Wikipedia:Reference desk/Archives/Computing/2010 March 21

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.

March 21

Local Javascript pop-up with pop-up blocking on

This is a somewhat odd request, and what I want to do might not work.

I am using Safari in OS X as my default browser. I've written a little Javascript page that will display some scheduling data to me quite quickly. That part of it is nice and works fine.

The way I have it set up, there is an icon on my dock which, when clicked, opens the file LAUNCHER.HTML in Safari. LAUNCHER then runs some Javascript that opens the schedule file as a pop-up (let's call it SCHEDULE.HTML), and then closes itself (just a window.open and window.close).

This works well when I have pop-ups enabled in Safari. But ideally I would have my "Block Pop-Up Windows" setting turned on. When I do have it turned on, LAUNCHER can't launch the pop-up. (Duh.)

The advantage of using a LAUNCHER popup rather than just linking to SCHEDULE.HTML is that I can strip away the address bar, status bar, and so forth of the pop-up. I don't think I can do that without doing a window.open(), right? And I suppose there's no way to enable pop-ups just for local HTML files.

The obvious easy "fix" is to scrap LAUNCHER and make SCHEDULE.HTML position itself accordingly and just put up with the extra address bar and status bar even though it is ugly (because really, the purpose of this application is just to be functional—it is just a home-made thing I use for my own stuff), but I'm curious if there is some other thing I have not thought of. --Mr.98 (talk) 00:42, 21 March 2010 (UTC)

No, the "obvious" easy fix is to run a script from the dock. Any modern scripting language has a built in GUI library and any built-in GUI library has an HTML window rendering class. So you just open a window using it and se it to render whatever you need. If you're on modern hardware then the fastest workaround is using a vm with safari in it, maybe under windows if its easier for you to get the cd. That should take you lile 12 minutes of yourtime plus download time/ waiting forthe guest os to install. The 12 minutes includes you figuring out how to use your vm software to make a checkpoint that your shortcut will open. —Preceding unsigned comment added by 80.187.97.181 (talk) 01:14, 21 March 2010 (UTC)

I'm not sure I understand what you are suggesting. Could you elaborate with fewer acronyms and more practical explanation? --Mr.98 (talk) 01:41, 21 March 2010 (UTC)

Agreed with above, I'm not sure how the use of a Virtual Machine applies to your question. It appears the basic stuck point is Safari's lack of security zones, which in browsers such as Internet Explorer or FireFox, would allow you to apply different browser settings to different categories of sites. So with Safari as your browser, your "obvious" fix is indeed the choice between a) unblocking popups for all sites, or b)rewriting your "launcher" to not rely on popups. Alternatives would include rewriting (assuming its local only) schedule.html as an HTA. This supports rendering HTML through client-side script but doesn't require launching the full browser. Cander0000 (talk) 04:59, 21 March 2010 (UTC)

It's too bad that Safari's popup blocker can't be configured. You might try looking at Mozilla Prism or another light little SSB that you could use for this one little HTML file. I tried using it once before, and it worked pretty well. Mozilla Prism website Indeterminate (talk) 04:24, 22 March 2010 (UTC)

Installing MSYS on 64bit vista

I am trying to make msys work on 64bit vista by following these instructions: http://old.nabble.com/Installing-MSYS-on-Windows-Vista-x64-td16904988.html. However I cannot seem to find the "MSYS-1.0.11-20071204.tar.bz2" file anywhere. Is there an easier fix for this issue? Is there an available substitute file that can be used in place of the required file. 72.188.46.69 (talk) 01:47, 21 March 2010 (UTC)CompuGuy

Yeah, that's a daily build which has probably disappeared off the internet. Did you try following the installation instructions on the project's website? Indeterminate (talk) 04:11, 22 March 2010 (UTC)

US federal government software

Works of the US federal government are in the public domain, and I think the government writes a lot of software. Is there a centralized Website repository somewhere containing all of this software? (I realize that (a) it must just license a lot of the software it uses, so I'm not going to expect every government payroll system and accounting system to be available; (b) I'm probably not going to be able to get their nuclear explosion simulators; and (c) I probably have little use for Medicare compliance certification audit software written in Ada.) Comet Tuttle (talk) 06:45, 21 March 2010 (UTC)

