my personal websites over from http to https using the "Let's Encrypt" free service. They seem to be working just fine. I'm wondering though about URL's I have embedded in HTML, JavaScript and PHP code... Is it best-practice these days to request https from other sites that I link to? All of them? Do I have to check each link manually to see which support it and which don't? I know that some browsers rewrite URLs that the user types in - but what about those embedded in other places? SteveBaker (talk) 15:55, 22 February 2016 (UTC)

This may have nothing to do with your question, but this is what your question made me think about... I just did a test in Google Chrome and in Firefox. I made a page that does an Ajax request to a page using http instead of https. I accessed the main page using https and let the ajax run. On the server, the ajax request is logged in access_log, not in ssl_access_log. So, I received it via http, not https (as I stated in the JavaScript). However, neither browser threw a warning. I hoped it would be like when you have a secure page that includes an image using http. You get a warning that insecure content is being included in the secure page. Therefore, the clients are not being told that unsecured content is being made via Ajax in their secure page. It is up to the developer to ensure that if "http" is included in the ajax, they must use "https". Better, don't use an absolute URL. A relative URL will maintain the https part of the page. 209.149.114.211 (talk) 16:27, 22 February 2016 (UTC)

Yeah - I already use relative URL's where I can - but it's not always possible to keep everything on the same server. But suppose I say something in HTML like "You can find out more about this on <A HREF="http://wikipedia.org">Wikipedia<\A>" ? Should that be 'https'? (Wikipedia supports https - but what about a site like slashdot that (currently) does not?) SteveBaker (talk) 16:51, 22 February 2016 (UTC)

A relative URL without a scheme (http: or https:) is valid, per RFC 3986: "Uniform Resource Identifier (URI): Generic Syntax", Section 4.2. It looks like this: <img src="//domain.com/img/logo.png"/> When making external links, try HTTPS first, if that does not work then link to the HTTP version. Using HTTPS Everwhere may be useful. The Quixotic Potato (talk) 17:30, 22 February 2016 (UTC)

I don't see the benefit of connecting to Wikipedia or Slashdot using HTTPS. Yes - it encrypts your traffic. No - it does not hide that you are going to Wikipedia or Slashdot. Wikipedia automatically redirects to HTTPS if you sign in. So, if you link to it using HTTP, it automatically switches to HTTPS. The same with Facebook. So, there is no point to use HTTPS when HTTP will automatically redirect as necessary. It won't throw a warning on the client because the link isn't being included in the source.

On a side note... I avoid using HTTPS at one place where I work. They have idiots in charge of IT. Instead of security, they just purchase and run a deep-packet inspection tool that automatically decrypts all HTTPS traffic. The client gets a fake certificate and has to accept it to continue. I don't want to fill my browser with fake certificates, so I use HTTP. 209.149.114.211 (talk) 18:51, 22 February 2016 (UTC)

I know there are some issues for some people with https - I'm not shutting off http access. But there is a strong trend for browsers to search for https sites first - so there is an incentive to support both.

The idea of https is not so much that the content arrives encrypted - or that you are somehow hidden from tracking - but that it guarantees against man-in-the-middle attacks. Your IT dept is (in effect) performing a man-in-the-middle attack on you - so it's good that you're being warned! SteveBaker (talk) 20:27, 22 February 2016 (UTC)

Listing duplicated filenames

On Fedora 20 Linux, I have a directory containing subdirectories and sub-subdirectories. These sub-subdirectories contain files whose names follow a regular pattern. Some of the names might be duplicated in separate directories. The directory structure is rigid, i.e. the files always reside at the same depth under the main directory. There are about a quarter million such files.

How do I get a listing of all the filenames that appear in more than one sub-subdirectory, and the directories they appear in? JIP | Talk 16:11, 22 February 2016 (UTC)

Well, if you do an "ls -R1" that'll get you a recursive list of all of the files - without the preceding pathnames - one per line. If you don't want to look for duplicated sub-directories, then you could pipe that through grep -v '\\' to get rid of the directory names. Pipe the results of that through 'sort' to get the identical names in consecutive order and then save the results into a file. Next, push the resulting file through 'uniq' to reduce it to a list without duplicates and save that to a second file. Then you can 'diff' the sort'ed list and the sort/uniq'ed list files to get a list of all of the duplicated files - which you'll want to push through 'uniq' again so you only have one copy of each filename.

That process gets you a list of duplicated filenames, one per line, with no directory names.

