←2009-11-11 2009-11-12 2009-11-13β†’ ↑2009 ↑all
00:02:34 -!- FireFly has quit ("Later").
00:11:22 -!- Sgeo has joined.
00:14:32 -!- boily has joined.
00:16:57 -!- boily has quit ("leaving").
00:25:07 -!- Azstal has joined.
00:25:43 -!- Asztal has quit (Read error: 60 (Operation timed out)).
00:25:48 -!- Azstal has changed nick to Asztal.
00:43:18 -!- Gracenotes has joined.
00:52:01 -!- fax has quit ("Leaving").
00:55:27 -!- oerjan has quit ("Good night").
01:26:35 -!- coppro has joined.
01:45:08 -!- Slereah has quit (Read error: 110 (Connection timed out)).
01:46:04 -!- Gracenotes has quit (Remote closed the connection).
01:46:13 -!- immibis has joined.
02:23:30 -!- augur has joined.
02:44:17 -!- Sgeo has quit (Read error: 54 (Connection reset by peer)).
02:57:22 -!- Sgeo has joined.
03:08:20 -!- Gracenotes has joined.
03:21:26 -!- Gracenotes has quit (Remote closed the connection).
03:37:31 -!- Gracenotes has joined.
03:48:27 -!- MizardX has quit (Read error: 145 (Connection timed out)).
04:07:23 -!- Rugxulo has joined.
04:07:53 <Rugxulo> Deewiant: the only place I can find the Funge '97 spec is here:
04:07:54 <Rugxulo> http://web.archive.org/web/20000903032408/www.mines.edu/students/b/bolmstea/mtfi/f97.html
04:08:35 <Rugxulo> and of all his files, only the mtfi121x.zip one is mirrored (8086 DOS version of his Funge '93/'96/'97 interpreter)
04:08:47 <Rugxulo> but it also compiles with DJGPP, so Linux etc. should work too
04:09:31 <Rugxulo> mostly I'm annoyed that one '93 example I wrote apparently doesn't work correctly in '98 (CCBI or FBBI)
04:10:11 <Rugxulo> apparently the whole "look for blank space and reflect back assuming its end of funge space" doesn't take account of string mode very well
04:10:23 <Rugxulo> so I consider that a bug
04:10:37 -!- puzzlet_ has joined.
04:11:30 <Rugxulo> (more annoying in that CCBI has no true '93 mode although FBBI and MTFI both have one)
04:11:40 <Rugxulo> ah, whatever
04:17:56 -!- MizardX has joined.
04:22:28 -!- puzzlet has quit (Read error: 110 (Connection timed out)).
04:24:16 -!- puzzlet_ has changed nick to puzzlet.
04:25:26 -!- Rugxulo has quit ("gxis revido").
04:31:16 -!- coppro has quit (Read error: 60 (Operation timed out)).
05:03:22 -!- immibis has quit (Remote closed the connection).
05:49:25 -!- Slereah has joined.
06:13:28 -!- Sgeo has quit (Read error: 104 (Connection reset by peer)).
07:15:08 <pikhq> Hmm. Bing has decided to suck more: results to come from W|A.
07:59:59 -!- clog has quit (ended).
08:00:00 -!- clog has joined.
08:10:46 -!- puzzlet has quit (Remote closed the connection).
08:10:52 -!- puzzlet has joined.
08:37:14 -!- {}_ has quit (Read error: 60 (Operation timed out)).
08:40:38 -!- AnMaster has joined.
08:45:56 -!- Asztal has quit (Read error: 110 (Connection timed out)).
10:03:51 -!- FireFly has joined.
10:17:23 -!- BeholdMyGlory has joined.
10:40:00 -!- Pthing has quit (Remote closed the connection).
11:45:20 -!- oerjan has joined.
11:47:51 <oerjan> <Rugxulo> apparently the whole "look for blank space and reflect back assuming its end of funge space" doesn't take account of string mode very well
11:49:50 <oerjan> well once you decide not to have a fixed fungespace size, it is hard to treat one end-of-funge-space space differently from another
11:54:19 <oerjan> unless you introduced a "nothing there" character, but that could lead to some awful debugging since most editors show that identical to space by default...
11:54:30 <oerjan> *identically
12:03:30 <oerjan> AnMaster: iwc
12:05:53 <fizzie> Funkiest thing evar: http://detexify.kirelabs.org/classify.html (shamelessly stolen from another #channel).
12:06:09 -!- ais523 has joined.
12:06:27 <fizzie> Doesn't seem to work so terribly well, but I like the idea of it.
12:07:37 <oerjan> AnMaster: also D&D. they're going to be _so_ happy when the real GM gets back :D
12:47:05 -!- ais523 has quit (Read error: 104 (Connection reset by peer)).
12:47:12 -!- ais523 has joined.
12:50:08 -!- oerjan has quit ("leaving").
13:22:32 -!- FireFly has quit ("Later").
14:33:42 <AnMaster> hi ais523
14:33:47 <ais523> hi
15:21:29 <AnMaster> hm where is ehird when you need him
15:21:54 -!- Asztal has joined.
15:42:55 * AnMaster wonders what on earth "italic correction" is in TeX.
15:43:12 <AnMaster> I'm getting a warning about it from chktex
16:16:39 -!- puzzlet has quit (Remote closed the connection).
16:16:44 -!- puzzlet has joined.
16:19:06 -!- GregorR has joined.
16:21:31 -!- Gregor has quit (Read error: 110 (Connection timed out)).
16:33:14 -!- mycroftiv has joined.
18:10:32 <fizzie> AnMaster: It's the small bit of spacing you need when you have an italic glyph followed by a roman glyph.
18:16:19 -!- FireFly[DS] has joined.
18:20:54 -!- kar8nga has joined.
18:21:11 -!- FireyFly[DS] has joined.
18:21:30 -!- FireFly[DS] has quit (Connection reset by peer).
18:21:41 -!- FireyFly[DS] has changed nick to FireFly[DS].
18:22:45 <AnMaster> fizzie, well. It is followed by a space. So I'm not sure if this even makes sense
18:22:50 <AnMaster> hm
18:23:12 <AnMaster> and, wouldn't LyX handle that automatically hm
18:24:15 <AnMaster> fizzie, it seems to be after an \emph{foo}
18:26:36 <fizzie> \emph is often in italics. I wouldn't know about automatics; you can add a manual "\/" command inside the emph to maybe placate chktex, though I'm not sure if that works.
18:35:51 -!- fax has joined.
18:39:34 <AnMaster> fizzie, it doesn't actually look bad though. And strangely enough chktex never warned before when I used \emph (and yes it produced italics then too)
18:41:16 <fizzie> Did you have something immediately after the \emph{} with no spacing? Since that's usually when it is needed.
18:41:40 <fizzie> "It" referring to the extra spacing.
18:42:50 <augur> ive decided that just as there is Occams Razor
18:42:58 <augur> there is a corollary
18:43:01 <augur> Augurs Razor
18:43:03 <augur> that says
18:43:33 <augur> given two equally adequate and equally simple theories, the one that is weirdest is best
18:44:24 <fizzie> It's not a problem with all letters; and there might even be some automatics nowadays, since the reasonably worst-case "\emph{f}k" and "\textit{f}k" render just fine here with a noticeable gap there.
18:45:40 <fizzie> But! The raw-TeX "{\it f}k" doesn't render well at all. Strange that chktex would complain about the working variants, though.
18:52:16 -!- FireFly has joined.
18:55:27 -!- FireFly[DS] has quit ("ClIRC - IRC client for Nintendo DS").
18:56:06 -!- cal153 has quit.
19:03:16 -!- cal153 has joined.
19:16:27 -!- puzzlet has quit (Remote closed the connection).
19:16:31 -!- puzzlet has joined.
19:24:39 -!- KingOfKarlsruhe has joined.
19:38:15 -!- oerjan has joined.
19:40:22 <oerjan> <augur> given two equally adequate and equally simple theories, the one that is weirdest is best
19:42:00 <oerjan> i'd assume that's implied in the famous bohr quote: "We are all agreed that your theory is crazy. The question that divides us is whether it is crazy enough to have a chance of being correct."
19:45:55 -!- sebbu2 has joined.
19:56:10 -!- iamcal has joined.
19:57:04 -!- augur has quit (Read error: 110 (Connection timed out)).
19:58:37 <AnMaster> oerjan, iwc. but read it without being connected to bouncer
19:58:42 <AnMaster> which means I count as not here
19:59:05 <AnMaster> oerjan, so I can just say iwc when you are not here and it will count too!
19:59:31 <oerjan> well i already did say iwc when you were apparently not here
19:59:51 <AnMaster> oerjan, yes that is the point. I always wait until you are here
20:00:30 <AnMaster> I see no reason to continue doing so
20:00:51 <oerjan> i always try to check if you are idle, but maybe 3 hours was a bit too large
20:01:14 <AnMaster> oerjan, 3 hours due to client loosing connection and auto-reconnecting even
20:01:22 <AnMaster> real idle time was closer to 12 hours
20:01:29 <oerjan> oh
20:01:52 -!- Rugxulo has joined.
20:02:19 <AnMaster> oerjan, you could easily have seen that in logs. That I reconnected about 3 hours before that, that is.
20:02:32 -!- sebbu has quit (Connection timed out).
20:02:35 <Rugxulo> <oerjan> well once you decide not to have a fixed fungespace size, it is hard to treat one end-of-funge-space space differently from another
20:02:35 <oerjan> hm true. i didn't think of that.
20:02:48 <Rugxulo> oerjan, it should know when string mode begins and ends, though, right??
20:03:05 <AnMaster> oerjan, wouldn't be exactly, client would spend a few minutes after reconnect rejoining channels (due to being in so many here on freenode)
20:03:11 <oerjan> Rugxulo: i was assuming you were talking about a string mode wrapping around the end?
20:03:17 <Rugxulo> no
20:03:32 <AnMaster> Rugxulo, what are you doing in before?
20:03:36 * Rugxulo wishes he had a small paste-able example ...