I don't think there's a central database. (There's not a central database for US federal government publications, much less software.) I also think they likely contract a lot of that work out, and contractors are not covered by the "works of the federal government" copyright clause. Nuclear weapons codes, even if they were declassified (and some old ones are), are technically works of a contractor (Los Alamos or Livermore labs, which are Government Owned Contractor Operated facilities). I did some software work for a US federal agency once, and they definitely weren't that interested in having it distributed (the question of whether it should be distributed actually did come up, but not because it was public domain—it's because the software was part of a database that theoretically a small-but-greater-than-zero number of the academic community might find useful). It was not sensitive or classified in any way, but they considered it solely useful to the internal functioning of their department. (Which actually exempts it even from FOIA if I recall.) The truth is, my particular bit of code was really quite boring and isolated and specific to the task. While one could imagine a well-organized code database in which I entered in component bits of my software, explained how they worked, made sure everything was commented well, and indexed it in such a way that others might find it useful... well, let's just say they weren't paying me to do that! And I suspect they aren't paying anyone to maintain such a central database, which in the end, with the federal government, is how anything exists at all (if it ain't funded, it ain't happening). Now, you probably could get certain types of software via FOIA requests, if you wanted them. But I don't think the government automatically distributes or collects them. --Mr.98 (talk) 14:16, 21 March 2010 (UTC)

As just an aside. The question of ownership of government v. contractor code comes up with some frequency in discussions of electronic voting. By contracting electronic voting out to private companies (e.g. Diebold), the government gives them the ability to make their code copyrighted and proprietary, and thus outside of public scrutiny. This has met with criticism by security people, unsurprisingly, who believe that security through obscurity on something as important as elections is a dangerous idea. Thus some have argued (I think Bruce Schneier is one in this camp?) that the government should code these things in-house, and thus the code should be public domain, and open to outside analysis. Just putting that out there, as it came to mind. --Mr.98 (talk) 14:29, 21 March 2010 (UTC)

My experience is that a lot of people don't care whether it's in the public domain, provided it's FLOSS. [1] Some don't even go that far, and simply argue the source code should be open to public scrunity. [2] Of course even if you do demand the source code be in the public domain, it doesn't automatically follow that they have to be produced by the US government since I believe it's possible for private companies to release things into the public domain in the US and there's no reason why that can't be a legal requirement for voting machines, just as FLOSS or open to public scrutiny could be. Nil Einne (talk) 17:55, 21 March 2010 (UTC)

Well, that's true. The issue of ownership matters largely because contractor produced code (and everything else) is not necessarily subject to information openness laws (like FOIA), and there is not a lot of precedent for contractor openness. --Mr.98 (talk) 18:04, 21 March 2010 (UTC)

For comparison, consider software written by a National Laboratory (which really are the premiere science and technology centers that are operated by the civilian side of federal government). Take a look at Argonne National Laboratory Mathematics and Computer Science Division. For example, here's the copyright statement for Message Passing Interface, which is authored by government employees, but may incorporate work from private contractors. This work is not exactly in the public domain, although "Permission is hereby granted to use, reproduce, prepare derivative works, and to redistribute to others," and "the Government is granted for itself and others acting on its behalf a paid-up, nonexclusive, irrevocable worldwide license in this computer software to reproduce, prepare derivative works, and perform publicly and display publicly." Similar sorts of license statements accompany any work released by the federal government. Nimur (talk) 18:54, 21 March 2010 (UTC)

GPU-assisted video transcoding on ATI?