OK - so now you need to know which directories those are in. The 'find' tool is a natural for doing that - so you'll want to use something like 'sed' to take every line of the file we just made and turn it from whatever into:

find . -name whatever -print

...so you now have a script with a 'find' command for each duplicated file. Then you can run that as a script to get a complete list of every duplicated filename and all of the paths that it occurs in.

If you might have spaces or weird characters in your file names - or if there are hidden files, devices, links, symlinks and other monstrosities that you need to be careful of...you'll need to think this through more carefully - but that's an outline of a shell-based means to do it.

Like everything you can do in shell - there are probably a dozen different ways to do it. SteveBaker (talk) 16:41, 22 February 2016 (UTC)

(ec)I don't know of a single command that does this. Just a train of thought... You can use ls -LR . to get a list of all files under the current directory (change . to the directory you are interested in). You can sort that list and then use uniq -c to get a list of unique names with a count. You can use grep -v "^\s*1\s" to omit any entry that has a count of 1. Then, you can use sed "s/^\s*[0-9]*\s*//" to remove the count from those that remain. Then, you can loop over the result list with locate (assuming you've run updatedb recently). This will create a list of duplicated filenames with full paths to each one: ls -LR . | sort | uniq -c | grep -v "^\s*1\s" | sed "s/^\s*[0-9]*\s*//" | while read x; do locate -bP '\\'$x; done

That is not "easy". However, it is what came to me as I went from "I have a list of files" to "I want a list of the locations of each duplicate file." I'm sure someone else will chime in with a command that does exactly what you want without all the extra work. 209.149.114.211 (talk) 16:48, 22 February 2016 (UTC)

I think we are talking about essentially the same approach here.

One thing you CAN'T do is to use wildcards to find files - with millions of files, you'll overflow the maximum length of the command line...and in any case, it would be insanely slow. SteveBaker (talk) 16:55, 22 February 2016 (UTC)

→List of file comparison tools --Hans Haase (有问题吗) 18:14, 22 February 2016 (UTC)

I believe that most (if not all) of those tools are for comparing the contents of files - not the names of files in different directories - which is what is being asked here. It's possible that one or two of the tools might do what our OP requests - but you're not going to discover that from List of file comparison tools. SteveBaker (talk) 20:22, 22 February 2016 (UTC)

I would do something like this

find . -type f | perl -e 'while(<>){chomp; $p=$_; ($f)=/.*\/(.*)/; push @{$d{$f}}, $p; } foreach $f (keys %d){ print(join(" ", @{$d{$f}}, "\n")) if @{$d{$f}} > 1 }'

If your filenames can contain spaces you may want to replace the separator in the join statement with some other character, just for readability. Mnudelman (talk) 20:07, 22 February 2016 (UTC)

I'm a little unclear on your question. Do you actually care about which directories the files are in? In other words if you have:

top1               top2
 |    \             |
sub1   sub2        sub1
 |  \    |          |
foo bar bar        foo

...do you want both "foo" and "bar" printed, or only "foo"? If it's the first, you're just asking for all files that have the same name, which is trivial. --71.119.131.184 (talk) 21:32, 22 February 2016 (UTC)

Yes, I want both "foo" and "bar" printed. JIP | Talk 05:46, 23 February 2016 (UTC)

If you want to do it all in shell scripting, you could do it by first using sort | uniq -d to pick out the repeated filenames, then find to report the instances of each one:

  find . -type f | sed 's:.*/::' | sort | uniq -d |
    while IFS="" read basename
    do
       echo "Repeated name $basename:"
       find . -type f -name "$basename"
       echo
   done

The IFS="" bit may be obscure: it's necessary in case any of the filenames may start or end with whitespace, because read was not originally intended for this sort of use and strips leading and trailing whitespace from the line it reads.

Of course this appraoch is potentially inefficient since it will use find repeatedly, but that may not be important. --69.159.9.222 (talk) 06:20, 23 February 2016 (UTC)

That script will still fail on filenames containing *?[ (special to -name) or \ (special to read) or newlines. It's O(n²) in the worst case, as you pointed out, which could be a problem when n ≈ 250000. sed, sort and uniq use the current locale, which could cause problems.