20:03:38 <oerjan> hm then i don't understand
20:03:54 <AnMaster> Rugxulo, some fungoid?
20:04:16 <Rugxulo> I'm talking about Befunge98 not accepting Befunge93 correctly
20:04:32 <Rugxulo> I think the whitespace inside quotes " x " (string mode) is confusing it
20:04:36 <AnMaster> Rugxulo, uh well yeah it isn't completely compatible. It is well known. The SGM spaces for example
20:04:43 <AnMaster> SGML*
20:04:49 <AnMaster> Rugxulo, sounds like SGML space rule
20:04:52 <AnMaster> see spec
20:05:13 <AnMaster> http://catseye.tc/projects/funge98/doc/funge98.html#Strings
20:05:14 <AnMaster> to be specific
20:05:43 <AnMaster> Rugxulo, further, if a befunge93 program depends on something like x just reflecting it won't work in 98 eithger
20:05:45 <AnMaster> either*
20:06:09 <Rugxulo> <v" x "
20:06:11 <Rugxulo> >,,,,,@
20:06:17 <AnMaster> the reason for SGML spaces is probably due to it wrapping
20:06:33 <oerjan> oh then that is something i didn't know about.
20:06:53 <ais523> 98 isn't identical to 93
20:06:55 <AnMaster> Rugxulo, what exactly do you expect that to do? I expect it to print (C string notation here): " x \0\0"
20:07:07 <Rugxulo> that's two spaces on each side
20:07:15 <AnMaster> Rugxulo, did you read the link above?
20:07:16 <ais523> multiple spaces count as one space in 98
20:07:18 <AnMaster> if not, do it
20:07:19 <AnMaster> please
20:07:21 <Rugxulo> but yeah, I'm not really familiar with Befunge98 (kinda confusing, honestly)
20:07:23 <AnMaster> it will explain it
20:07:26 <Rugxulo> I'm reading it now
20:07:34 <ais523> basically, because otherwise you'd get a near-infinite number of spaces if a string wrapped
20:07:43 <AnMaster> ais523, "near"?
20:07:49 <Rugxulo> but the string doesn't wrap, just has some whitespace in it
20:07:52 <AnMaster> ais523, at least in efunge you would get infinite
20:08:02 <Rugxulo> but I guess since it could in theory ...
20:08:06 <ais523> AnMaster: arguably, you'd get 2^32 for a 32-bit funge
20:08:21 <AnMaster> ais523, yes but efunge is bignum
20:08:46 <AnMaster> so you would get infinite in theory (of course in practise it would run out of memory before that)
20:09:04 <AnMaster> Rugxulo, yes it applies to all strings
20:09:21 <Rugxulo> is string wrapping really that useful or common??
20:09:34 <Rugxulo> I know, silly question, but still ... ;-)
20:09:46 <AnMaster> Rugxulo, I have done it I'm pretty sure. Can't remember when though
20:09:57 <AnMaster> probably when something didn't fit
20:10:07 <Rugxulo> you mean in 93 or 98?
20:10:09 <AnMaster> Rugxulo, I'm quite sure that mycology tests it.
20:10:12 <AnMaster> Rugxulo, in 98 of course
20:10:18 <AnMaster> I don't code in 93 at all
20:10:19 <Rugxulo> so how didn't it fit?
20:10:41 <Rugxulo> I mean, if 98 isn't limited like 93, then why need to wrap strings?
20:11:01 <AnMaster> Rugxulo, didn't fit due to surrounding code. I try to code compact
20:11:36 <AnMaster> Rugxulo, and even when not wrapping it is useful
20:11:51 <AnMaster> say you need code to cross, but don't have the space for a # in that direction
20:12:05 <AnMaster> and there a space one space off, and you can extend that line a bit
20:12:11 <AnMaster> well then you have the solution
20:12:29 * Rugxulo still wonders what Befunge98 examples exist ...
20:12:31 <AnMaster> alternative would be to rewrite parts of the code of course
20:12:49 <AnMaster> Rugxulo, it is more for writing application programs in? Large programs I mean
20:12:52 <AnMaster> ;P
20:12:53 <Rugxulo> some would say "rewrite rewrite rewrite ..."
20:13:08 <Rugxulo> BTW, I have never seen anything about Befunge '96 or '97
20:13:24 <Rugxulo> late last night I posted a (WayBack) link to the '97 spec, though (Ben Olmstead's old page)
20:13:34 <Rugxulo> e.g. o for nop (instead of z as in 98)
20:13:39 <AnMaster> Rugxulo, I would prefer "keep patching, until it looks so improbable and messy that not even the original author can read it"
20:13:39 <Rugxulo> kinda weird
20:13:59 <AnMaster> <Rugxulo> BTW, I have never seen anything about Befunge '96 or '97 <-- I have seen a few recovered snippets
20:14:08 * AnMaster thinks he may have some stuff around somewhere
20:14:12 <Rugxulo> I mean, I saw *one* example program in '97 (GPL, too, heh)
20:14:22 -!- cal153 has quit (Read error: 113 (No route to host)).
20:14:46 <Rugxulo> Ben's MTFI ran on Linux or DOS (16-bit or 32-bit)
20:14:47 <AnMaster> ah yes there it is
20:14:54 <Rugxulo> but he included no examples
20:14:56 <AnMaster> Rugxulo, an 1.9 MB mailing list dump
20:14:58 <AnMaster> interested?
20:15:03 <Rugxulo> uh ...
20:15:07 <AnMaster> it conains 97 stuff
20:15:12 <AnMaster> spec drafts and such
20:15:13 <Rugxulo> sure, why not? :-)
20:15:35 <Rugxulo> I've got the spec (from WayBack), just no examples (besides that one signature that draws an ellipse)
20:15:38 <AnMaster> Omploaded 'bef_maillist_0_520.txt' to http://omploader.org/vMnI2dg
20:15:39 <AnMaster> there
20:15:43 <Rugxulo> thanks ;-)
20:15:54 <AnMaster> Rugxulo, also you meant 96 not 98 right?
20:16:04 <AnMaster> Rugxulo, link to the spec?
20:16:20 <Rugxulo> yes, apparently there was a '96 too
20:16:25 <Rugxulo> no idea where '96 spec is
20:16:29 <AnMaster> Rugxulo, never heard of 98
20:16:36 <AnMaster> Rugxulo, I meant to 97
20:17:06 <Rugxulo> well, Pressey's latest site doesn't have it, so that doesn't help
20:17:07 <Rugxulo> http://web.archive.org/web/20000903032408/www.mines.edu/students/b/bolmstea/mtfi/f97.html
20:17:55 <Rugxulo> here's the only '97 example I know of: http://www.bedroomlan.org/hacks/signature-befunge
20:17:57 <Asztal> http://catseye.tc/projects/funge98/library/TOYS.html does mention it though.
20:19:03 <Rugxulo> ... barely
20:19:09 <AnMaster> Rugxulo, contact that person and ask for what interpreter he used?
20:19:30 <Rugxulo> well, MTFI works, he probably used that
20:19:59 <AnMaster> MTFI?
20:20:21 <Rugxulo> hold on, here's '96 spec (I think): http://web.archive.org/web/20001008142309/www.mines.edu/students/b/bolmstea/mtfi/96.html
20:20:34 <Rugxulo> yeah, MTFI is the '93 / '96 / '97 interpreter for Linux and DOS by Ben Olmstead
20:22:08 <Rugxulo> only the WayBack link to the 8086 version still works, but the source is included, also compiles under DJGPP, so it should work on Linux too (since he had an explicit "Unix" port too)
20:22:12 <Rugxulo> http://web.archive.org/web/20000831062334/http://www.mines.edu/students/b/bolmstea/mtfi/mtfi121x.zip
20:22:27 <AnMaster> mhm
20:22:47 <Rugxulo> crap, the Unix link works on this version of the page!
20:22:49 <Rugxulo> http://web.archive.org/web/20000831062334/http://www.mines.edu/students/b/bolmstea/mtfi/mtfi-1.21.tar.gz
20:22:59 <AnMaster> why "crap"
20:23:00 <AnMaster> ?
20:23:00 <Rugxulo> too bad no MCBC :-(
20:23:06 <AnMaster> MCBC?
20:23:07 <Rugxulo> crap as in "wow, oops, surprise"
20:23:14 <Rugxulo> MCBC = 386 NASM Befunge compiler
20:23:18 <AnMaster> Rugxulo, as for 98 "examples"? what exactly do you mean by that?
20:23:25 <AnMaster> ^source
20:23:25 <fungot> http://git.zem.fi/fungot/blob/HEAD:/fungot.b98
20:23:31 <AnMaster> there you have one?
20:23:52 <AnMaster> Rugxulo, or not happy with that?
20:23:59 <Rugxulo> is that the IRC bot?
20:24:02 <AnMaster> Rugxulo, yes
20:24:08 <Rugxulo> then yes, that counts ;-)
20:24:12 <AnMaster> Rugxulo, that is the code for fungot itself
20:24:12 <fungot> AnMaster: i was just wondering if any current scheme system today. more details, pls.
20:24:19 * Rugxulo just thought '98 seemed overrated for something no one codes in ...
20:24:20 <AnMaster> Rugxulo, there is mycology too. As you probably know
20:24:44 <Rugxulo> yeah, but a test suite doesn't do much of anything (no offense), I wanted something interesting ;-)
20:24:54 <AnMaster> Rugxulo, people do code in it. Just generally it is complex stuff compared to 93 ones. Thus it isn't really "examples" but "applications"
20:25:38 <AnMaster> Rugxulo, anyway that code file won't work on it's own. It uses a bootloader. Which sets up nick, server and such, then loads the main file
20:25:40 <Rugxulo> okay, I just never saw many (well, any)
20:25:53 <AnMaster> http://git.zem.fi/fungot/blob/HEAD:/fungot-load-freenode.b98
20:25:54 <fungot> AnMaster: what's the big difference is in the source fnord it isn't in the index. once this is false, another is a sgi indy ( mips, running irix) and a
20:25:58 -!- KingOfKarlsruhe has quit (Remote closed the connection).
