Wikipedia:Reference desk/Archives/Computing/2013 December 18

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.

December 18

Why do people still say "w w w dot" when naming websites aloud?

I just saw a commercial for Lumosity, and in the commercial they say something like "just go to w w w dot ..." I haven't typed www. anything in years (does anyone?) You just type imdb.com, or google.com or whatever. Why do people still say it and is there any purpose in actually typing it, like some millisecond faster connection (that would not make up for the time saved in not typing it in the first place)?--108.46.96.174 (talk) 06:11, 18 December 2013 (UTC)

Your IP address geolocates to the USA, which explains a bit. Sites in other countries have country codes on the end (e.g. .uk, or .au), which must be typed, or an American site may come up. You must also only ever go to sites with ".com" on the end, or use other means to get to this site. If you look at the url for this page, you'll notice it's got a ".org" there. Typing the full url still has its place at times for some of us. As for why American advertisers do it, maybe they're not all that computer literate. HiLo48 (talk) 07:31, 18 December 2013 (UTC)

The question is about the www. prefix, which is equally useless everywhere in the world. -- BenRG (talk) 07:50, 18 December 2013 (UTC)

Well to access the uk version of yahoo it's uk.yahoo.com not www.Dja1979 (talk) 10:55, 18 December 2013 (UTC)

There are plenty of less-popular web sites that only support www.foo.bar (or even only foo.bar). It isn't hard to support both but it isn't automatic either. If foo.bar redirects to www.foo.bar then the latter will be faster (and vice versa). I think that advertisers like Lumosity worry that some people would not know how to "go to lumosity.com" without the www to make it clear that it's a web site, but that's pure speculation on my part -- BenRG (talk) 07:50, 18 December 2013 (UTC)

Human idiocy is unquestionable. About 10 people a day attempt to call our fax number at work every day, despite our business cards and catalogues stating FAX:xxxxxxxx and Telephone:xxxxxxxxx very clearly. I'm sure it's the same people who use email addresses as website addresses. Thanks ツ Jenova20 ^(email) 10:28, 18 December 2013 (UTC)

"Does anyone?" Well I do, quite often. While Ben points out that www.foo.bar is a different address from foo.bar, not every where automatically redirects from one to the other, therefore I often put the www. in even if it is strictly not needed. Astronaut (talk) 11:13, 18 December 2013 (UTC)

I'm sniggering at Austronaut and his www.-triggered carpal tunnel syndrome.--108.46.96.174 (talk) 14:00, 18 December 2013 (UTC)

I work at a college here in the UK, and we have to type www,www1,www2 and www3 depending on where were want to get to on our website servers. We also have cases were www isn't required, or is replaced with something else, eg vdi. I don't know why but that's the way it is and we just accept it (and yes it can be confusing). --TrogWoolley (talk) 16:22, 18 December 2013 (UTC)

I can go one better; not only do I type in the www, I also usually type in the http://. I'm well aware it's not necessary, but I picked the habit up in the old days when you had to specify between HTTP and FTP on the browser (I'm assuming Netscape Navigator, but possibly other early ones as well). I find it useful to annoy my daughter as well, so the practice is not without merit. Matt Deres (talk) 17:17, 18 December 2013 (UTC)

• One use of saying "w w w" out loud is to signal to the listener that a URL is coming. In some oral contexts that can be very useful. It may well be that such a signal is found conducive to an audience which also listens to an ad, besides watching it. For a radio ad it's very useful, I imagine, even if redundant. Drmies (talk) 19:23, 18 December 2013 (UTC)

The Ali G multiplication problem?

What is meant by it?

Input: two d digit numbers (mostly 9s)
Output: the product of all the numbers

Is it something you would find in the serious literature? OsmanRF34 (talk) 13:32, 18 December 2013 (UTC)

It seems to be meaningless babble. 'd' is undefined and "mostly" doesn't help much. The phrase "product of all the numbers" is odd - you'd expect it to say "product of these two numbers" or something...which implies to me that you don't have the question right. As given, it seems like there are an insane number of answers for any given 'd'. What is the context of the question? Maybe there is more information about it there that would help us to get you an answer. Certainly there is nothing in serious math literature that would be written this badly. SteveBaker (talk) 14:30, 18 December 2013 (UTC)

