Perl 6 - the future is here, just unevenly distributed

IRC log for #rosettacode, 2011-07-18

| Channels | #rosettacode index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
00:18 BenBE I'm pulling a first snapshot of the sources from the server to check the script for a first time.
00:18 BenBE I'll post a link soon.
00:21 BenBE http://benbe.home.omorphia.de:43815/geshi-misc/langdetect/learn.php --> Script to relearn all the files
00:21 BenBE http://benbe.home.omorphia.de:43815/geshi-misc/langdetect/detect.php --> Script that reads all examples and tries to guess the language.
00:27 BenBE Still downloading the snapshot, so stand by.
00:34 BenBE Looks like a REALLY good ADA coverage on RC ;-)
00:44 BenBE Currently best results show for ADA, Algol68, Autohotkey
00:44 BenBE Interestingly there are some languages that won't be detected at all ...
00:44 BenBE I.e. some languages always get misdetected as another language.
00:46 BenBE Examples are ABAP, ActionScript, Applescript, ASM, Autoit, AWK, Bash.
00:48 ttmrichter joined #rosettacode
00:49 Mathnerd314 left #rosettacode
00:55 TimToady ADA is not Another Darn Acronym, and should be spelled Ada
00:57 TimToady because it's really named after Miss Lovelace, who did not spell her name ADA. :)
00:58 TimToady either that, or you'll have to change to HASKELL to be consistent :)
00:59 BenBE :P
00:59 TimToady sorry, hot button with me...
00:59 BenBE I noticed ;-)
01:00 TimToady and you're also getting some backscatter from the topic of whether operators are functions in Perl 6 :)
01:00 BenBE Well, Haskell is yet another way for mathematicians to express themselves so that really nobody has a chance of understanding them :P
01:00 BenBE There was a discussion about this? Didn't notice :P
01:01 TimToady Haskell is easy to learn; I've done it five or six times already...
01:01 BenBE Why that often?
01:01 BenBE Did you forget it that easily again?
01:01 TimToady when you're as old as I am, you can do something five or six times without it being "often"
01:03 BenBE Any idea for how detecting a source code's language could be done?
01:03 TimToady feed it to the compiler and see if it barfs
01:04 BenBE The code I'm currently investigating uses Markov Chains to analyze the conditional probabilities of how likely it is for one character to follow another in the source code.
01:05 TimToady sounds like something that could be done by training a neural net, but then it couldn't tell you *why* it thought it was bf instead of j
01:05 TimToady definitely amenable to lexical and statistical analysis though
01:05 BenBE Yeah. BF and J are hard to tell apart ^^
01:05 TimToady how does Google Translate tell what language you're feeding it?
01:06 BenBE Basically Markov Chains plus some word probabilities I guess ...
01:06 BenBE So basically a simular starting point.
01:06 BenBE That's also why I'm currently collecting the "sorted" code samples from RC ;-)
01:07 BenBE I need some training data ;-)
01:07 Mathnerd314 joined #rosettacode
01:07 TimToady maybe you should use Watson :)
01:08 BenBE There's a small bird. Let's use the /really/ big gun ;-)
01:09 BenBE Or better yet the inter-galactic Death Star laser cannon ;-P
01:12 TimToady that would be Holmes, not Watson.  :)
01:12 BenBE My aim currently is to create a detection mechanism that I can train with source samples of the language to be detected OR the information available in the GeSHi language files.
01:13 TimToady or just write a parser for each language in Perl 6...
01:23 BenBE Well ... no :P
01:29 ttmrichter Has Perl 6 actually been finalized yet?  I kind of got bored and wandered off ten years ago.
01:30 mbishop haha
01:31 * ttmrichter was being serious.
01:32 ttmrichter I'm not up to date on Perl 6's latest and greatest.  My most recent "news" was that PUGS' lead developer quit.  That's pretty damned old news.
01:42 BenBE Well, I just heard that Linus Torvalds is waiting for Perl 6 to be released so he can port Hurd to it.
01:43 ttmrichter So that there's a platform for Duke Nukem For....  Oh, never mind.  That finally got released.  :D
02:42 mikemol TimToady, BenBE: Indeed. My first daughter will be named Ada. First son will be named Pascal. :)
02:44 mikemol BenBE: Perhaps there's something more fundamental than language that GeSHi could detect and highlight on.
02:45 sorear ttmrichter: are you trying to troll
02:46 sorear ttmrichter: but yes, the Perl 6 design is more or less finalized
02:46 sorear now the implementations are racing to catch up
02:46 mikemol ttmrichter: In all honesty and seriousness, the one rule I require on RC is participation in good faith. Please don't troll unless it's understood you're kidding. If you're not kidding, but have non-hostile intent, find a non-hostile way to make the query.