20:26:04 <AnMaster> ^style
20:26:04 <fungot> Available: agora alice c64 ct darwin discworld europarl ff7 fisher ic irc* jargon lovecraft nethack pa speeches ss wp youtube
20:26:14 <ais523> ^style youtube
20:26:14 <fungot> Selected style: youtube (Some YouTube comments)
20:26:15 <AnMaster> Rugxulo, you weren't aware of it being in befunge?
20:26:18 <AnMaster> -_-
20:26:29 <ais523> fungot: I doubt what you spam will look much different from the original, here
20:26:29 <fungot> ais523: wie alle auf einmal opern singen wollen... dreck. who really knows aviation will tell you something, it's dark blond
20:26:38 <AnMaster> heh
20:26:44 <Rugxulo> AnMaster, I probably fun^H^H^Hforgot
20:27:06 <Rugxulo> fungot, what no fjords?
20:27:07 <fungot> Rugxulo: lol that was really funny, holding up random stuff made of gold. like time magazine, airliners dot net and at wikipedia? i dont remember seeing any angels in the spotlight, and didn't react fast enough to pull up
20:27:15 <AnMaster> Rugxulo, what?
20:27:22 <Rugxulo> fungot, fnord yourself
20:27:23 <fungot> Rugxulo: ok, farinelli... i got the hell
20:27:32 <AnMaster> fungot, ...
20:27:32 <fungot> AnMaster: agreed, i know the song? :p), there is terror and pain caused by pilot error, didn't know that you've been to france.
20:27:53 <AnMaster> arliners running on .NET code sounds scary
20:27:57 <AnMaster> airliners*
20:28:36 <AnMaster> fungot, why so much about aviation?
20:28:36 <fungot> AnMaster: she makes me jizz in my pants. from reading these 488 hideous comments, heres the conclusion of all the spectators. i liked the scottish budweiser commercial the best
20:28:47 <AnMaster> hehe
20:29:26 <Rugxulo> airliners on .NET? they do have British nuclear subs running Windows though (supposedly)
20:31:06 -!- puzzlet has quit (Read error: 60 (Operation timed out)).
20:32:34 <AnMaster> interesting (from the 97 spec):
20:32:35 <AnMaster> In a more ideal universe, ~ may suspend only it's own the thread. Suspending the entire program is the usual and entirely forgivable real behaviour.
20:32:36 -!- puzzlet has joined.
20:34:30 <Rugxulo> the latest version of Ben's copy of the '96 spec is here:
20:34:38 <Rugxulo> (not sure if it differs from what I listed before)
20:34:40 <Rugxulo> http://web.archive.org/web/20040221081031/www.mines.edu/students/b/bolmstea/mtfi/96.html
20:35:10 <AnMaster> it does
20:35:14 <AnMaster> say diff
20:35:26 <AnMaster> -<html><head>
20:35:26 <AnMaster> -<meta http-equiv="content-type" content="text/html; charset=ISO-8859-1"></head><body><center><h1>Befunge-96</h1></center>
20:35:26 <AnMaster> +<center><h1>Befunge-96</h1></center>
20:35:28 <AnMaster> for example
20:35:30 <AnMaster> very strange
20:39:39 <Deewiant> Rugxulo: Good finds, by the way. I've spent many a time looking for those without success.
20:42:02 <Rugxulo> "If you wrap in stringmode, the result is undefined." -- MTFI's "notes" file
20:45:18 <Deewiant> I don't think that's ever been true as far as the spec is concerned. It may have made sense in '96, with a bigger playfield but without SGML spaces, and the poor 16-bit implementations.
20:46:05 <Rugxulo> possibly so
20:46:39 <Rugxulo> BTW, MTFI "Unix" and "8086" both seem identical except for makefiles and readmes
20:47:26 <Rugxulo> and I can't say for sure, but I'd bet it's GPL because he uses GNU's getopt_long
20:48:08 <Rugxulo> (hmmm, actually that part is LGPL here, but if you get it from BinUtils, for example, it's GPL)
20:48:26 <SimonRC> oh dear, a case of the viral GPL
20:48:35 <Rugxulo> *BSD has its own getopt_long these days
20:49:22 <Rugxulo> BTW, what's the deal with MS being braindead about writing temporary files to C:\
20:49:45 <Rugxulo> the 8086 MS C .EXE tries doing it (forbidden under Vista default user account), MSVCRT tries doing it, etc.
20:49:48 <Rugxulo> ... annoying
20:50:34 <Rugxulo> you'd think they'd test their own stuff, but obviously not as well as we'd like!
20:50:44 <Rugxulo> (rant off)
20:55:41 * Rugxulo reading '96 mailing list dump
21:09:56 <AnMaster> Rugxulo, uh. not allowing writing temp files in c:\ seems pretty sane
21:10:11 <Rugxulo> yes, but they could virtualize it (but don't)
21:10:18 <AnMaster> Rugxulo, why should they?
21:10:22 <SimonRC> but then things would break silently
21:10:30 <Rugxulo> because MS C and MSVCRT.DLL are the ones that expect to be able to do so !!!!
21:10:40 <SimonRC> why don't you ask Raymond Chen? That is his kind of question
21:10:42 <AnMaster> it makes no fucking sense to allow that outside some sort of /tmp or ~/tmp
21:10:46 <Rugxulo> they don't even workaround their own bugs! (argh)
21:10:46 <SimonRC> he might alreadhave covered it
21:10:48 <AnMaster> well not those obviously
21:10:55 <AnMaster> but equivalents
21:11:02 <AnMaster> Rugxulo, is it a bug?
21:11:17 <Rugxulo> MSVCRT.DLL 's tmpfile() assumes C:\ is writable
21:11:20 <AnMaster> <Rugxulo> because MS C and MSVCRT.DLL are the ones that expect to be able to do so !!!! <-- MSVCRT what version?
21:11:30 <AnMaster> 80? 90?
21:11:36 <Rugxulo> I think from MSVC 6, the one MinGW usees
21:11:38 <Rugxulo> *uses
21:11:45 <AnMaster> Rugxulo, well that's outdated as hell afaik
21:11:55 <Rugxulo> yes, but MinGW still uses it, so blame them not me :-P
21:16:02 -!- augur has joined.
21:16:06 <AnMaster> Rugxulo, well lets put it another way: I don't care a "shit" about windows breaking or not breaking.
21:16:14 <AnMaster> actually that is not true
21:16:28 <AnMaster> I would prefer windows to fail, rather than work
21:16:53 <Rugxulo> it's just sad when something works worse than before instead of better ... that's all
21:16:56 <AnMaster> there are way better operating systems out there. and yes I did try windows7
21:17:17 <Rugxulo> there are too many OSes, languages, cpus, etc.
21:17:24 <AnMaster> Rugxulo, sadly, this one isn't likely to affect the majority of windows users. Would have been a lot better if it did
21:17:25 <Rugxulo> no one-size-fits-all
21:17:55 <Rugxulo> no, it wouldn't matter then either, Windows bugfixing isn't a democracy, meritocracy, or anything more than "random"
21:18:05 <AnMaster> Rugxulo, mac OS X for those who just want stuff to work. Linux for those that want free software or being able to easily modify major parts of the system
21:18:09 <AnMaster> freebsd too there
21:18:43 <AnMaster> at least OS X has a decent shell
21:18:52 <Rugxulo> ?
21:19:10 <Rugxulo> Bash, you mean?
21:20:30 <AnMaster> Rugxulo, yes. cmd.exe is just unusable for anything serious. And windows "powershell" seems like a joke.
21:20:46 <AnMaster> sure powershell can do more, but it seemed rather awkward to me.
21:21:02 <Rugxulo> Total Command? (TCC)
21:21:16 <Rugxulo> Powershell isn't even included in Vista by default (despite all the hype)
21:21:27 <AnMaster> isn't it in 7 iirc?
21:21:31 <Rugxulo> OS X used to include tcsh until 10.2 or so
21:21:36 <Rugxulo> yes, in 7, from what I heard
21:21:44 <AnMaster> Rugxulo, I'm aware of os x used to use tcsh
21:21:46 * Rugxulo didn't feel 7 was worth upgrading to
21:21:47 <AnMaster> of that*
21:22:11 <Rugxulo> definitely not for $119, ugh
21:22:18 <AnMaster> never heard of TCC. Does it allow the wide flexibility of the shell on a *nix system? With the huge number of programs that you can use to do lots of things
21:22:45 <Rugxulo> I'm not sure exactly how to classify it as I've never used it, but it's by the guy who made 4DOS and 4NT
21:22:48 <AnMaster> could you just use that instead of the GUI?
21:22:54 <AnMaster> on linux that would be feasible
21:23:01 <AnMaster> in fact I administrated lots of server over ssh
21:23:05 <AnMaster> with no X
21:23:18 <Rugxulo> no, you can't remove the GUI from Windows (that I know of, anyways)
21:23:31 <Rugxulo> MinWin was just a tech demo, I think
21:23:45 <AnMaster> Rugxulo, well, I didn't require that. I just required it to not require you to use the GUI ever
21:24:09 <Rugxulo> TCC? well, it is a GUI app I think (with tabs, etc.), but console-ish
21:24:14 <Rugxulo> I'm not entirely sure, never used it
21:24:15 <AnMaster> ugh
21:24:17 <Rugxulo> at least TCC/LE is free
21:24:24 <Rugxulo> -ware
21:24:33 <AnMaster> Rugxulo, so completely irrelevant then to my question?
21:25:11 <AnMaster> my point is that with *nix you can use the GUI but don't *need* it. On windows so far it seems you will need it for some things
21:25:12 <Rugxulo> oops, not Total Command but Take Command
21:25:28 <Rugxulo> http://www.jpsoft.com/tccledes.htm
21:25:43 <Rugxulo> "formerly known as 4NT" ... okay
21:26:41 <Rugxulo> yeah, seems pretty crippled except in the high end (non-freeware) versions
21:26:48 <pikhq> Rugxulo: Well, Windows Server does at least offer a very barebones install.
