00:00:00 --- log: started haskell/01.12.27 00:37:15 --- quit: julien ("Client Exiting") 00:41:37 --- join: julien (~julien@uu212-190-122-70.unknown.uunet.be) joined #haskell 01:24:54 --- join: shapr (~user@195.156.199.185) joined #haskell 02:24:26 --- join: antiClauz (~alife@sense-sea-MegaSub-2-56.oz.net) joined #haskell 02:24:54 --- nick: antiClauz -> dmiles 02:27:33 --- quit: shapr (Read error: 54 (Connection reset by peer)) 02:31:47 --- join: shapr (~user@195.156.199.185) joined #haskell 03:08:30 --- quit: juhp (Ping timeout: 180 seconds) 05:55:19 --- quit: shapr ("brb") 06:10:17 --- join: shapr (~user@195.156.199.185) joined #haskell 07:59:45 --- join: sebras (~sebras@) joined #haskell 08:00:11 hi 08:00:28 shapr: hi... rather quite channel, is it not... :-/ 08:00:34 nah 08:00:39 we talk 08:01:17 xbill: oh... it's like when you go into a room and everybody just get quiet 08:01:20 yah, just depends on what people have to say 08:01:52 sadly, I'm at work, banging my head against X.509 08:02:23 shapr: oh, well I didn't expect a haskell channel to be nosiy... 08:03:09 you could always ask what a Monad is ;) 08:03:29 shapr: "What the hell are Monads?"... 08:03:37 shapr: read that already... :-) 08:03:38 everyone asks that 08:04:27 I still don't really understand the answer. 08:04:44 shapr: but I could ask you what compilers you use. I've only tried hugs, but that doesnt compile natively... is GHC better? 08:05:04 shapr: well, not _better_ but is it feasible to use GHC for native compilation? 08:05:23 yes 08:05:29 -fvia-native 08:06:11 yup, I prefer to use GHC for everything 08:06:12 xbill: do you know of any papers/articles/whatever that compares ghc with other compiles (haskell, C, C++, what have you)? 08:06:34 comparing with non-Haskell compilers is apples and oranges 08:07:42 xbill: I know that haskell and non-haskell languages are dissimilar, but it would be interesting to know what speed one could get from a ghc compiled program... 08:07:49 lots :) 08:08:22 sebras: there are quite a few differences in the execution model, the operations happening will make a large difference 08:08:34 xbill: I read some article that stated that ocaml was faster than c++ today. that problably makes ocaml a feasible langauge to use, as opposed to hugs... 08:09:08 shapr: 08:09:15 shapr: url me! :-) 08:09:33 shapr: at least if you have any articles handy... 08:09:55 old news, ocaml generates higher-quality code than gcc by a longshot and C++ isfairly crappy 08:09:55 ghc is a fine compiler, i've not heard many complains that it is slow 08:11:08 * shapr gets tired of X.509 08:11:10 I'm goin home 08:11:11 it's usually not the compiler that makes programs slow, it's the programmer 08:11:24 yup, program design makes all the difference 08:11:27 xbill: I know its old news, but I recently discovered ocaml...! C++ is icky no matter what... :-/ 08:11:28 event-driven programs 08:11:45 allyou have to do is poke events fast enough 08:11:48 --- quit: shapr ("hometime") 08:11:55 dennisb: algorithms make the difference... 08:12:02 response time is the measure of performance 08:12:50 sebras: yes, and understanding the hardware and lower level languages even when you program in a high level language is also important 08:13:23 xbill: I used a theorem proving program with some GUI. all was developed in haskell, but I don'nt know what compiler the used. it was slow as h-ll... 08:13:54 theorem proving is usually slow by itself.. 08:13:59 don't know what to tell you 08:14:00 what prover did you use? 08:14:04 xbill: hence my distrust of native haskell compilers... haven't read anything by ghc though... 08:15:10 dennisb: it's called alga I think... some guy a Chalmers (Gothenburg, Sweden) Institute of technology though that making proofs was interesting... 