I don't understand why people go to all the effort of trying to get bash to do anything reliably. There are much better programming languages out there. Here's a solution in Python:

    import collections, os
    dirs_by_name = collections.defaultdict(list)
    for dir, subdirs, files in os.walk('.'):
        for file in files:
            dirs_by_name[file].append(dir)
    for file, dirs in dirs_by_name.items():
        if len(dirs) > 1:
            print('%r is in %r' % (file, dirs))

-- BenRG (talk) 09:36, 23 February 2016 (UTC)

Google date

Please can you tell me how to find out when the "Google Street View" photographic car will be visiting a certain area? I do NOT wish to be photographed so I want to stay indoors in the day it is due in my area. Thank you. — Preceding unsigned comment added by Haridtanton (talk • contribs) 20:27, 22 February 2016 (UTC)

They give rather general info on this page. It is important to note that Google doesn't completely control every camera used to get images for Google Maps. From that page, you will see links to information about how the images are obtained. You should also note that Google has put a lot of effort into blurring out the faces of people or, often, completely removing people from images. The next goal is to completely remove cars from the images. The goal is to remove temporary obstacles to make the images better. The goal is not to snap of photo of you standing on your front porch in you bathrobe. 209.149.114.211 (talk) 20:43, 22 February 2016 (UTC)

Add to this that you can ask Google to remove pictures from their service where you or your house are depicted. --Scicurious (talk) 22:28, 22 February 2016 (UTC)

See How To Remove Your House From Google Street View The Quixotic Potato (talk) 01:03, 23 February 2016 (UTC)

Windows 7 computer crashed

A week ago my Dell rack mount R5400 crashed. The OS is Windows 7. Every user worst nightmare. I do have an external Toshiba drive with ample memory for backups and I have created system repair disks. The last time I created a repair disk was last October. Last night a friend of mine, a very competent professional software developer and I, decided to try to restore the system. We went through some motions: changed the boot from the system drive to the onboard CD, etc. It seemed everything worked until it began "restoring" and all of a sudden a message appeared "No disk that can be used for recovering the system disk can be found." That was an affront. We saw a list of dates when the system image was created, last time on Feb 15th, just a week ago. There were others, earlier dates as well. What is the problem? Googling showed that it is a common situation facing folks who try to restore their systems. My rackmount computer is rather old although functionally efficient. I doubt I have the original windows restoration disk somewhere and also after so many updates over the years that will be useless. What options do I have?

One web site suggested to contact Microsoft for a bootable disk. But they are now promoting Win10 nonstop, I doubt I will get a lot of help. I will appreciate any suggestions.

Thanks, - Alex --AboutFace 22 (talk) 22:23, 22 February 2016 (UTC)

The Dell™ Latitude™ E5400 is a laptop, do you mean the Dell™ Precision R5400? There are some suggestions on various sites saying that this problem occurs when the target partition is smaller than the original source partition. The target DISK size must be at least as large as the source DISK, regardless of the size of the PARTITION you're trying to restore. Using Diskpart also seems to be useful. See here. I would ask Dell, not Microsoft. Try this. My experience with Microsoft support is that they are slow. You may be able to boot from USB if you enable legacy usb emulation in the BIOS. Does the computer still have the restore partition? If so press F8 at boot and select Repair The Computer. I also found this and this and that. The Quixotic Potato (talk) 00:47, 23 February 2016 (UTC)

February 23

Pausing applications

Does anyone know of a way to pause applications on a PC (Win10)? There was a program called uPause a few years ago, but I don't know if it went anywhere. — kwami (talk) 05:01, 23 February 2016 (UTC)

Well, there is the old keyboard "Pause/Break" key, on the upper right corner of my keyboard. That's mostly an obsolete key left over from mainframe days, but I've occasionally been surprised to find some modern application that obeys it (along with the Scroll Lock key right next to it). StuRat (talk) 05:29, 23 February 2016 (UTC)

You can suspend a process with, for example, Process Hacker (right-click on the process and choose "Suspend"). This will prevent it from using the CPU until you resume it. pssuspend does the same thing from the command line. -- BenRG (talk) 07:48, 23 February 2016 (UTC)

A question about RAM

When computers got 1GB of RAM (Random Access Memory) for the first time? 192.114.23.208 (talk) 09:34, 23 February 2016 (UTC)

Time computers often claim to have the first home computer to hit 1GB RAM. The problem is that they put out the hardware with an OS that couldn't access 1GB of RAM. Super computers obviously hit that much earlier, but they do not have the same design as average computers, so they tend to measure memory in words, such as "1 million words of memory." You can translate that to bytes (e.g. if a word is 32 bytes, 1 million words is 32MB). Because memory is not used in the same way, that comparison doesn't really make a lot of sense. 209.149.114.211 (talk) 13:34, 23 February 2016 (UTC)

Computer memory part

How many part of computer memory and how it work adn what is the imortant each of one? By Aanaya, — Preceding unsigned comment added by Scalemodel (talk • contribs) 09:53, 23 February 2016 (UTC)

See Random Access Memory. The Quixotic Potato (talk) 10:11, 23 February 2016 (UTC)