Is there another GPU-assisted video transcoding program available for ATI? The Avivo program provided by ATI is too limited in its options. thanks F (talk) 09:14, 21 March 2010 (UTC)

Yes Cyberlink PowerDirector [3] and CyberLink MediaShow Espresso [4]. If I had to guess, these use the same internal encoder although the feature set may vary. These were evaluted here which doesn't come out very favourable for ATI although that was in August 2009 which can probably be said to be a long time ago given that GPGPU is still a fairly rapidly developing field. In particular they used PowerDirector 7 but 8 is now out and while they don't seem to say what MediaShow Espresso they used (I'm not surprised, I've never been particularly impressed with the quality of Pc Perspective's reviews) I believe 5.5 was only released in 2010 [5]. In any case this found the results tended to vary depending on the source material and this didn't find any substanial quality differences. Windows 7 also uses ATI Stream for the drag and drop transcoding (for portable devices) [6] [7] but I doubt you get much control. It may be part of SHED [8]. There's also an AMD plugin for Sony Vegas Movie Studio 9 [9] [10] [11] and Adobe Premiere Pro CS4 [12]. How much control you have with these, I don't know. Personally though, while I don't know if things are better on the Nvidia side (with other apps) from what I've just read GPU assisted transcoding still doesn't sound particularly interesting to me, except perhaps for cases when I absolute need extreme speed, any case where I care about quality I'd stick with x264 (presuming your talking about h.264). And I do have a Nvidia card altho I do rarely transcode. Nil Einne (talk) 17:44, 21 March 2010 (UTC)

Thanks, that's a very detailed answer. I gave MediaShow Espresso a go. Yes, it is ~4 times faster than CPU encoding, but the quality is clearly inferior compared to CPU encoding at the same bit rate. Do you know why? F (talk) 09:51, 22 March 2010 (UTC)

Fortran

I write a fortran program but I can't run it and I can not get output please tell how it compile and run I use windos xp.Supriyochowdhury (talk) 09:57, 21 March 2010 (UTC)

Which coding environment did you use? Presumably one supplied by one of the tools you downloaded and installed after your previous question was answered. Astronaut (talk) 11:30, 21 March 2010 (UTC)

Do you get an error when you compile it ? If so, what is the error ? Note that a file may have been produced in the directory where you did the compile, which contains the error messages. StuRat (talk) 17:17, 21 March 2010 (UTC)

Also, if you will provide the Fortran source code, I could debug it for you. StuRat (talk) 17:18, 21 March 2010 (UTC)

Did you compile the code? It's not clear from your question, but I wonder if you have merely written the source code but have not yet tried to compile it (since you did not mention any compiler error message). FORTRAN is a compiled language, so you need to process the text file containing your program source to generate the executable program. Nimur (talk) 22:22, 21 March 2010 (UTC)

Updating to Norton 360 v.4 from v.3 for free --- is that possible?

I have just installed Norton 360 v.3 in my laptop. Is there anyway I can update it to v.4 for free? Please give me outright instructions. I'd be so glad to read from you. Thanks, everybody! —Preceding unsigned comment added by 112.202.208.174 (talk) 11:48, 21 March 2010 (UTC)

If you have a current (i.e., unexpired) subscription, go here. Symantec does a pretty good job of hiding this page, because they want you to think you have to upgrade when you renew your subscription. The only place I could find a link to it was right on the Norton front page, but it is very well-hidden. It is named "Norton Product Updates" and is located right above "Upgrade to latest version". To a novice user it would sound like "Norton Product Updates" is where you go to get, say, the latest definitions, not a new version. Quite misleading. Xenon54 / talk / 12:53, 21 March 2010 (UTC)

fortran

which mistake frequently doing a student first time during run a fortran program that is why showing error.Supriyochowdhury (talk) 14:59, 21 March 2010 (UTC)

Before you get any more answers about Fortran, you must answer the questions you are being asked in return. What program are you using to write Fortran? What program are you using to compile Fortran? Without answers to those questions, it is not possible to help you. -- kainaw™ 15:45, 21 March 2010 (UTC)

Maybe a variable which is declared incorrectly ? Often a problem for me. For example, I make a single variable when I start writing the program, then, in the middle of writing it decide I need it to be an array, but don't rememebr to go back and change the declaration. StuRat (talk) 17:22, 21 March 2010 (UTC)