It was from a ppt presentation, in the context of computability. The source is [[1]]. It is from the CS department of the Univ. of Virginia. See [[2]] too for the "10012 Things Every Self-Respecting Computer Scientist Should Know." OsmanRF34 (talk) 14:38, 18 December 2013 (UTC)

The problem comes from this episode of Ali G. The point made in the presentation (and by Ali G), as I understand it, is that we'll never be able to build a computer that can solve any multiplication problem, simply because the input factors can be arbitrarily long (d has no bounds). --Wrongfilter (talk) 14:57, 18 December 2013 (UTC)

One of the Powerpoint slides shows an Excel spreadsheet in which the product of a column containing 999999999 and four 99s is computed to be 96059600903940400 (the correct/exact answer is ...399). The next slide concludes "There are many decidable problems we cannot solve with real computer: the numbers do matter". Yes, they matter, but this is a terrible example. Exact multiplication of d-digit integers for very large d is easy (unlike, say, integer factorization). Excel got an approximate answer because it's using fixed-precision arithmetic. Maybe it shouldn't, but that's not a computer science issue. This example is not something you'd find in the literature; it was made up for this Powerpoint presentation (along with the name). In my opinion the whole presentation is pretty bad. It covers the usual computer science cliches but isn't well researched or insightful. And Ali G is an idiot, of course. -- BenRG (talk) 21:33, 18 December 2013 (UTC)

Ali G is an idiot like a fox. --Trovatore (talk) 05:50, 19 December 2013 (UTC)

The problem seems more about the input and output. If you could input the two numbers and be able to output the answer then you could write a program to do the multiplication using Arbitrary-precision arithmetic. Such a system would be able to handle any number Ali G could think of but not any number a computer scientist could think of. --Salix alba (talk): 19:07, 18 December 2013 (UTC)

If you have a turing-engine (ie a computer) and sufficient memory to hold both input numbers and the output, then you can compute the result in finite time. Clearly the numbers might be too big to fit in memory - but that's a problem no matter how you do the computation. The observable universe itself isn't large enough to hold arbitrarily large numbers. The problem is nothing to do with computers - it's to do with storage space. You can't multiply two arbitrarily large numbers on an abacus either if you don't have enough paper to write the numbers onto! It doesn't really make a very interesting point. There are much better examples of uncomputable problems that might even seem deceptively easy. For example: Given seven 3×3 matrices with integer entries, can they can be multiplied in some order, possibly with repetition, to yield the zero matrix? No possible computer can solve this - not because the numbers involved won't fit in memory...any modern computer can easily hold the definition of the problem...but the problem can't be solved for every set of input matrices with any kind of turing engine. SteveBaker (talk) 23:49, 21 December 2013 (UTC)

FORTRAN Compiler

I've posted here recently (To hire a programmer for a job, etc). I've got a few good suggestions and I am grateful for them. I've decided in the end that I will need a FORTRAN compiler. Years and years ago I did a lot of FORTRAN, so I hope the programming part won't be that difficult although now it seems the FORTRAN allows to do graphics, etc. The old FORTRAN of course did not have it.

I think StuRat, one of the participants here stated that he does a lot of his work in FORTRAN. I hope someone will recommend which compiler I should download. I checked the web. There is this crazy Intel FORTRAN "Composer" costing over a grand. I will need the compiler for a single task only so for me it won't be a good investment. There are "free" compilers but I am afraid to download from a website I don't know. So, I count on some compromise. BTW, my Windows 7 is 64 bit. Thanks, AboutFace 22 (talk) 16:03, 18 December 2013 (UTC)AboutFace_22

Have you looked at gfortran?--Aspro (talk) 16:57, 18 December 2013 (UTC)

This is a bad decision. The only valid reason to use FORTRAN nowadays is if you have legacy code. Otherwise you are much better off using C -- the available tools are vastly superior. Looie496 (talk) 19:03, 18 December 2013 (UTC)

If the OP is the guy who was asking how to do a web-page, I'd say he needs neither Fortran (it's not FORTRAN anymore, BTW), nor C. He would be fine with html + CSS + JS. OsmanRF34 (talk) 19:39, 18 December 2013 (UTC)