21:26:50 <Rugxulo> oh well, just saying, it's better than CMD at least
21:27:13 <pikhq> It kinda *has* a GUI, but it only has cmd.exe running in it.
21:27:34 <Rugxulo> last updated two days ago, too
21:27:35 <pikhq> Pity they've not made it without the GUI, though.
21:27:39 <AnMaster> pikhq, so how usable is it to administrate it from there?
21:28:01 <pikhq> AnMaster: Decently usable, actually -- the Microsoft server programs, at least, got CLIs.
21:28:21 <AnMaster> pikhq, okay. But that is cmd.exe you are talking about
21:28:27 <pikhq> Perhaps a bit nicer if they had a better shell, though.
21:28:35 <pikhq> (you could, say, use Mingw bash)
21:28:47 <AnMaster> pikhq, how would you do stuff like loop over all files in the current directory, executing some command
21:28:53 <AnMaster> or something like:
21:29:29 <pikhq> Most sanely? Probably a .bat file.
21:29:46 <fax> is this #erotic?
21:29:47 <pikhq> (goto, I'd imagine, is hard to use on a prompt)
21:29:48 <AnMaster> find . -name '*.html' -exec chmod o-w {} +
21:29:51 <Rugxulo> for %a in (*.txt) do sed -i -e "s/foo/bar/" %a
21:29:54 <AnMaster> pikhq, do that in one as short command :P
21:29:59 <AnMaster> bat file is cheating
21:30:06 <Rugxulo> not a .BAT (else I'd have to use %%)
21:30:07 <AnMaster> has to be done on a single command line
21:30:11 <Rugxulo> straight from prompt
21:30:12 <pikhq> AnMaster: bash -c find . -name '*.html' -exec chmod o-w {} +
21:30:13 <pikhq> :P
21:30:25 <AnMaster> Rugxulo, sed isn't cmd.exe
21:30:26 <Rugxulo> C:\TEMP> for %a in (*.txt) do sed -i -e "s/foo/bar/" %a
21:30:34 <Rugxulo> so? find isn't Bash either
21:30:36 <AnMaster> pikhq, I think you need some quotes there. not 100% sure
21:30:51 <pikhq> AnMaster: Probably.
21:30:57 <Rugxulo> fax: no
21:31:01 <AnMaster> Rugxulo, sorry, wrong term. "Not included by default installation of the OS"
21:31:01 <pikhq> The first thing I do on any Windows box is install bash.
21:31:03 <AnMaster> which find is
21:31:11 <AnMaster> on all *nix I have seen
21:31:16 <fax> this whole time I was just waiting for somethig to happen :(
21:31:18 <Rugxulo> fax: not unless this excites you
21:31:19 <Rugxulo> 12481> #+?\# _.@
21:31:21 <pikhq> And yeah, find is POSIX.
21:31:24 <fax> a bit...
21:31:34 -!- Pthing has joined.
21:31:53 <AnMaster> pikhq, so is sed btw
21:32:08 <Rugxulo> POSIX sed doesn't support "-i" anyways
21:32:26 <AnMaster> Rugxulo, true, don't use it a lot. easy to work around anyway. GNU sed has it
21:32:35 <Rugxulo> since 4.x, yes
21:32:35 <AnMaster> so does freebsd sed iirc
21:32:37 <Rugxulo> yes
21:32:41 <AnMaster> and so on
21:32:43 <Rugxulo> but not NetBSD or OpenBSD
21:33:07 <Rugxulo> or Minix (although they probably have GNU sed by now)
21:33:22 <Rugxulo> `befunge 12481> #+?\# _.@
21:33:26 <HackEgo> No output.
21:33:28 <Rugxulo> !befunge 12481> #+?\# _.@
21:33:37 <Rugxulo> gah, I can never remember
21:33:49 <Deewiant> The first one seemed correct, since it responded.
21:34:18 <AnMaster> Rugxulo, indeed. still why not do: sed 's/..../..../' "$i" > "$i.tmp" && mv "$i.tmp" "$i"; done A bit more if you don't trust the directory to not contain such stuff (create a temp dir in /tmp and do it in there then)
21:34:29 <AnMaster> Rugxulo, who uses minix?
21:34:45 <Rugxulo> Minix 3 is getting better all the time (or so I hear) ;-)
21:35:08 <pikhq> AnMaster: Try "mktmp".
21:35:18 <pikhq> Erm. mktemp.
21:35:19 <Rugxulo> pikhq, try Perl ;-)
21:35:25 <AnMaster> Rugxulo, the "for %a in (*.txt) do" I didn't know about. when was it added to cmd.exe? I'm pretty sure command.com didn't have it at least. So it must have been in NT or later?
21:35:40 <Rugxulo> COMMAND.COM has had it for as long as I'm aware of, actually
21:35:42 <AnMaster> pikhq, well yes. that is what you would use for said temp dir in /tmp
21:35:47 <pikhq> ... File...
21:35:51 -!- nice has joined.
21:35:52 <AnMaster> Rugxulo, guess it was poorly documented
21:35:55 <AnMaster> pikhq, that would work too
21:36:03 <AnMaster> Rugxulo, since I never seen any docs mentioning it
21:36:06 <Rugxulo> AnMaster, nope ... it's DOS history, actually, which you missed ;-)
21:36:22 <AnMaster> Rugxulo, windows help center certainly didn't mention it on xp (which was last time I checked)
21:36:23 -!- nice has changed nick to KingOfKarlsruhe.
21:36:42 <Rugxulo> "help for" will show you what you want
21:36:42 <AnMaster> and if the OS doesn't even come with proper documentation, how can anyone use it?
21:37:07 <pikhq> AnMaster: Ask Windows. :P
21:37:11 <AnMaster> Rugxulo, what will tell me there is a help or a for command there though. Sure that would be reasonable to guess. But far from everything is.
21:37:36 <Rugxulo> ask that Paperclip ... or Rover ;-)
21:37:47 <AnMaster> I thought that was office only. And shudder
21:37:55 <Rugxulo> :-D
21:38:38 <Rugxulo> no, I'm on XP now, and the "Search" in Start Menu (optionally) has Rover ;-)
21:39:53 <Rugxulo> take that, Linux :-P
21:40:33 <Rugxulo> anyways, no seriously, even MS-DOS 6 had HELP.EXE, and you can usually get short help by doing "<command> /?"
21:41:08 <Rugxulo> DR-DOS had DOSBOOK.EXE (with a HELP.BAT alias), FreeDOS has HTMLhelp (HELP.EXE)
21:41:48 <Rugxulo> so it's not like they're trying to leave users in the cold, it's just that docs are always secondary importance, plus it's too much work to document everything
21:43:18 <Rugxulo> but there is no sed equivalent (that I know of) on Windows
21:43:26 <Rugxulo> find and findstr are Windows equivalents of grep, though
21:43:38 <Rugxulo> QBasic hasn't been included since Win95? (98?)
21:43:49 <Rugxulo> debug still exists though, so technically you could probably do anything ;-)
21:43:58 <Rugxulo> debug and edlin, heh
21:44:44 <Rugxulo> (but DOS apps!, so only 32-bits Windows supported, kthxbai)
21:44:58 <Rugxulo> TCC/LE says it does regex
21:45:15 <Deewiant> edlin is the sed equivalent on Windows :-P
21:45:31 <Deewiant> Who needs things like long filenames
21:45:31 <Rugxulo> findstr supports regex, but find and edlin don't
21:45:38 <Rugxulo> ;-)
21:45:46 <Rugxulo> kids today :-D
21:46:07 <Deewiant> It's also 16-bit and thus not provided on 64-bit Windowses (or maybe it is, but it won't run)
21:46:17 <Rugxulo> right, that's what I meant
21:46:32 <Rugxulo> DOSBox will run it though (although "why" is probably a better question)
21:46:33 <Deewiant> Ah right, you said so, I missed that.
21:49:18 <Rugxulo> BTW, Deewiant, what changed in latest CCBI? (still no latest build for Win32)
21:49:20 -!- FireFly has quit ("Later").
21:49:44 -!- BeholdMyGlory has quit (Read error: 54 (Connection reset by peer)).
21:50:10 <Deewiant> There's a Changelog.
21:50:34 <Deewiant> FILE's R now reflects on EOF.
21:50:38 <Deewiant> That's the only change.
21:50:46 <Rugxulo> FILE's R now reflects on EOF.
21:50:50 <Rugxulo> oops, beat me to it ;-)
21:51:31 <Deewiant> The fact that the change wasn't that big explains why I haven't bothered to build it on Windows :-P
21:51:56 <Rugxulo> lazy ;-)
21:52:16 <Rugxulo> BTW, have you heard of "Go" yet?
21:52:20 <Rugxulo> (since you like D)
21:52:23 <Deewiant> Yes, several times.
21:52:52 <Deewiant> (Two D-related IRC channels, LLVM IRC channel, D newsgroups, couple of posts on reddit, post on /.)
21:53:06 <Deewiant> And now, #esoteric. :-P
21:54:05 <Rugxulo> mixed reviews (although not too critical), but I suspect it will mature / morph a lot in the next year or two
21:54:20 <Deewiant> Shrug.
21:54:40 <fizzie> Was this the Google thing?
21:54:46 <Rugxulo> yes
21:54:51 <fizzie> comp.lang.forth of all places had a short thread about it.
21:55:40 <Rugxulo> yeah, because of the upcoming ChromeOS which may or may not utilize it
21:55:44 <Deewiant> Young people are excited because it's by Google, older people because it's by the C/Unix/Plan 9 folks.
21:55:48 <Rugxulo> (relevant 'cause Forth can often serve as OS also)
21:56:03 <Rugxulo> and *BSD people because it's BSD-licensed
21:56:23 <fizzie> Given how much stuff is BSD-licensed, *BSD people must lead pretty exciting lives.
21:56:25 <Deewiant> What, as opposed to proprietary? :-P Most languages are at least that free.
21:56:32 <Deewiant> As for the language itself: meh. I'll see what happens.
21:56:42 <Rugxulo> no, not all languages are free
21:56:50 <Deewiant> I didn't say that they were.