Most programmers would say "Your first mistake is using Fortran in the first place." - it's a horribly obsolete and completely nasty programming language. But there is no way we could hope to help you without at least seeing the source code and the error message(s) that the compiler is giving you. SteveBaker (talk) 00:11, 22 March 2010 (UTC)

And many of us Fortran programmers consider C to be non-intuitive and unnecessarily complex, such as when dealing with string handling. StuRat (talk) 02:00, 22 March 2010 (UTC)

Obligatory non-quote from McCarthy: "You're doing it completely wrong." -- kainaw™ 03:28, 22 March 2010 (UTC)

It depends what you are programming. If it's pure time-critical number crunching, Fortran is still the fastest language out there (short of assembler).213.160.108.26 (talk) 21:22, 22 March 2010 (UTC)

That's complete nonsense! All of the major programming languages are going to share the same arithmetic/logic compiler back-ends. C, C++, Pascal, Fortran - all of them get the same optimisation, etc. There are no linguistic features of Fortran that make it any better than the other major compiled languages as regards doing number-crunching. Fortran has zero strengths - only weaknesses (someone is going to mention complex numbers very soon now - and I'm already laughing). SteveBaker (talk) 04:11, 24 March 2010 (UTC)

My understanding was that Fortran, being a rather pointerless language, was not prone to Aliasing (computing), and hence was easier for the optimizer to optimize. -- Coneslayer (talk) 13:19, 24 March 2010 (UTC)

Top ten websites

What are the top ten websites, most used on the internet?? --Extra999 (Contact ^me + _contribs) 16:44, 21 March 2010 (UTC)

Search engines and porn sites (not in that order). ¦ Reisio (talk) 17:13, 21 March 2010 (UTC)

Those are categories of sites, not individual sites. --Mr.98 (talk) 18:01, 21 March 2010 (UTC)

Alexa Internet ranks websites here: [13]. There are questions as to how accurate they are, but they're something to start with. --Mr.98 (talk) 18:01, 21 March 2010 (UTC)

In terms of the Alexa rankings, I know for http://www.wikipedia.org after a poll we decided to choose and order the top 10 by the most visited which many would feel makes sense. But bizzarely even though these are entirely wikimedia run projects and so producing stats on most visited was never going to be that difficult (and in fact I think they were always available) we originally used Alexa Meta:Talk:Www.wikipedia.org template#Do not use Alexa as main source! Meta:Talk:Www.wikipedia.org template#most visited wikipedias and when we changed we changed to using our local stats it made quite a big chance to the order of some and even what we listed. I also noticed [14] which while probably not suitable for the above article may be of interest. However I agree in the absence of anything else, Alexa is an okay starting point Nil Einne (talk) 19:58, 21 March 2010 (UTC)

I was always baffled by reliable sources that rely on Alexa rankings. The Alexa toolbar is nothing short of spyware - it intentionally installs itself with the professed motive of monitoring a person's web browsing habits. Do we really want to represent the internet's usage patterns based on the sub-sample of users who knowingly or unknowingly install spyware and relay all their activities to a third-party? Nimur (talk) 16:34, 22 March 2010 (UTC)

Thanks everybody. --Extra999 (Contact ^me + _contribs) 04:04, 22 March 2010 (UTC)

If it's not too late, the top 100 produced by Nielsen is reported here. AlmostReadytoFly (talk) 14:51, 24 March 2010 (UTC)

Putting text into a database

I have a large number of text files. In each text file one or more entities are described in a standardised way i.e. a short descriptor such as "Name" "Location" "Type" is followed by the appropriate detail. The order of those descriptors is the same for every entity. Sometimes the content such as an address will continue on further lines after the descriptor.

Is there any easy (which means quick for a non-programmer to do) and free way of making a simple database out of this? I have something in mind like a card file. Thanks 92.29.149.119 (talk) 21:13, 21 March 2010 (UTC)

Free, yes. Easy as in non-programming... I doubt it. It is not a hard job, as far as they go, to get that data out and into some kind of database file (it is a pretty standard and straightforward scripting job, assuming things as as regular as you describe them), and going from the database file to a simple database application is pretty straightforward (could be done with PHP/MySQl or something like OpenOffice Base)... but you still need that initial step of extracting the data from the files into the database format. While not difficult for someone with a small amount of scripting, it would probably require some scripting (that is, you probably won't find a solution that does it without a little scripting). The reason for this is that you need to tell the script how to interpret the text files (what is a field name, what is a field value), and since it is non-standard you would need to write a custom little script to do this for you.