I am the guy who asked how to build web pages. So what? I tried to do this project with Microsoft Visual Studio and C# recently but could not get the precision I wanted. As a matter of fact the iterative errors were overwhelming. At the same time years ago I did a similar job with one of the old versions of FORTRAN compilers and everything worked fine. I do not remember if I used double or even quadruple precision. Most likely it was double, 16 digits. I am waiting for StuRat's comment. He is using FORTRAN every day in his work. BTW, "Fortran" is highlighted by Wikipedia as an error. Thanks, AboutFace 22 (talk) 20:56, 18 December 2013 (UTC)AboutFace_22

So what? Ỹour purpose is not completely clear, but you seem completely lost. Anyway, I don't know what you mean by "highlighted by Wikipedia." Fortran is clear that the name is "Fortran (previously FORTRAN)." Maybe you are confusing your spell checker with Wikipedia. OsmanRF34 (talk) 21:17, 18 December 2013 (UTC)

There are still valid reasons to use Fortran these days, mainly computing speed: due to the simplicity and rigidity of the language optimizing compilers are usually more successful with Fortran code than with other languages. This is pretty much only an issue for heavy number crunching that is cpu bound. Precision, however, is not a reason to use Fortran. All modern languages implement floating point arithmetic as per IEEE-754, so the results will be the same in any language. I don't know what your webpage is doing, but if this is some algorithm to compute something that doesn't converge or result in large computational errors, then your algorithm is wrong and not the language used. Maybe you need to talk to an applied mathematician and not a programmer? 86.128.183.4 (talk) 23:13, 18 December 2013 (UTC)

It is an issue of numerical integration with special functions (Spherical Harmonics). The trouble is I need Associated Legendre polynomials of a very high order l about 200 at times. For this I naturally have to use recurrence formulas. If you are familiar with the subject you should appreciate that many iterations will be involved. AboutFace 22 (talk) 01:37, 19 December 2013 (UTC)AboutFace_22

FINALLY you tell us what problem you are actually trying to solve! Finally! It would have saved a shitload of time to say so from the beginning. The Numerical Recipes books -- the bible for people who need numerical algorithms -- contains a section on how to calculate spherical harmonics, section 6.8. There are C and FORTRAN versions of the book, and each of them gives you source code for calculating Associated Legendre polynomials, using a robust method. It also warns against half-a-dozen other methods that seem like they ought to work but in practice are too sensitive to precision (exactly the problem you ran into earlier). Looie496 (talk) 03:09, 19 December 2013 (UTC)

Thank you. It is interesting! I will check the book, for sure. AboutFace 22 (talk) 04:02, 19 December 2013 (UTC)AboutFace_22

I just checked the book. There is a copyright snag in there. Just read the customers' reports on Amazon. No, thanks. The thing is programming for this project is elementary. It is a snap. I now have my own code in C#. I had written it in Fortran years ago. I need the compiler. Using so many iterations kills the result unless it is in Fortran, this is my experience. AboutFace 22 (talk) 04:18, 19 December 2013 (UTC)AboutFace_22

The gfortran package could be one option, if you are willing to use Linux, and Fortran 77, maybe up to 95. Alternatively, if it's non-commercial, use the Intel compiler for free [[3]]. OsmanRF34 (talk) 08:37, 19 December 2013 (UTC)

That book will explain why you're having problems in other languages and how to properly do the calculation to avoid the errors. Even if the code can't be copied, you can learn from the text and write your own implementation of the robust algorithm. It is definitely the best reference you're going to find for this problem. Katie R (talk) 13:22, 19 December 2013 (UTC)

Thanks to Osman and Katie (?), the last two contributors for your suggestions. Every drop of thought might be helpful eventually. AboutFace 22 (talk) 16:20, 19 December 2013 (UTC)AboutFace_22

I know that it's a waste of time, but I'll try once more to explain. YOU ARE USING A BAD ALGORITHM. The algorithm you are using is numerically unstable. If you don't switch to a more stable algorithm, all this other stuff is useless. Trying to make an unstable algorithm work by operating at ultra-high precision is always a losing proposition. The solution is to replace it with a more stable algorithm. The Numerical Recipes section that I pointed to explains what you need to do. Looie496 (talk) 17:00, 19 December 2013 (UTC)