21:57:07 <Rugxulo> well, "most" implies that almost all are, which I doubt is the case
21:57:16 <Rugxulo> (although they probably should be)
21:57:38 <fizzie> Deewiant: Yes, well, the comp.lang.forth thread has apparently now degenerated into a "discussion" whether Chrome OS is "an evil scheme to take control of everybody's computer?" (direct quote).
21:57:39 <Deewiant> "most" implies that more than half are, which probably is the case, excepting all kinds of used-in-only-one-project type DSLs.
21:57:42 <ais523> my guess is that the vast majority of languages are proprietary, but none of the proprietary ones are massively popular
21:57:49 <Deewiant> fizzie: :-D
21:58:00 <ais523> it seems that half of companies have a proprietary language or two of their own
21:58:19 <Deewiant> Yeah, and I'd exclude those if they're only for their internal use.
21:58:28 <Rugxulo> are you really surprised? Earth has like 6000 living human languages :-/
21:59:11 <Deewiant> Who's surprised by what?
21:59:34 <Rugxulo> nobody, just saying it's obvious that there are way too many languages out there
22:01:37 <Rugxulo> `befunge 01g0g,01g1+:83*-!#@_01p
22:01:38 <HackEgo> No output.
22:01:41 <Rugxulo> gah
22:01:46 <Rugxulo> I know I'm doing it wrong
22:02:46 <Deewiant> What, are you trying to do?
22:02:47 <Rugxulo> (quoting, lol): "STYLE GUIDELINES. Every Befunge programmer has their
22:02:48 <Rugxulo> own style. As a Befunge programmer your style should
22:02:50 <Rugxulo> be unique, ugly, and incomprehensible to others."
22:02:51 <Deewiant> s/,//
22:03:01 <Rugxulo> run a program that actual has output!
22:03:04 <Rugxulo> somehow it's not working
22:03:20 <Deewiant> `befunge 55+"ihaO",,,,,@
22:03:21 <HackEgo> No output.
22:03:29 <Deewiant> Hmm
22:03:39 <Deewiant> I managed to typo, even.
22:03:41 <Rugxulo> ehird knows how, but he's not here and I can't remember
22:03:49 <Rugxulo> `befunge 55*.@
22:03:50 <HackEgo> No output.
22:03:56 <Rugxulo> `run befunge 55*.@
22:03:57 <HackEgo> No output.
22:04:03 <ais523> `help
22:04:03 <HackEgo> Runs arbitrary code in GNU/Linux. Type "`<command>", or "`run <command>" for full shell commands. "`fetch <URL>" downloads files. Files saved to $PWD are persistent, and $PWD/bin is in $PATH. $PWD is a mercurial repository, "`revert <rev>" can be used to revert to a revision. See http://codu.org/projects/hackbot/fshg/
22:04:14 <GregorR> Since people were talking about Google Go the other day, here is my brief opinion from poking arounda t it a bit: Probably the worst language design in at least a decade. Easily the worst toolchain design in the entire history of toolchains. If I never touch this language again, it will be too soon.
22:04:14 <ais523> `run which befunge
22:04:15 <HackEgo> No output.
22:04:18 <Rugxulo> `befunge 55*.@ > /dev/null
22:04:18 <HackEgo> No output.
22:04:21 <ais523> there's your problem
22:04:26 <ais523> `run ls /usr/bin
22:04:27 <HackEgo> X11 \ [ \ a2p \ addpart \ addr2line \ apropos \ apt-cache \ apt-cdrom \ apt-config \ apt-extracttemplates \ apt-ftparchive \ apt-get \ apt-key \ apt-mark \ apt-sortpkgs \ aptitude \ aptitude-create-state-bundle \ aptitude-run-state-bundle \ ar \ arch \ as \ awk \ base64 \ basename \ bashbug \ bdftopcf \ bdftops \ bdftruncate
22:04:33 <ais523> `run ls /usr/bin/b*
22:04:34 <HackEgo> /usr/bin/base64 \ /usr/bin/basename \ /usr/bin/bashbug \ /usr/bin/bdftopcf \ /usr/bin/bdftops \ /usr/bin/bdftruncate \ /usr/bin/bsd-write
22:04:39 <Rugxulo> `bef 55*.@ > /dev/null
22:04:39 <HackEgo> No output.
22:04:40 <ais523> see, no befunge installed
22:04:45 <ais523> `run ls /usr/bin/c*
22:04:46 <HackEgo> /usr/bin/c++ \ /usr/bin/c++filt \ /usr/bin/c2ph \ /usr/bin/c89 \ /usr/bin/c89-gcc \ /usr/bin/c99 \ /usr/bin/c99-gcc \ /usr/bin/c_rehash \ /usr/bin/cal \ /usr/bin/calendar \ /usr/bin/captoinfo \ /usr/bin/catchsegv \ /usr/bin/catman \ /usr/bin/cc \ /usr/bin/chage \ /usr/bin/chattr \ /usr/bin/chcon \ /usr/bin/chfn \ /usr/bin/chkdupexe
22:04:46 <Rugxulo> `run which bef
22:04:47 <HackEgo> No output.
22:04:48 <GregorR> !help
22:04:53 <GregorR> Oh, EgoBot isn't here.
22:04:56 <GregorR> That ain't right.
22:05:12 <Rugxulo> GregorR, worst this decade????
22:05:20 <ais523> given that HackEgo has neither cfunge nor ccbi, I'm not sure why you're expecting befunge commands to work
22:05:23 <GregorR> Rugxulo: In terms of "real" languages. It's completely horrible.
22:05:29 <Rugxulo> compared to what?
22:05:32 <Rugxulo> Java? C#?
22:05:35 <ais523> (there might be another befunge interp on there, but the c* command checks the most two common)
22:05:38 <Deewiant> !befunge 55+"iahO",,,,,@
22:05:49 <Deewiant> That'd require EgoBot, who is absent.
22:05:49 <ais523> Deewiant: EgoBot isn't here
22:05:52 <GregorR> Rugxulo: Literally every language created for serious use.
22:05:53 <Rugxulo> `run which fbbi
22:05:55 <HackEgo> No output.
22:06:01 <ais523> Rugxulo: fbbi?
22:06:06 <GregorR> Rugxulo: There is no language I can compare it against without saying "Wow, Go is a steaming pile of shit compared to that language."
22:06:09 <ais523> `run ls /usr/local/bin
22:06:10 <HackEgo> No output.
22:06:10 <Deewiant> ais523: Flaming Bovine Befunge-98 Interpreter
22:06:15 <ais523> oh, the original
22:06:26 <Rugxulo> GregorR, main reasons?
22:06:42 <Deewiant> GregorR: COBOL?
22:06:43 <ais523> some of its advertised features are comparable to INTERCAL's
22:06:48 <ais523> Deewiant: no, Issue 9
22:06:49 <Rugxulo> `run ls /usr/bin
22:06:50 <HackEgo> X11 \ [ \ a2p \ addpart \ addr2line \ apropos \ apt-cache \ apt-cdrom \ apt-config \ apt-extracttemplates \ apt-ftparchive \ apt-get \ apt-key \ apt-mark \ apt-sortpkgs \ aptitude \ aptitude-create-state-bundle \ aptitude-run-state-bundle \ ar \ arch \ as \ awk \ base64 \ basename \ bashbug \ bdftopcf \ bdftops \ bdftruncate
22:06:52 <GregorR> Deewiant: COBOL is not in the last decade ;)
22:07:02 <Deewiant> ais523: That's an interesting statement
22:07:15 <ais523> Deewiant: reddit have decided to call the language Issue 9 rather than Go
22:07:27 <Deewiant> I know; I was talking about your previous statement
22:07:28 <Rugxulo> that's because somebody already has "Go!" as a langauge
22:07:32 <GregorR> Rugxulo: Its type system is insane, its toolchain is horrible, its method of "encapsulation" is capitalization, it has no consistent interoperability story, lesse...
22:07:35 <Deewiant> Regarding INTERCAL.
22:07:40 <GregorR> There is literally not a single redeeming feature.
22:07:44 <ais523> Deewiant: things like light-weight coroutines
22:07:45 <ais523> and duck typing
22:07:51 <Rugxulo> what's so bad about the toolchain?
22:07:54 <Deewiant> Sheesh, this shell server is lagging.
22:07:59 <GregorR> Rugxulo: THE BINARIES ARE NAMED PER FUCKING ARCHITECTURE
22:08:05 <GregorR> 6g? 8g? WTF?!
22:08:05 <Deewiant> GregorR: Plan 9 style, I hear.
22:08:09 <ais523> (people nowadays seem not to keep track of INTERCAL development)
22:08:12 <GregorR> Yes, plan 9 is fucking stupid that way.
22:08:15 <Rugxulo> GregorR, it's very fast, it's supposed to be good for concurrency
22:08:18 <Rugxulo> that's two pros, no?
22:08:25 <Deewiant> I have yet to see any benchmarks.
22:08:26 <GregorR> Fast is not a pro unless we're compiling in 1965.
22:08:32 <ais523> well, ick is pretty fast at compiling too
22:08:37 <GregorR> And the compiled code isn't fast, the compiler is fast.
22:08:40 <Deewiant> And yeah, fast compiling is pretty much a no-brainer these days.
22:08:42 <ais523> admittedly, it then gets slowed down as you have to compile the C as well
22:08:49 <Rugxulo> uh ... I sure wish GCC was faster
22:08:52 <Rugxulo> lots faster
22:08:53 <Deewiant> Only C++ is too slow for comfort.
22:09:01 <GregorR> It has fork-join-ish concurrency, which makes it "good" in a "hey if we restrict you to a ridiculous degree you won't make mistakes" sense.
22:09:12 <Deewiant> (And CCBI2, but that's due to compiler bugs.)