It's the kind of thing where if you posted maybe five of the text files here that were representative of the variety, we could probably write the conversion script for you pretty easily. It's probably about 10-20 lines of code total depending on the language, nothing complicated. --Mr.98 (talk) 00:39, 22 March 2010 (UTC)

This sounds like something you could do with a spreadsheet. 66.127.52.47 (talk) 07:15, 22 March 2010 (UTC)

Only if you wanted to reenter all of the data manually. I don't know of any spreadsheet program that can extract the data from multiple text files in this way. If you have it in a spreadsheet, you essentially have it in a database (converting between a spreadsheet and a database table is trivial). A simple (custom) script could get it into spreadsheet format; it's not essentially different than making it into a flat file database. --Mr.98 (talk) 14:21, 22 March 2010 (UTC)

I think you are being rather pessimistic about entering or extracting the data. It could quite easily be converted into comma seperated values by batch replacing for example "Name" with "Name," and so on, or some other delimiter rather than a comma. Assuming I've then got it in CSV format, what free software is there that would allow me to see one "card" at a time and search particular fields of the cards? There would not be much point loading the CSV file into a spreadsheet as that wouldnt do anything more than what a plain text file could do. I suppose I could write a BASIC program myself, but that would take up a lot of time and perhaps there is something ready-made. Thanks. 84.13.47.185 (talk) 15:36, 22 March 2010 (UTC)

If you could convert it to CSV, then it's just a matter of finding a good displayer/reader. What you want sounds like FileMaker Pro. I don't know the free database software very well, but there are a few out there. I've not had good experiences with OpenOffice.org Base but it has been a long time. Kexi looks a little more straightforward? Maybe others have tried these out. Each of these should be able to import CSV without difficulty. Creating an application that is a simple cardfile should not take very long at all with any of these. --Mr.98 (talk) 16:34, 23 March 2010 (UTC)

There are basically two approaches to reading data from a file. One is using a separator, like a comma, between fields. The other is to use fixed length fields. Both methods can fail, however. The separator may sometimes be absent or extras may be present. For example, if a space is used, there may be a space within some names. The length of some fields may also be wrong, such as if a tab character is sometimes used to space out columns and other times the space bar is used. So, in either case, it's likely some manual edits will be required to fix such problems, before importing into a DB. StuRat (talk) 15:52, 22 March 2010 (UTC)

So things havnt really changed from when you had exactly the same considerations when using BASIC decades ago. 78.149.193.98 (talk) 20:18, 22 March 2010 (UTC)

No, not really. Custom data is still custom data. Getting it into a standardized form is still a pain. --Mr.98 (talk) 16:34, 23 March 2010 (UTC)

Clever AI

We havnt got anywhere near having intelligent robots yet. What is the cleverest AI system that currently exists? I'n not expecting it to talk. Thanks 92.29.149.119 (talk) 21:18, 21 March 2010 (UTC)

It might help to define (what you mean by) "clever". At one point in time, one might have guessed that a computer would have to be "clever" to beat the world chess champion, but Deep Blue accomplished that. However, I don't think anyone would class Deep Blue itself as "clever" as it basically brute-forced the problem, although some of the algorithms which went into pruning the search space might have been (but the cleverness there was on the part of the programmers). On the other side of the spectrum, you have things like genetic algorithms, which routinely come up with solutions a human might not have ever considered, so may be classed as "clever" in that sense. Genetic algorithms, though, proceed through a pretty stupid guess-and-check system, so some may argue that the algorithms themselves are not "clever", merely diligent. - As a final note, saying a horse is "clever" merely means that it is well behaved, which may be a much easier standard for an AI to reach. -- 174.21.243.94 (talk) 22:00, 21 March 2010 (UTC)