Looie496, I will get the book. I will order it tonight. It is not a problem. The problem might be that there is no way to improve the algorithm. It is a simple iterative procedure, using recurrence formulas. If you look at them (the recurrence formulas) in Wikipedia (Associated Legendre Polynomials) you will see that there is nothing to improve there, since the operation is so straightforward. If you talk about Numerical Integration which is also one of the components of the whole project of mine, then it is a different story. In that area the algorithms have been optimized and improved for years. But this is a different part and I am not discussing it here (yet!). Of course any criticism is helpful anyway. Thanks AboutFace 22 (talk) 17:42, 19 December 2013 (UTC)AboutFace_22

You don't even have to order it, you can read it online at http://www.nr.com/oldverswitcher.html. Looie496 (talk) 18:43, 19 December 2013 (UTC)

Sorry I didn't see this thread until now (you could have contacted me on my talk page to speed up my response). I concur that gfortran is the way to go, I use it myself. It's a bit tricky to download, as I recall, as the links keep taking you back to the gcc compiler (I think gfortran is bundled with gcc). But, gfortran is free, works well, and supports recent versions of Fortran, as well.

And I would certainly argue that there are many reasons to use Fortran instead of a C variant. Fortran string handling is much simpler, you don't need to include libraries for basic functions, you don't have to worry about pointers and addresses unless you want to, etc. StuRat (talk) 18:18, 19 December 2013 (UTC)

StuRat, thank you. It is all I wanted: your endorsement. It clinches it. I will download it tomorrow. Looie496, thanks for that suggestion also (reading it online)AboutFace 22 (talk) 00:26, 20 December 2013 (UTC)AboutFace_22

You're welcome, and be sure to let us know how it goes. StuRat (talk) 12:34, 20 December 2013 (UTC)

Request assistance from a C++ programmer

I'm attempting to compile the C++ code listed in this article, ISAPI Filter to reject HTTP/1.0 requests. However, I am getting the following error:

Error 1 fatal error C1189: #error :  Building MFC application with /MD[d] (CRT dll version) requires MFC shared dll version. Please #define _AFXDLL or do not use /MD[d]
c:\program files\microsoft visual studio 8\vc\atlmfc\include\afx.h	24	

I'm not a C++ programmer. Can someone please help me to get this code to compile? AnonComputerGuy (talk) 18:26, 18 December 2013 (UTC)

What this error message is saying is that you cannot compile the code in question unless you have Microsoft Visual Studio. MFC stands for Microsoft Foundation Classes. They have multiple files to support the compilation and then run the code (dll's, etc) MFC is a part of VS. You can purchase the VS with MSDN (Microsoft Developer Network) - about $800.00 - professional, 2013. AboutFace 22 (talk)AboutFace_22

I do have Visual Studio installed. I followed the instructions given in the article[4] and I'm getting the above error. Do you (or anyone) know what is wrong? If you (or anyone) have a copy of Visual Studio, can you try it and let me know if you get the same error? AnonComputerGuy (talk) 21:14, 18 December 2013 (UTC)

On a more careful examination it seems the compiler wants a specific header file: afx.h. Could you go down the chain of sub directories and verify that the header in question is there, and if so, open it and check if there is some content inside. Headers are just text files. AboutFace 22 (talk) 21:55, 18 December 2013 (UTC)AboutFace_22

I just found your header on the web. Everything is available! Amazing! I don't know how you post the links. Anyhow, the URL is http://www.cppdoc.com/example/mfc/classdoc/MFC/AFX.H.html AboutFace 22 (talk) 22:00, 18 December 2013 (UTC)AboutFace_22

I don't think that's the problem. But I did check and I have multiple versions of the file, one for each version of Visual Studio installed on my PC. Since this is Visual Studio 2005, I believe the one in C:\Program Files\Microsoft Visual Studio 8\VC\atlmfc\include\ is the one we. I opened it up and it looks fine. But I just installed Visual C++ 2005 yesterday, so it's unlikely the file would have been tampered with.