08:15:32 yes, alfa 08:15:34 hbc was not exactly a speed demon 08:15:37 xbill: thats ok... :-) 08:16:19 Thomas Hallgren is the writer, it uses fudgets which is a widget set written fully in haskell 08:16:27 dennisb: alfa, close at least... the lack of speed might be because of that fact that I ran the program on a solaris machine... 08:17:16 well, it does a lot of work also, i've seen it on normal pc:s in linux and it's not that bad 08:17:27 dennisb: you obviously know the program.. I attended Thomas presentation of alfa in lund a few weeks back... 08:17:51 Oh, I thought he was in usa 08:18:14 dennisb: :-D well... lund isn't that far away... 08:18:24 sebras: I'm at chalmers so... 08:19:10 But I bet it would be snappa and nice if it used some standard gui toolkit 08:19:52 snappy even, which might not be a english word anyway :-) 08:19:56 dennisb: its compiled with ghc today? 08:20:13 No, I don't think so. 08:20:37 dennisb: wouldn't it be marvelous to have the outgoing traffic pass through ispell before sending it? >;-) 08:20:41 He talked about it once, that he had fixed it so he could compile it with ghc, but I don't think that is what he uses 08:21:03 what does he use? 08:21:09 I like these undelinings that ms word does.. 08:21:12 hbc 08:21:16 dennisb: then ghc might be wort a try... 08:21:58 ghc is fine, it's cool with the interpreter in ghc5 where you can mix compiled and interpreted code 08:22:12 you get something that works like hugs, but where parts can be compiled 08:22:25 (or everything of course) 08:23:35 dennisb: sounds nice... I might very well try that... afaicr it's linux-friendly right? 08:23:59 The fudgets are not like other widgets in the unixworld, I guess some of the slowness is there 08:24:08 sebras: ghc works great in linux 08:24:44 dennisb: hm... I wonder if debian has a precompiled (up-to-date!) binary... 08:24:54 yes, it should be 08:25:11 both debian and rpm-based binaries should be fine 08:25:28 I think ghc is even in the standard debian 08:25:39 (i run redhat so I don't really know) 08:26:00 just do some "apt-get ghc" or what it is you write... :-) 08:27:22 My only complaint with ghci (the interpreter) used to be that it uses a lot of memory, but now I have lots of memory so that's no problem. 08:27:24 apt-get install ghc5 08:27:36 ghci could be much faster 08:27:41 If you are on a small machine you probably want hugs 08:27:42 I think it leaks, too. 08:27:47 xbill: yes, it 08:27:56 's not perfect, but not bad 08:28:09 it's very good even 08:28:34 dennish 08:28:37 I agree, the last version works very well for me, the first ghci 5 was not so good 08:28:54 sebrash 08:29:06 dennisb: ghc5 does exist in woody... 08:29:33 yes, that's what i've heard (and xbill showed me the syntax to get it) 08:29:37 dennisb: how much memory are we talking about here? 192MB is enough? 08:29:59 If you have trouble with hugs, just let me know (wli@debian.org) 08:30:04 you don't want to have less then that 08:30:16 but it depends on your program 08:30:35 dennisb: naturally... 08:31:08 --- quit: pHa (Ping timeout: 180 seconds) 08:31:09 apt-cache show hugs 08:31:20 dennisb: oh, and btw. up at chalmers functional programming (and most non-imperative programming) is used in the undergraduate courses often right? 08:31:28 yes 08:31:38 we teach it as the first language 08:32:01 but they don't really get to see all the nice functional stuff 08:32:55 dennisb: but either way; it's probably more educational than the way we "learn" programming in lund... :-/ java is the language of choice... 08:33:12 well, we have java as the second laguage 08:33:43 xbill: hm... dec2001-release of hugs is out... 08:33:59 but it is a good first language, both for people who have not programmed at all and for people that think they are gods when it comes to c-programming 08:34:02 sebras: I'll upload a package soon then. 08:35:07 dennisb: I wonder why a course which is a survey of different programming languages and what tradeoffs one has to make when deciding what language to use to solve a problem one has to do.... 08:35:48 sebras: ? 