I consider the disorganized set of distributed computing resources available via the Internet, through search engines and wikis, to be a "clever" system. A natural language query can be submitted; a series of computers translates that query into machine-readable form, transmits it over a network, processes the query, locates resources, and delivers a meaningful answer - entirely transparently. For example, I can ask "the internet" a question, like "what do pandas eat", and get an answer - without ever knowing anything about how computers process data. The ensemble of technologies, including computer networking and routing, HTTP, web services, natural language processing, wikis, and distributed computing, all result in an answer to virtually any kind of question I can pose. Nimur (talk) 22:27, 21 March 2010 (UTC)

I sat through a painful lecture a long time ago title "AI is Dead". The point is that the ideas of AI in the 60s and 70s were abandoned during the 80s (to be overly general). The main reason is that we don't have a need for another human, so why make an artificial human? We need machines that are better at certain tasks than humans are. So, we end up with "clever" machines for specific uses. For example, it is well known that computers can trade stocks better than humans. It used to be assumed you had to be real clever to profit in the stock market. Now, you just have to be a computer running a clever program. -- kainaw™ 23:59, 21 March 2010 (UTC)

I don't buy the "We gave up doing AI because we don't need it" argument. It was flat out too hard. So we're doing all of the bits around the edges - speech, language comprehension, vision, robotics, knowledge representation, etc. All of those fields are producing great results - and what was being attempted before wasn't - an nobody likes failing. SteveBaker (talk) 00:08, 22 March 2010 (UTC)

Play with http://www.20q.net for a while and tell me it's not clever? ;) But it's not really AI.. Vespine (talk) 00:14, 22 March 2010 (UTC)

Clever? It guessed Tori Amos when I was obviously thinking of Liz Phair. Pshhh. Not even remotely the same. Nimur (talk) 15:08, 22 March 2010 (UTC)

And presumably any true A.I. would need to combine all of these things. Our own brains have separate centers for language, visual processing, speech, etc., and even within those centers separate places for spatial thinking, abstract thinking, mathematical thinking, etc. I think cog sci has shown pretty clearly that the idea that brain intelligence is just emergent from throwing a bunch of neurons together has been pretty dead for a long time (just like throwing a bunch of transistors together does not make a CPU—there's a lot of complicated engineering involved to get everything to work together correctly). Today we would probably say that A.I. is not just one problem but a bunch of interconnected problems. Artificial intelligence has more details on this. As for "cleverest A.I." system, I think I'd have to agree with those who say that by itself that is not quite a well-formed question (or at least no better formed than asking who the cleverest person is). We have machines that can do very clever things with communication, data, visuals, language, etc. None of them can do anything more than an illusion of "general intelligence," and not a great one at that, but in specific intelligences, there are probably sure winners. --Mr.98 (talk) 00:56, 22 March 2010 (UTC)

20Q is basically a slightly complexified version of the venerable "guess animals" game - which I first saw in about 1974 - and I'm pretty sure it was old even then! You can write a minimal version of that software in 50 lines of C++ code - or a pretty version in about 300 lines - and it seems every bit as amazing as 20Q. You can download my version of it here.

OK, you think of an animal - and I'll try to guess what it is.
Is it furry?  n
Does it live underwater? n
Does it have more than four legs? n
Does it slither? n
Is its tail nearly as long as its body? n
Does it have really big ears? y
Is it an Elephant? y
Hooray! I got it right!  It was an Elephant!

OK, you think of an animal - and I'll try to guess what it is.
Is it furry? n
Does it live underwater? y
Does it have scales? n
Does it have to surface to breathe? n
Is it more or less as wide as it is long? y
Is it too small to see without a microscope? y
Does it have pseudopods? y
Is it an amoeba? y
Hooray! I got it right!  It was an amoeba!

SteveBaker (talk) 04:34, 23 March 2010 (UTC)

20Q learns from the answers of players which means it's vunerable to misinformation or disagreements and differing definitions. For example:

It is Kuala Lumpur.

Is it located in the southern hemisphere? You said No, 20Q was taught by other players that the answer was Yes.