22:10:07 <Rugxulo> I've never heard anyone say, "Nah, I don't need faster compiles"
22:10:18 <ais523> GregorR: see INTERCAL's concurrency model, it's richer than pretty much all common languages up to about the Haskell level
22:10:19 <Deewiant> Nah, I don't need faster compiles
22:10:30 <Deewiant> You want a .wav of that, too? :-P
22:10:47 <Rugxulo> Deewiant, you must not compile a lot of stuff
22:10:53 <Rugxulo> or else extremely patient
22:10:53 -!- GregorR has set topic: Go is a no-go | http://tunes.org/~nef/logs/esoteric/?C=M;O=D.
22:11:14 <pikhq> I prefer Haskell's concurrency model over most other things.
22:11:15 * Rugxulo jokingly suggested to rename it "!Go" (not Go) or "Go2" (goto) ;-)
22:11:31 <pikhq> "Shared state"? What does that even mean when you don't have state? :P
22:11:52 <Deewiant> What kind of stuff do you compile that makes you feel such a need for speed?
22:12:12 <Rugxulo> any non-trivial stuff with latest GCC
22:12:33 <Deewiant> I compile LLVM with latest GCC, I don't really mind how long that takes.
22:12:39 <Rugxulo> it's not for nothing that LLVM + CLang are gaining popularity
22:13:18 <Deewiant> LLVM's codegen bit isn't very fast, GCC is often faster :-P
22:13:20 <Rugxulo> Deewiant, what machine do you use?
22:13:29 <Deewiant> Linux niΓ°avellir 2.6.32-rc3-deewiant #1 SMP PREEMPT Fri Oct 9 17:07:44 EEST 2009 x86_64 Intel(R) Core(TM)2 Quad CPU Q9550 @ 2.83GHz GenuineIntel GNU/Linux
22:13:37 <Rugxulo> yes, obviously GCC is more mature (but slow ... oh so slowwwwwwww)
22:13:54 <Rugxulo> 2.6.32-rc3 ... wow, slow adopter, eh? ;-)
22:13:58 <Deewiant> Clang is a faster frontend, but LLVM's backend is typically slower in my experience.
22:14:21 <Rugxulo> Core 2 Quad x86-64, should've known :-P ;-0
22:14:23 <Rugxulo> :-)
22:14:59 <Rugxulo> sure if you have one of the fastest machines on the market it's "good enough", but be realistic, it could be loads better
22:15:24 -!- KingOfKarlsruhe has quit (Remote closed the connection).
22:15:29 <Deewiant> Sure, it probably could be
22:15:34 <Deewiant> I just don't feel the need for it. :-P
22:15:36 <Rugxulo> not probably, definitely!
22:15:46 <Rugxulo> ... because you have a damn fast machine, one of the fastest out there now
22:15:59 <Deewiant> I don't know about definitely since I haven't looked at GCC's code.
22:16:15 <Deewiant> And yes, that's likely a reason I don't feel the need. :-P
22:16:23 <Rugxulo> it's like Bill Gates saying, "I think I'll give away a billion", it's because he's rich, not because 1 bil ain't worth nothing!!
22:16:26 <ais523> don't look at GCC's code
22:16:29 <ais523> you may end up wishing you hadn't
22:16:32 <Deewiant> I don't intend to.
22:16:36 <Rugxulo> thar be dragons
22:16:49 <Rugxulo> can't be worse than Befunge98, though ;-)
22:16:56 <ais523> it is
22:17:10 * Rugxulo was joking
22:17:11 <ais523> for instance, gcc's .md files are actually written in two different languages
22:17:17 <ais523> you have to write them as polyglots
22:17:38 <Rugxulo> I've briefly looked at GCC, I'm not saying it's not complex ... just that I'm sure it could be better
22:17:53 <Deewiant> It's just one language that coincidentally happens to be the intersection of two other languages
22:18:52 <ais523> an interesting viewpoint
22:19:31 <SimonRC> and I thought that polyglots were just a game
22:20:00 <pikhq> ais523: Speaking of GCC, how goes gccbf?
22:20:09 <pikhq> Rugxulo: GCC's code is a crime.
22:20:11 <ais523> pikhq: it doesn't, I'm busy with loads of other things
22:20:18 <pikhq> ais523: Alas.
22:20:29 <ais523> although, it's now complete enough that it feels like a very buggy program, rather than an incomplete one
22:20:32 <pikhq> And its build system should be put out of its misery.
22:20:38 <Rugxulo> lots of code is horrible, it's rewriting everything that is hard
22:22:02 <ais523> pikhq: gcc's build system actually inspired ick's
22:22:17 <ais523> in the sense of, I looked at gcc's build system and resolved to do something different
22:22:26 <ais523> preferably as different as possible, whilst still using autoconf
22:22:41 <Rugxulo> ais523: what is this ick you speak of? updated Intercal-to-C proggie?
22:22:51 <pikhq> ais523: Ah.
22:23:03 <ais523> Rugxulo: the C-INTERCAL compiler
22:23:07 <ais523> ick(10
22:23:09 <ais523> * ick(1)
22:23:48 <Rugxulo> ick15d12.zip, ick-c024.tgz, ick-c025.tgz (all I have on this cpu, and I never really bothered learning it)
22:24:43 <ais523> 0.25 is rather old
22:25:01 <Rugxulo> okay, yes, obviously, I'm implying that you should tell me what's new, where to get it, etc. ;-)
22:25:04 <ais523> 0.-2.0.29 is available at http://overload.intercal.org.uk/
22:25:08 * SimonRC likes Go's ability to retrofit interfaces onto classes
22:25:13 <SimonRC> compile-tim duck typing!
22:25:15 <ais523> it's a beta, but should work fine on Linux and UNIX systems
22:25:23 <ais523> (the reason it's beta is that it hasn't been tested on other architectures yet)
22:25:27 <SimonRC> but it has no parameterised types
22:25:30 * Rugxulo wonders about DJGPP ...
22:25:30 <SimonRC> like WTF
22:25:41 <Rugxulo> Go or Intercal? ;-)
22:25:43 <SimonRC> welcome to the 80s
22:25:44 <ais523> Rugxulo: older versions were actually tested on DJGPP
22:26:04 <SimonRC> intercal has the excuse that it is actually from the 70s
22:26:16 <ais523> Rugxulo: download 0.28 if you're on DJGPP, the DOS build is supported there
22:26:21 <Rugxulo> BTW, wasn't there a Befunge interpreter written in Intercal?
22:26:24 <ais523> Rugxulo: yes
22:26:29 <SimonRC> Rugxulo: what was that ping for?
22:26:33 <ais523> it's in the distribution, I forget which version it was added in though
22:26:44 <Rugxulo> SimonRC: seemed you were still on about Go (twenty minutes late) ;-)
22:26:52 <SimonRC> yeah
22:27:01 <ais523> oh, added in 0.28, with a bug fixed on 0.29
22:27:03 <ais523> *in
22:27:07 <SimonRC> on several channels at once
22:27:12 <Rugxulo> why so many releases on April Fools? coincidence? joke? :-)
22:27:16 <ais523> Rugxulo: deliberate
22:27:22 <ais523> it's a classic day for releasing INTERCAL versions
22:27:23 <Rugxulo> of course
22:28:14 <Rugxulo> bugfix in 0.29 ("0.-2.0.29") doesn't help me much
22:28:24 <Rugxulo> unless you want me to try 0.29 in DJGPP also
22:28:49 <Rugxulo> why such huge size increases? hmmm ...
22:30:02 <Rugxulo> and what's the diff b/w C-INTERCAL and CLC-... oh well
22:30:48 <Rugxulo> what, what the ... pax + gz ???
22:31:09 <Deewiant> It's INTERCAL, it can't be too easy.
22:31:24 <ais523> C-INTERCAL and CLC-INTERCAL are entirely different compilers
22:31:36 <Rugxulo> obviously but how / why?
22:31:39 <ais523> also, .pax.gz is the POSIX-mandated format for tarballs
22:31:46 <Rugxulo> *sigh*
22:31:47 <ais523> Rugxulo: completely different codebases and feature sets, different maintainers
22:31:53 <ais523> Rugxulo: you'll find tar decompresses it just fine
22:31:58 <ais523> it's backwards-compatible
22:32:14 <Deewiant> Only GNU tar, though, or?
22:32:16 <Rugxulo> I've heard of pax, don't get me wrong, just ... too obscure IMHO
22:32:21 <Deewiant> I can't remember if we checked that.
22:32:29 <ais523> Deewiant: any POSIX-compatible tar
22:32:37 <ais523> although the ones that don't know the format will create junk dotfiles
22:32:40 * Rugxulo is starting to hate POSIX ...
22:32:55 <AnMaster> <Rugxulo> anyways, no seriously, even MS-DOS 6 had HELP.EXE, and you can usually get short help by doing "<command> /?" <-- yeah but how do you find out what built in control structures to look for? I would do man bash. but well yeah what is the equivalent of that...
22:33:20 <ais523> AnMaster: help.exe had a command list if you started it with no arguments
22:33:22 <Rugxulo> you mean for the shell?
22:33:26 <Rugxulo> yeah
22:33:36 <ais523> also, help cmd, in more recent versions
22:33:49 <Rugxulo> or FreeCOM (from FreeDOS) prints a nice word list if you type "?" by itself
22:33:52 <Deewiant> ais523: Yeah, but I'm wondering if anything other than GNU is actually POSIX-compatible in that respect. :-P
22:34:02 <ais523> Deewiant: it works in practice
22:34:24 <ais523> basically, the primary goal of INTERCAL is to act differently from everything else
22:34:32 <ais523> and if everything else packages incorrectly, INTERCAL should package correctly
22:34:51 <Rugxulo> then use .zip.7z or something silly like that (or .lbr + .sq, heh)
22:35:46 <AnMaster> <Rugxulo> (quoting, lol): "STYLE GUIDELINES. Every Befunge programmer has their <-- where from is that?