02:48 * mikemol didn't intend authenticating to NickServ to be quite that theatrical...
02:56 dagnyscott left #rosettacode
03:34 ttmrichter mikemol: I was not intending to troll.  I really am way out of the Perl 6 loop and was asking an honest question.  If, however, this was viewed as a troll, I apologize for not communicating intent properly.
03:50 BenBE left #rosettacode
04:42 Sky_Eagle^ joined #rosettacode
04:42 Sky_Eagle^ zup all?
04:42 sorear hi
04:43 mwn3d_phone Yo
04:43 Sky_Eagle^ been looking at the rosetta code site for the last couple of days.  Is anyone in a position to answer a couple of questions for me?
04:44 mwn3d_phone Ask and we can see what we can answer
04:44 mikemol Sky_Eagle^: What kind of questions?
04:44 mikemol (And depends on how many; I should be heading to sleep soon...)
04:44 Sky_Eagle^ well been reading some of the things on a couple of languages and wondering if there are some set criteria for "solutions"?  I'm specifically wondering:
04:45 Sky_Eagle^ 1) I see some tasks that are definite OS tasks.  Call the API, API does it, returns results.   Just wondering why there would be any interest in those specifically for what the purpose of the site seems to be?  Or are you looking for "Linux" vs. "Windows", etc?
04:45 mikemol (First, there's no fundamental rule for what constitutes a solution which could be said to apply evenly across the whole site. The site has some pretty aged content)
04:45 Sky_Eagle^ and
04:46 Sky_Eagle^ 2) I notice some tasks that are possible and known to accomplish in the language(s) in question, but can be done in Win32 API as well.   Which is acceptable?
04:46 Sky_Eagle^ (for example, you can do a MD5 algo for a string, but you can call the Win32 Crypto API and get the same thing)
04:47 mwn3d_phone "Idiomatic" is a word we like
04:47 mikemol Solutions to RC tasks need to [more or less] meet the requirements set forth in the task.
04:48 mikemol So if the task says that something has to do something entirely within the language, then an API call is out.
04:48 mikemol However, as I've told people I've explained the site to, "if that's how you do it in language X, that's how you do it in language X." If you would normally call out to a library to solve a problem, then, well, go ahead.
04:49 mikemol On the question of using APIs to do things which *can* be done in-language, and even things which already have solutions, I really don't have a problem with it.
04:50 ttmrichter left #rosettacode
04:50 Sky_Eagle^ yeah speaking of which, what about cross language solutions?  Like for MD5, I already have a solution using Delphi, but it has a couple of ASM pieces for optimization sake.  Those ASM pieces can be made into Delphi, but still...
04:50 mikemol While most of the site and the site's users revolves specifically around comparing languages, I realize that libraries face their own degrees of competition, and that there is a need for similar comparison. If you wanted to go through and demonstrate appropriate usage of the Win32 API for most of the tasks, that would seem reasonable to me.
04:50 mikemol Mixing languages. Hm.
04:51 Sky_Eagle^ yeah basically ASM made into functions, then called within the Delphi.
04:51 Sky_Eagle^ (but I'm thinking myself that it would probably be out, but that's just my inclination.  Doesn't hurt to ask though)
04:51 mwn3d_phone Id say OS and hardware-specific solutions could be added but should be separate from other examples in the language and notated
04:52 mikemol In those cases, the way I'd see the content being interesting and educational would be to show two examples; one using the asm, one in pure delphi. This demonstrates that sometimes you'd want to drop to asm, that this part of delphi in this problem might be appropriate for that, and that this is how you'd do it.
04:52 mwn3d_phone Then let the community talk about it on the talk pages
04:52 * mikemol nods
04:52 Sky_Eagle^ yeah the ASM opt is mainly to use an instruction or two that Delphi doesn't support innately.
04:52 mikemol Sky_Eagle^: Generally speaking, I look for ways to usefully incorporate content which doesn't strictly conform to pattern.
04:57 Sky_Eagle^ If there isn't any other comments, one last thing.  I'm not seeing how to submit things.  How is that done?
04:58 Sky_Eagle^ wiki edit the page, or is there another process in place?
04:58 mwn3d_phone Nope. Just the regular wiki way.
04:58 mwn3d_phone Keep things in alphabetical order by language name
04:59 Sky_Eagle^ okay thanks.