Did you try to compile the code? Did you get the same error? AnonComputerGuy (talk) 23:54, 18 December 2013 (UTC)

File list for afx.h AnonComputerGuy (talk) 23:54, 18 December 2013 (UTC)

C:\>dir afx.h /s Volume in drive C has no label. Directory of C:\Program Files\Microsoft Visual Studio\VC98\MFC\Include 06/15/1998 12:00 AM 60,113 AFX.H 1 File(s) 60,113 bytes Directory of C:\Program Files\Microsoft Visual Studio .NET 2003\Vc7\atlmfc\include 10/01/2002 04:27 PM 55,851 afx.h 1 File(s) 55,851 bytes Directory of C:\Program Files\Microsoft Visual Studio 10.0\VC\atlmfc\include 11/04/2009 10:43 PM 60,070 afx.h 1 File(s) 60,070 bytes Directory of C:\Program Files\Microsoft Visual Studio 11.0\VC\atlmfc\include 06/05/2013 09:31 AM 61,674 afx.h 1 File(s) 61,674 bytes Directory of C:\Program Files\Microsoft Visual Studio 8\VC\atlmfc\include 12/01/2006 10:55 PM 62,490 afx.h 1 File(s) 62,490 bytes Directory of C:\Program Files\Microsoft Visual Studio 8\VC\ce\atlmfc\include 09/23/2005 01:23 AM 73,323 afx.h 1 File(s) 73,323 bytes Directory of C:\Program Files\Microsoft Visual Studio 9.0\VC\atlmfc\include 04/23/2008 05:08 PM 54,980 afx.h 1 File(s) 54,980 bytes Directory of C:\Program Files\Microsoft Visual Studio 9.0\VC\ce\atlmfc\include 04/30/2007 06:44 AM 61,038 afx.h 1 File(s) 61,038 bytes Total Files Listed: 8 File(s) 489,539 bytes 0 Dir(s) 14,088,654,848 bytes free

Do you have #include statement for this header in your C++ code? AboutFace 22 (talk) 01:30, 19 December 2013 (UTC)AboutFace_22

Yes, I have this header in my C++ code. I followed the instructions in this article, ISAPI Filter to reject HTTP/1.0 requests. Can you (or anyone besides myself) try to compile the code? If so, do you get the same error? AnonComputerGuy (talk) 05:08, 19 December 2013 (UTC)

• Often the most effective way to find a solution to a problem is to paste the error message into Google and see what comes up. In this case I find this Microsoft page, which gives instructions for fixing it that hopefully will work for you. Looie496 (talk) 03:40, 19 December 2013 (UTC)
  • Often the most effective way to find an answer to a question asked on the Reference Desk is to pay attention to what the OP is asking and try it yourself. In this particular case, I am following the instructions given in this article, ISAPI Filter to reject HTTP/1.0 requests. All I am asking for is a C++ programmer to try to execute these instructions and try it for themselves.
  • Are there any C++ programmers on the Computing Reference Desk? Is so, can you please read this article and see if you can get this code to compile? There's not a lot of code, so this should only take a few minutes to try it. AnonComputerGuy (talk) 05:19, 19 December 2013 (UTC)

I can get to it later today - I'll use VS2005 since that's what you mentioned above. However, could you try to add "#define _AFXDLL" to the top of your source or turn off the /MD switch? It will be somewhere in your project configuration - each item in the configuration pages mentions which compiler switch it applies, so you should be able to find it. Katie R (talk) 13:16, 19 December 2013 (UTC)

Good news! I got it to compile. It looks like the article is missing an instruction. To get it to compile, I went into Project Properties | Configuration Properties and changed Use of MFC to Use MFC in a Shared DLL. AnonComputerGuy (talk) 17:01, 19 December 2013 (UTC)

Facebook linked to Gmail, maybe? [moved from "Miscellaneous"]

I had an email exchange with a Gmail user, and to my surprise my original message, in her response to me, had the layout of a Facebook wall post, complete with my Facebook avatar (though it had my real name, not my Facebook alias, next to the avatar). This is odd: the sender has a Facebook account too, but we're not friends, and she's not a member of the only group in which we could conceivably share membership. So, how does she (or her Gmail account) know my Facebook avatar? Drmies (talk) 05:46, 18 December 2013 (UTC)

You might want to ask this on the computer desk, given the subject and that that desk gets a lot more attention than this one. μηδείς (talk) 17:13, 18 December 2013 (UTC)

Will do (I mean, done)--thanks. Drmies (talk) 19:20, 18 December 2013 (UTC)