22:35:46 <ais523> that isn't correct, that's just being silly for the sake of it
22:36:00 <ais523> admittedly, INTERCAL does a lot of silly for the sake of it, but silly for a good reason is even better
22:36:03 <Rugxulo> AnMaster, from that mailing list log you sent me
22:36:55 <Rugxulo> ais523, then you should write a PAX unpacker in B98 or Intercal, stat! ;-)
22:37:03 <ais523> Rugxulo: just use tar, it unpacks pax just fine
22:37:09 <Rugxulo> lazy ;-)
22:37:31 <ais523> and you need quite a bit of knowledge about your specific OS to write a pax unpacker, anyway
22:37:45 <ais523> I /am/ interested to know if the 0.29 beta runs unmodified on DJGPP
22:37:49 <ais523> as I haven't tested that at all
22:37:58 <ais523> and it's marked as a beta precisely because of the lack of testing
22:38:27 <AnMaster> <ais523> you have to write them as polyglots <-- huh?
22:38:37 <ais523> AnMaster: a gcc .md file is compiled into two different object files
22:38:42 <ais523> each of which interprets the .md file a different way
22:39:31 <AnMaster> oh about gcc, I was surprised yesterday when gcc optimised a printf() into a puts()
22:39:37 <AnMaster> that I hadn't expected at all
22:39:57 * SimonRC recalls Wumpus in Bef93. Spectacular.
22:40:05 <AnMaster> ais523, what does the md file do?
22:40:15 <Rugxulo> .md = machine description ??
22:40:18 <ais523> AnMaster: it's one of the major files that describes a target (such as x86 or ARM)
22:40:21 <SimonRC> I spent hours tracing the code path, and it only just fits in the grid
22:40:47 <ais523> it basically specifies a) how to translate RTL into asm, and b) what sort of RTL is best to generate to produce decent code on that platform
22:41:02 <ais523> and it compiles into a GIMPLE->RTL translator and an RTL->asm translator
22:41:03 <Rugxulo> yeah, but some platforms are better covered than others
22:41:16 <AnMaster> ais523, why
22:41:26 <Rugxulo> is GIMPLE the 4.x one or the 3.x (treelang??) ?
22:41:35 <ais523> Rugxulo: neither, it's an internal representation
22:41:47 <ais523> that's used between the frontend and backend
22:41:55 <ais523> AnMaster: because optimisations take place on the RTL
22:42:14 <ais523> the polyglot basically specifies the GIMPLE, RTL /and/ asm versions of bits of code
22:42:25 <ais523> but you have to do a lot of messing about to get it to work
22:42:39 <Rugxulo> no worse than AutoConf, heh ;-)
22:43:13 <ais523> Rugxulo: autoconf can and does work fine in C-INTERCAL 0.29
22:43:22 <ais523> I spent weeks learning how it actually worked, rather than how everyone uses it
22:43:27 <Rugxulo> "in" or "used by"?
22:43:27 <ais523> together with the rest of autotools
22:43:33 <ais523> well, used to build it
22:43:36 <Rugxulo> ;-)
22:43:42 <Rugxulo> good, otherwise I'd have a heart attack
22:43:53 <ais523> as a result, C-INTERCAL 0.29 has a build system that is both a) good, and b) based on autoconf
22:43:58 <ais523> a combination which most people thought impossible
22:43:58 <AnMaster> ais523, why does it have to be one file. Couldn't it be two?
22:44:12 <ais523> AnMaster: there'd be quite a bit of duplication if there were two; I suppose that was the reasoning
22:44:39 <AnMaster> <Rugxulo> "in" or "used by"? <-- nice idea there.
22:44:49 <AnMaster> would fit well into intercal somehow I feel
22:45:02 <ais523> AnMaster: write an m4 fingerprint for cfunge
22:45:13 <AnMaster> ais523, what about autoconf configuring itself? Surely that can't be too bad
22:45:29 <ais523> AnMaster: it mostly doesn't, autoconf is a script not a binary
22:45:31 <AnMaster> ais523, a what?
22:45:37 <AnMaster> oh
22:45:38 <AnMaster> hah
22:45:46 <AnMaster> ais523, I wouldn't know where to start with m4
22:46:00 <ais523> Rugxulo: one of the new features in 0.29 is the ability to compile Funge-98 and INTERCAL together
22:46:01 <AnMaster> I avoid autoconf, I know it a bit, but I don't know any "raw" m4
22:46:11 <ais523> raw m4's actually quite a nice language
22:46:19 <AnMaster> ais523, well, automake then? Or something
22:46:19 <ais523> it has an eso feel to it, too
22:46:24 <ais523> AnMaster: automake's Perl
22:46:25 <ais523> apparently
22:46:27 <AnMaster> ais523, so does malbolge
22:46:28 <AnMaster> ...
22:46:32 <ais523> at least, written in Perl
22:46:39 <AnMaster> mhm
22:46:44 <ais523> but you don't write bits of Perl to expand it, you just use more m4
22:47:10 <AnMaster> bbiab....
22:47:16 <pikhq> Automake's Perl and m4.
22:47:31 <ais523> well, it's automake (perl) plus a library for autoconf (m4)
22:47:53 <ais523> Rugxulo: how's your C-INTERCAL build going?
22:48:13 <Rugxulo> I'm trying to be a good boy and check the "README" first (instead of asking blindly, "just configure and make??"
22:48:15 <Rugxulo> )
22:48:23 <ais523> fair enough
22:48:27 <ais523> the README used to be essential reading
22:48:32 * Rugxulo didn't want to hear, "Read the README, n00b!"
22:48:34 <ais523> (what version are you trying, btw?)
22:48:39 -!- kar8nga has quit (Remote closed the connection).
22:48:40 <Rugxulo> 0.29
22:48:46 <Rugxulo> you said 0.28 works, so ...
22:48:53 <ais523> fair enough
22:49:01 <ais523> 0.28 works, 0.29 is untested (on DJGPP)
22:49:27 <Rugxulo> yes, you can configure out-of-tree, but your make must support VPATH (BTW)
22:49:55 <ais523> I'm not completely certain that's needed, given the typical automake insanity, but it probably is
22:50:09 <Rugxulo> obviously haven't updated README, it says DJGPP works fine
22:50:24 <ais523> ugh, good catch
22:50:28 <ais523> I think I wrote what was intended to happen
22:50:31 <ais523> rather than what actually does
22:51:14 <ais523> yep, needs VPATH to build out of tree
22:51:19 <ais523> most makes have it nowadays, though, IIRC
22:51:29 <Rugxulo> also, the bit about cfunge is obviously moot to DOS (no mmap)
22:51:40 <Rugxulo> ais523: "most" means more than just GNU and *BSD, I hope
22:51:41 <ais523> yep, it works fine without though
22:52:06 <ais523> and you can still build in-tree even if your make's lacking out-of-tree features
22:53:02 -!- Gracenotes has quit (Remote closed the connection).
22:53:32 <Rugxulo> Autoconf 2.61 (should work okay, 2.63 has typo, 2.64 exposed a now-fixed DJGPP Bash bug)
22:54:42 -!- puzzlet has quit (Remote closed the connection).
22:54:43 <Rugxulo> Deewiant, it may be an artifact of DJGPP (or Windows in general, e.g. Cygwin) but Autoconf is *slow*
22:54:46 -!- puzzlet has joined.
22:54:59 <Rugxulo> even 2.64 ("30% faster") isn't nearly fast enough for comfort
22:55:07 <ais523> Rugxulo: the reason it's slow is that it has to invoke the compiler and linker tens of times
22:55:09 <Rugxulo> " ... faster on Cygwin"
22:55:10 <ais523> on lots of trivial programs
22:55:17 <ais523> so, it's actually gcc that's being slow, rather than autoconf
22:55:24 <Deewiant> Rugxulo: Windows.
22:55:29 <Rugxulo> it checks too much cruft
22:55:36 <Deewiant> It's not gcc, it's Windows's process invocation.
22:55:41 <Rugxulo> (I know I know, tweak my config.site, bah)
22:55:48 <ais523> Rugxulo: on some systems, those crufts may have unexpected values!
22:56:06 <Deewiant> I've clocked ./configure --help taking longer on Windows than the whole ./configure on Linux, on the same machine.
22:56:17 <ais523> Deewiant: wow
22:56:19 <Rugxulo> no, I mean, some of the things it checks for are bogus, i.e. not required for the program (you could also blame the package maintainer I guess, but ...)
22:56:19 <ais523> was there a cache?
22:56:27 <ais523> Rugxulo: I've tried to be careful wrt that
22:56:30 <pikhq> Windows process invocation is *very very slow*.
22:56:34 <Rugxulo> ais523: it's done already
22:56:42 <Deewiant> ./configure uses some kind of cache always, no?
22:56:43 <ais523> if the configure went fine, try a make
22:56:44 <Rugxulo> luckily not as painful as other packages I've built before (e.g. ZILE)
22:56:45 <Deewiant> But it really doesn't matter
22:56:53 <Rugxulo> -C is optional
22:56:57 <ais523> Deewiant: I mean, a pre-existing cache
22:57:03 <pikhq> Deewiant: Not really; caching is a bit broken.
22:57:04 <Deewiant> ais523: No in both cases.
22:57:05 <ais523> if you don't use -C, it caches just for one run then deletes the cache again
22:57:20 <Rugxulo> ais523: configure didn't bomb out, so here goes nothing ... "make"
22:57:58 * Rugxulo should've done --disable-dependency-tracking but didn't want to push too hard ...
22:58:23 <Rugxulo> libidiot.a, a lib after my own heart!
22:58:24 <pikhq> The dependency tracking bit only matters for rebuilds.
22:58:29 <Deewiant> ais523: In any case, ./configure on Windows is, in my experience, slower than it would be for me to type the values in manually.
22:58:33 <Rugxulo> yes, I know, and slows everything down too
22:58:47 <Deewiant> Autoconf-generated configures, that is.
22:58:57 <Rugxulo> ais523: completed without error, now what? test some stuff in pit?
22:59:02 <ais523> Rugxulo: if you like
22:59:07 <ais523> or install if you prefer
22:59:21 <Rugxulo> what is convickt?
22:59:25 <Rugxulo> and oil?