Was it involved in World War II? You said Yes, 20Q was taught by other players that the answer was Doubtful.

Does it have a subway system? You said No, 20Q was taught by other players that the answer was Probably.

Is it a honeymoon destination? You said Sometimes, 20Q was taught by other players that the answer was No.

Is it an influence on science and technology? You said No, 20Q was taught by other players that the answer was Probably.

1 is clearly wrong. 2 is highly debatable, while not a front Kuala Lumpur was captured by the Japanese as with most of Malaysia. 3 it depends on what you mean by a subway system, parts of the RapidKL Light Rail Transit is underground although it isn't that big a part (4.4km according to some sources) and it isn't commonly thought of as a subway system from my experience however given that yes and no would be acceptable. 5 I was thinking this at the time it's a somewhat problematic question anyway most stable tourist destinations are going to be a honeymoon destination for a resonable number of people (even places like Iraq and Afghanistan but be an occasional honeymoon destination although rare enough it's perhaps fair to say it's irrelevant). 6 I was wondering when answering what to say and in retrospect sometimes was probably a better answer here but even though I may not like it, Malaysia and therefore KL's contribution to science and technology isn't much, the publishing record of academics for example tends to be fairly dire (as much as the government wants Malaysia to be at the forefront of science and technology, they haven't really achieved that much).

Of course any case when the answer is "Probably" or "Doubtful" clearly indicates it doesn't know enough.

Incidentally in both the Kuala Lumpur case and when I tried Auckland (and also got debatable information), it failed to do it in 20Q and gave me completely wacky guesses (in the KL case, it asked me if it was Vietnam then later Ho Chi Minh city, in Auckland it asked me if it was Zanzibar, Bali and Mauritius). And I was being kind to it by choosing relatively big cities. When I tried the classic version with rambutan I found it doesn't even exist in the database... (It did win with durian.)

Nil Einne (talk) 13:46, 23 March 2010 (UTC)

That's the thing really - you KNOW you're going to sometimes collect garbage "knowledge" from members of the public - so the trick is to build a system that can robustly recover from that. In "Guess Animals", if you decided that an elephant is "furry" (well, they have some hair - so maybe) - then the system will fail to guess "Elephant" and maybe come up with "Aardvark" or something. But when it fails, it asks you to enter a question that will distinguish an elephant from an aardvark - and you're probably going to say "Does it weigh more than a ton?". Now, the next time someone thinks that elephants are furry, it'll correctly guess "Elephant" despite that error. This results in it asking more questions (although rarely as many as 20) without the need for "Maybe" and "Irrelevant" options (which really only confuse the system IMHO). When I had Guess Animals on my web site, it filled up with a gazillion animals and an awful lot of other junk too (there were dozens of things like "My brother", "Car" and "Oak tree" in there - along with a lot of obscenities and typos) - but it still guessed real animals correctly better than 99% of the time. I'd say that its biggest failing is when (for example) someone is thinking of specifically an "Indian elephant" - the system will guess "Elephant" and the user is left feeling a bit disappointed when they have to say that the computer got it right...because in a sense, it didn't. They may then decide to say "No" and then they'll be asked to enter a question to distinguish an "Elephant" from an "Indian Elephant" - and that gets a bit tortured too. But if the user enters "Does it live in Asia" then the next time someone is thinking of an Indian Elephant, they'll get the more specific answer. What really screws up Guess Animals is when someone puts in a really weird question high up in the 'tree' of answers - like "Can it swim?". Because most people don't know whether (say) a Leopard can swim or not - you eventually end up with "Leopard" on both sides of the 'tree' - and that results in a lot of redundancy in the resulting data structure. SteveBaker (talk) 13:44, 24 March 2010 (UTC)