Even if you are using Tor, proxying your IP address, using a newly-created email address, and avoiding disclosure of any personal information, it is still possible to uniquely identify which human you are. This is called data mining, and it is a reality of the 21st century. At all times, without even realizing it, you are leaking personal information - in digital form, and in other forms. A dedicated attacker (like Google or Facebook or a government police force) can use a strategy of centralizing trillions of pieces of insignificant information into one common repository for the purposes of producing meaningful connections. The exact nature of the connections, the exact details of which pieces of data forge the connections - are irrelevant. What matters is that if an organization collects enough tiny insignificant pieces of information - it can construct a very complete picture of the whole "situation." Two specific technologies that are used include Facebook Beacon and Google Analytics; these surveillance tools collect information about users on a wide variety of websites, even when the users perceive that they have no interaction with Google or FaceBook. No doubt, there are many, many, many trillions more pieces of information that these corporations use to "complete the picture." Nimur (talk) 20:24, 18 December 2013 (UTC)

It's often possible through police work to identify people who are trying to hide, and one way they can do that is by analyzing web server logs. But that isn't an automated process and it's obviously not relevant to what happened here. I don't think there are any companies that go out of their way to do automated forensic analysis of logs in order to deanonymize you for the purpose of inserting a photograph into your email that reveals that they've deanonymized you. It doesn't seem like a winning business proposition. -- BenRG (talk) 00:25, 19 December 2013 (UTC)

The case of the Harvard bomb hoax is pretty specific. They had a starting point (all Harvard students with exams approaching and who could use a little bit more time), and, although what goes through Tor is hidden, they can still know who entered the network. PUt on top of it that there are enough human resources to track down this kind of things. A 'normal' user, who fears being tracked for commercial purposes, as it's the case of a user of FB and Google, would be pretty anonymous through Tor or similar strategies. It's important not to get paranoid here. Also notice that the Harvard guy was not caught using data mining, but something more like computer forensics. OsmanRF34 (talk) 08:48, 19 December 2013 (UTC)

While I would agree we should be careful about reading too much in to the Harvard case, it and in particular your post does actually raise some important considerations. You mentioned 'they can still know who entered the network', but the obvious question is 'do they'? In the Harvard case they (meaning the FBI) likely had the advantage that the purpose was using the Harvard university network who likely keeps logs, and they would almost definitely have had Harvard's cooperation and were given access to these logs after the fact. This would apply in some other law enforcement cases but it gets more tricky if you have no particular ISP to target but lots of possibilities.

On the other hand, there's also the question over who's keeping track in the general case. It always seemed likely and the recently leaks seem to affirm, that intelligence agencies particularly US ones and probably other ones are in fact keeping track of those using Tor and also have access to a lot of key routes (and probably run a few of the entry nodes themselves) so they do probably do know the precise time any specific IP in the US and most likely IPs in other places access/make requests to Tor. Having this sort of information, combined with access to what comes out of Tor exit nodes which they likely do all have, does of course make these sort of correlation attacks (link doesn't cover this sort of thing) much more feasible in the generic case as well.

And it's perhaps worth remembering even if you're using HTTPS, even without breaking the encryption or having access to either end, adversaries can often guess at what you're accessing on the other end by the size of your requests etc.

Nil Einne (talk) 03:06, 20 December 2013 (UTC)

Indeed, Tor has its weak points, [| The Guardian ] claims 'We will never be able to de-anonymize all Tor users all the time' but 'with manual analysis we can de-anonymize a very small fraction of Tor users'. But what do you expect when you are trying to hide from any government agency, which has much more resources than the average user and can make the rules to seize any device and subpoena any information? OsmanRF34 (talk) 13:06, 20 December 2013 (UTC)

1. What email provider and email application (or webmail) are you using? 2. How sure are you that this Facebook information was not actually in the email you sent to her? It's extremely unlikely that Gmail added it. A Google+ avatar maybe, but not a Facebook avatar. -- BenRG (talk) 00:25, 19 December 2013 (UTC)

• I'm using Outlook Webmail and my university email, connection courtesy of Charter. I don't use Facebook to email or to make email connections (don't even know how to do it). I'm not friends with this person on Facebook and, again, the name accompanying the avatar in her response was my name, not my Facebook alias. Thanks, Drmies (talk) 15:42, 19 December 2013 (UTC)