08:35:50 dennisb: I think such a course would be of great importance. today students just see either functional or imperative programming... 08:36:24 dennisb: a couse that explains what different languages are suitable for and why and for what problems they are not suitable 08:37:49 well, it's hard to get that knowledge unless you know these languages, and once you know them you don't need a course to compare them (that's my view at least) 08:37:55 dennisb: we wrote some entertaining memory mind and tower of hanoi games in our functional programming couse. oh, and a music program, but no "real" applications... 08:38:19 music programs can be very real 08:38:35 that's the bad thing, students thing one can not write "real" program. 08:39:09 but a first course in java would probably not be real programs either 08:39:27 dennisb: true, at least to some extent. but for those that has knowledge _only_ of the languages taught a univ. this would be enlightening... 08:40:17 "Here's a simulator an assembler and a C compiler, write an OS" 08:40:24 xbill: I know of no other program that constructs chords from a melody and it's key... :-) 08:40:33 one of the standard laborations at chalmers is for the students to write the unix-command sort but in haskell 08:40:45 xbill: that would also be needed... 08:41:31 it's very simple in haskell, but also pretty real. Then they get to write it in Java in the next course. And I promise they all prefere the haskell solution :-) 08:41:41 dennisb: sounds interesting... maybe I ought to inform my teachers of this... 08:41:56 And when they try the speed the see that haskell is usually faster also 08:42:05 dennisb: at least if it uses a quicksort approach... :-) 08:42:22 but it depends on the programs of course 08:42:47 dennisb: nice... could you send me the url to the laboration? 08:42:48 but java have never been about speed 08:42:54 radixsort =) 08:43:27 xbill: there is always a couple of students that brings up radixsort, yes :-) 08:43:44 no need to bring it up, just write it 08:44:28 sebras: i'm not sure where the lab is, i'm not involved in this course. I just know about it 08:44:41 dennisb: oh, ok... 08:45:03 dennisb: got to go now. nice talking to you. 08:45:52 xbill: I'll get hugs when you put it up. and ghc5 also... I think I've been inspired... haskell...!, here I come... *runs into the sunset* 08:47:15 sebras: maybe they don't use it no more, I just found some other problems 08:47:40 writing a UNIX shell in Haskell could be interesting 08:47:58 some program that gives statistics of text files seems to be this years lab, both in haskell and java 08:48:03 something like wc I guess 08:48:44 xbill: didn't someone do that? 08:49:09 --- quit: sebras ("Leaving") 08:51:23 I wrote a small demon on haskell, it worked fine. It used the double fork trick just like in C and it worked just like in C but guaranteed without buffer overruns and such bugs 08:52:01 I wrote something like fetchmail in Haskell once. 08:52:13 somehow managed to write dirty Haskell =) 08:52:20 xbill: yes, I know. haskell works fine for these kind of things 08:52:39 not fine 08:52:42 better 08:52:46 some parts looks just like in c, but there is nothing wrong with that 08:54:26 gtk+hs is something I really like, to bad that gtk+ is such a small part of what you usually wants.. 08:54:36 --- quit: julien ("Client Exiting") 08:54:52 just "small" things like gettext needs to be made bindings for.. 10:20:39 --- quit: xbill (Ping timeout: 180 seconds) 10:36:08 --- join: xbill (wli@DOMINIA.MIT.EDU) joined #haskell 10:41:25 --- join: pHa (sjh@aircomm-1blk-220-26.air.on.ca) joined #haskell 13:39:22 --- join: Wittekind (~Autosoph@A9883.pppool.de) joined #haskell 13:42:20 --- quit: xbill (Ping timeout: 180 seconds) 13:59:01 --- join: xbill (wli@DOMINIA.MIT.EDU) joined #haskell 19:20:41 --- quit: Wittekind ("[x]chat") 23:59:59 --- log: ended haskell/01.12.27