Nil Einne points out that a rigorous, objective, universal knowledge system is corruptible by public perceptions, invalid information, and so forth. That's why I am so amazed that a web search engine can accurately infer the intended meaning, even when the exact query is vague or unqualified. For example, if I query Google for "orange", there are thousands of legitimate ways for the system to qualify and interpret such a vague request - do I want produce, paint, orange juice, am I looking for history of Holland, or what? But rather than play 20-questions, Google can take into account my search history, my location, and send me to Gryphon Strings - an Orange Amplifier dealer five blocks from my house, which currently has one Orange Tiny Terror 15-watt amp in stock. Then Google gave me the price, a map to the dealer, and their phone number. That was a pretty obscure query - but they got all that information for me! How did Google know what I wanted? Well, they are tracking me and my purchase habits; they know where I have clicked before; they know what I have bought before; they know my location, my interests, and all kinds of other meta-information about me. When I am being paranoid and proxying or otherwise hiding my personal details, Google needs to play "20 Questions" with my search queries. From the standpoint of privacy, I'm terrified of this capability - but from the standpoint of artificial intelligence, I have to say that this is pretty darned amazing. If I walked up to a friend down my hallway or an office-mate who knows me well, and said "ORANGE" with no further qualifications, I doubt I could get such a meaningful response. (I know, because I have tried this). So, the internet has proven that targeted marketing, which is one of the most profitable applications of expert systems, can actually outperform a human in terms of "clever" interpretations of responses - in specific circumstances. Nimur (talk) 17:02, 23 March 2010 (UTC)

Complete System restore for Medion Akoya

Hi everybody we have a problem with our Medion Akoya, model number: WIM 2180. We have a password on it which my dad has forgotten. We need to perform a complete system restore on the laptop. Is there anybody who knows how to do this? Any help appreciated. Thanks, Hadseys 21:23, 21 March 2010 (UTC)

Is the password just the windows administrator password? If so, google for "reset vista password", or "reset XP password" as appropriate, and you'll find methods to do this without zapping the whole system. Or do you mean the BIOS password? -- Finlay McWalter • Talk 21:31, 21 March 2010 (UTC)

TCP/IP weirdness.

Network programming is not my strong point. I'm trying to get a very basic TCP/IP communication going between a Linux box and a Windows-7 machine for a little project I'm doing in my spare time at home.

To keep the explanation simple, I'm using the code from [15] to send a TCP/IP packet from a client to a server - which simply sends the packet content back again. I'm compiling the identical code under g++ under Linux and Cygwin under Windows. I have the pair of programs (client and server) use port 5010.

1. The program works fine when I run the server on Windows-7 - the client works on both the Linux box and the Windows box.
2. The program works fine when I run the server under Linux - but ONLY if the client is also running under Linux.
3. Nothing works if I run the server under Linux and the client under Windows.

The Windows machine is running "Symantec Endpoint Protection" and the firewall is turned on. I tried creating a custom firewall rule in case port 5010 was disabled - but I've never played with that stuff before - and it's perfectly possible that I messed up. However, because everything works in case (1) above, I don't see how it could be a problem of port blocking.

This is all running on my home LAN via a LinkSys router.

Help!

SteveBaker (talk) 22:44, 21 March 2010 (UTC)

So the server, if running on Linux, is only reachable from the machine itself. Sounds like the Linux machine has a firewall enabled too ? Unilynx (talk) 22:51, 21 March 2010 (UTC)

Also check to ensure SELinux is not blocking network traffic. -- kainaw™ 23:54, 21 March 2010 (UTC)

<smacks forehead> that was it - disabling the Linux firewall fixes it. Many thanks! SteveBaker (talk) 00:05, 22 March 2010 (UTC)

As in all such matters, Wireshark is your friend - in this case you have the luxury of running it on both parties, so you can see if the packet was sent ok, received ok, and delivered (os->app) ok, and you get a better insight into failure that ECONNREFUSED will give you. It might be instructive to temporarily turn one or t'other firewalls back on to see what they really do - some barf your SYN immediately with a RST/ACK, others drop the SYN and vexingly play possum. -- Finlay McWalter • Talk 00:30, 22 March 2010 (UTC)

Yeah - I should probably do that - however, I was hoping for a quick fix and I got one! Putting in a custom "Additional Service" for my new software even allowed me to turn the Linux firewall back on. SteveBaker (talk) 00:58, 22 March 2010 (UTC)