04:59 mwn3d_phone There should be some sort of help page somewhere for adding examples...can't remember the title atm
04:59 mwn3d_phone I'll try to find it quick and link it here
05:01 mwn3d_phone [[Rosetta Code:Solve a Task]]
05:01 fedaykin http://rosettacode.org/wiki/Rosetta_Code:Solve_a_Task
05:02 sorear I would say that if language A is designed to integrate closely with language B, such that A-users often reflexively solve specific tasks by using B, then it makes sense to allow solutions which use B, as long as they are idiomatic for A
05:02 sorear as an extreme example, nobody cares about C code examples using the printf interpreter for output
05:04 Mathnerd314 left #rosettacode
05:05 Sky_Eagle^ Alright, thanks again!
05:06 Sky_Eagle^ left #rosettacode
10:19 FireFly joined #rosettacode
10:19 FireFly left #rosettacode
10:19 FireFly joined #rosettacode
10:40 BenBE joined #rosettacode
10:40 BenBE mikemol There probably is And I'm positive someone will have a proper idea on how to do it ;-)
10:40 BenBE I'm just trying to test a random idea and see how well it works ...
13:05 BenBE Currently working on an update for VB.net ...
13:05 fedaykin "VB.NET"
13:08 BenBE IIRC the ".net" stands for "OHNOOOOOO!"
13:08 BenBE (Comprehensible for german dialect speakers only)
13:31 Hypftier There are worse languages out there, honestly.
13:34 dagnyscott joined #rosettacode
13:47 mikemol BenBE: God half-way through my thought before I had to deal with something else, and missed continuing it when I got back.
13:48 mikemol BenBE: I was thinking about things like Icon vs Unicon, or even C vs Pascal; I wonder if it's practicable to identify syntax classes of languages and keyword-meaning classes (i.e. printf will mean the same thing in just about every language, right?), and highlight based on those things.
13:50 mikemol While Google Analytics show site activity to be somewhat flat, I've been noticing a lot of new, active community memebrs in the last few weeks.
13:50 mikemol Pretty neat. :)
13:55 Hypftier GA might be pretty skewed on a technically-oriented site, though. It's blocked by default for plenty of users.
14:00 mwn3d_phone Well I, for one, welcome our new information-gathering overlords
14:00 mwn3d_phone Not blocked here
14:02 BenBE mikemol I think another basic step might be looking for keyword coverage ...
14:02 BenBE The more keywords match the more likely it is for the language to be the correct one.
14:04 BenBE The more coverage you have the more whitespace to text ratio you get when removing everything that was highlighted.
14:05 BenBE Leave out the symbols and you should get pretty close ...
14:05 BenBE But as I said: I'll be providing the dump of all the language samples so people can experiment with them ;-)
14:05 * mikemol nods
14:05 mikemol Just don't forget their license.
14:06 BenBE GFDL 1.2 ...
14:11 mikemol Hypftier: It's probably a bit skewed. I don't know how much, though; I'd have to compare it with chewings on access logs.
14:12 mikemol One of the nice things about GA, though, is it doesn't tend to include bot crawls, which are harder to filter using log analyzers I've played with. (Though I haven't really played with log analyzers in five or six years...)
14:49 dagnyscott left #rosettacode
14:53 kpreid left #rosettacode
15:01 dagnyscott joined #rosettacode
15:49 kpreid joined #rosettacode
16:05 kpreid left #rosettacode
16:06 kpreid joined #rosettacode
16:08 MigoMipo joined #rosettacode
16:41 kpreid left #rosettacode
16:42 kpreid joined #rosettacode
17:26 mikemol Ugh. Google Reader is falling back into the habit of duping items from RC's Recent Changes feed.
17:43 mwn3d_phone1 joined #rosettacode
17:47 mwn3d_phone left #rosettacode
19:03 mwn3d_phone1 is now known as mwn3d_phone
19:04 mwn3d_phone I imagine this applies analogously to programming languages and programming problems: http://www.cracked.com/article_18823_5-insane-ways-words-can-control-your-mind.html
19:04 fedaykin "5 Insane Ways Words Can Control Your Mind | Cracked.com" http://rldn.net/culJ
20:58 dagnyscott left #rosettacode
21:39 MigoMipo left #rosettacode
23:37 lambdabot left #rosettacode
23:37 ashleydev left #rosettacode
23:43 lambdabot joined #rosettacode
23:43 ashleydev joined #rosettacode
23:47 FireFly left #rosettacode

| Channels | #rosettacode index | Today | | Search | Google Search | Plain-Text | summary