22:59:26 <ais523> Rugxulo: a character set converter
22:59:44 <ais523> say, if you get an INTERCAL program in EBCDIC, you can convert it to ASCII to run it with C-INTERCAL
22:59:52 <ais523> and OIL is a domain-specific language for writing optimiser idioms in
22:59:56 <Rugxulo> yeah, like that'll happen ;-)
23:00:09 <ais523> the optimiser got better really quickly after I wrote OIL
23:00:13 <ais523> because it became so much easier to change
23:00:58 <pikhq> ais523: Your .pax files really truly should be .tar; pax creates archives in USTAR format, just like GNU Tar and BSD Tar...
23:01:19 <ais523> pikhq: no, USTAR and .pax are slightly different IIRC
23:01:23 <pikhq> No.
23:01:27 <AnMaster> <Rugxulo> yes, you can configure out-of-tree, but your make must support VPATH (BTW) <-- doesn't all?
23:01:40 <Rugxulo> apparently not
23:01:52 <Rugxulo> as they always warn about that
23:02:09 <Rugxulo> ick +help doesn't work (should it?)
23:02:13 <ais523> man 5 tar says that pax follows the ustar format, but contains more data
23:02:19 <ais523> Rugxulo: no, +help is for programs you created
23:02:22 <ais523> try ick -@
23:02:38 <ais523> (as in, ick supports -@, programs you create using ick support +help)
23:03:00 <pikhq> Hmm. Sure enough -- pax adds an extended header.
23:03:35 <ais523> if you look at a pax archive using a non-pax untar tool, such as Emacs, you see extra dotfiles in the archive
23:04:23 <pikhq> It appears that pax archives are as capable as GNU tar, but a bit different.
23:04:45 <ais523> GNU tar isn't POSIX-compatible
23:04:56 <ais523> whereas pax is compatible with old tars, and with GNU tar as well
23:05:14 <pikhq> Yes, GNU tar is USTAR with additional features that aren't all that POSIX-compliant.
23:05:21 <pikhq> (IIRC)
23:05:30 <ais523> in fact, I think pax is compatible even with ancestral tar
23:05:38 -!- Gracenotes has joined.
23:05:44 <ais523> which just ignores all the data in there it doesn't understand, as it's placed into reserved fields and-or dotfiles
23:05:46 <AnMaster> <Rugxulo> yes, I know, and slows everything down too <-- never noticed that. Half a second difference on a project taking 3-4 minutes to build. But that is within the error margin I think. Since the build time varied with like 5 seconds anwyway
23:05:52 <AnMaster> anyway*
23:06:15 <ais523> how much dependency tracking slows it down depends on the compiler
23:06:16 <Rugxulo> try on a non-Core2 billion GB machine ;-)
23:06:17 <pikhq> ais523: According to this, it isn't required to be: http://www.opengroup.org/onlinepubs/009695399/utilities/pax.html
23:06:29 <ais523> for compilers that can't produce dependency info as a side-effect, it slows it down by a factor of 2
23:06:39 <Rugxulo> ais523: it doesn't work ... or at least "ick -g hail_mary.3i" isn't doing anything (nor "-c" etc.)
23:06:57 <ais523> Rugxulo: not even producing hail_mary.c or hail_mary?
23:07:01 <ais523> try ick -y hail_mary.3i
23:07:08 <ais523> that should bring up a debug prompt
23:07:21 <Rugxulo> nope, keeps saying "NO SKELETON IN MY CLOSET"
23:07:35 <Deewiant> Gotta love INTERCAL error messages <3
23:07:35 <AnMaster> <ais523> for compilers that can't produce dependency info as a side-effect, it slows it down by a factor of 2 <-- gcc can do that though
23:07:48 <ais523> Rugxulo: oh, make sure syslib.3i is in the right place
23:07:58 <ais523> if you haven't installed, try copying it into the current directory so it can find it
23:08:08 <ais523> (ick-wrap.c may need to be copied there too)
23:09:08 <Rugxulo> bunch of compilers errors when using "-g"
23:09:55 <ais523> ok, that's more interesting
23:09:57 <ais523> could you paste them?
23:10:08 <Rugxulo> hold on, probably 'cause I'm trying to run inside /pit
23:11:20 <ais523> that shouldn't cause the errors...
23:11:59 <Rugxulo> bah, too many files
23:12:07 <Rugxulo> I don't know what headers and libs need to be where :-/
23:12:19 <ais523> make install should put them all into the right places for you
23:12:30 <ais523> if you don't mind installing
23:12:40 <Rugxulo> doubt it, I've tried that before on other packages, and it didn't work correctly
23:12:43 <ais523> otherwise, it'll look in the current directory, and a couple of binary-relative places
23:12:51 <ais523> but binary-relative is quite possibly broken on DOS
23:12:56 <ais523> due to having a different path separator
23:14:35 <Rugxulo> (trying on a "clean" install of DJGPP)
23:15:06 <Rugxulo> no errors! (-g)
23:15:15 -!- immibis has joined.
23:15:35 <Rugxulo> 11/12/2009 05:15 PM 216,066 hail_mary.exe
23:15:50 <ais523> hmm, it failed on an unclean install, but worked on a clean one?
23:15:53 <ais523> I wonder what the issue was
23:15:54 <Rugxulo> (okay, very very boring example, heh, but it seems to work)
23:16:01 <Rugxulo> no no no
23:16:24 <Rugxulo> I mean I "installed" DJGPP in a separate place (so as not to pollute my main install) and did "make install" there, and it installed ICK correctly where it works
23:16:33 <ais523> ah
23:16:46 <Rugxulo> 1000000 might take a while (snore)
23:17:04 <Rugxulo> doesn't even print the number it's on *sniff*
23:17:46 <Rugxulo> I can already guess a __dpmi_yield() call would be nice ;-)
23:18:01 <Rugxulo> okay, it finished
23:18:11 <ais523> what does __dpmi_yield do?
23:18:26 <ais523> remember, DOS is single-process, so on pure DOS you couldn't yield execution at all, sensibly
23:18:29 <Rugxulo> release the time slice to OS
23:18:34 <Rugxulo> DOS != DPMI
23:18:53 <AnMaster> Rugxulo, hit ctrl-c instead?
23:19:04 <AnMaster> night
23:19:08 <Rugxulo> OS/2, Novell / DR-DOS 7, Win 3.x on up all support it (int 2fh, 168fh) ... technically DPMI 1.0 but all of those are 0.9 only, so ...
23:19:21 <ais523> Rugxulo: you can run DPMI on pure-DOS, though; does it just ignore the call when you do that?
23:19:39 <ais523> I wouldn't mind putting a __dmpi_yield in behind a #ifdef __DJGPP__ barrier or whatever
23:19:49 <Rugxulo> on DR-DOS 7.03? only if not multitasking (which needs its own DPMI server)
23:20:07 <Rugxulo> but yeah, I'm pretty sure all other DOS-only DPMI servers ignore it
23:20:14 <Rugxulo> (would have to check to be sure, though)
23:20:56 <Rugxulo> dpmi.h -> __dpmi_yield()
23:21:25 <Rugxulo> e.g. Bash uses this to not hog cpu time completely
23:22:09 <Rugxulo> but yes, under FreeDOS it's moot (dunno about DOSEMU though, would be vaguely interested to find out)
23:22:17 <Rugxulo> DOSEMU has its own DPMI also
23:22:45 <AnMaster> <Rugxulo> e.g. Bash uses this to not hog cpu time completely <-- what the hell are you on about. If running under windows, wouldn't it do it's own scheduling?
23:22:46 <Rugxulo> but DOSBox has no built-in DPMI server, so you have to use whatever would be used in real DOS
23:23:03 <Rugxulo> AnMaster, NTVDM will gladly hog up a lot of your processor
23:23:10 <ais523> DJGPP comes with a DPMI server, IIRC
23:23:15 <AnMaster> Rugxulo, of half you mean. Dual core :P
23:23:24 <Rugxulo> sure, pre-emptive multitasking is still in control, but it's fairly lenient in letting you hog almost everything else if you wish
23:23:46 <AnMaster> Rugxulo, that's strange. I'm pretty sure the linux one isn't
23:23:46 <fizzie> ais523: DJGPP comes with CWSDPMI, which ignores the __dpmi_yield function.
23:24:01 <ais523> that's what I was thinking of
23:24:07 <Rugxulo> DJGPP is not designed only for CWSDPMI, though, but for any compliant DPMI 0.9 server
23:24:24 <Rugxulo> CWSDPMI is just for those who don't have any other
23:24:29 <fizzie> The 0x2f interrupt handler only handles function 1686h, "Get CPU Mode".
23:24:31 <AnMaster> ais523, why not port ick to classic mac os?
23:24:35 <AnMaster> or has that already been done?
23:24:44 <ais523> AnMaster: port? in theory it should run unmodified on anything without porting
23:24:46 <ais523> ofc, that's just theory
23:24:57 <AnMaster> ais523, you need to do something about the console
23:25:01 <AnMaster> like, emulate one yourself
23:25:01 <ais523> so if classic mac is around, it would be nice if someone would try it and let me know where it fails
23:25:19 <ais523> AnMaster: well, if it doesn't have any sh-equivalent, then a new build system would be needed
23:25:20 <AnMaster> ais523, I have one emulator with MPW I could try tomorrow. Runs OS 9/PPC.
23:25:48 <Rugxulo> you could probably hack your own config.h if needed
23:25:50 <AnMaster> ais523, indeed it doesn't. Well MPW has it's own weird shell-like thingy. You select some text press a key combo to execute it.
23:26:02 <AnMaster> and the make isn't the usual syntax
23:26:03 <AnMaster> at all
23:26:06 <ais523> presumably it doesn't do sh commands
23:26:17 <AnMaster> MPW make just outputs a series of commands you can run
23:26:30 <ais523> simply tracking which commands make calls and putting them into a text file would be enough to create a relatively universal build system
23:26:38 <ais523> if, as you say, config.h was set correctly
23:26:44 <AnMaster> ais523, the C compiler is called MrC iirc
23:26:49 <AnMaster> cc won't work
23:26:59 <ais523> symlinks can fake that easily enough
23:27:08 <ais523> on the system you use to make the makescript
23:27:15 <Rugxulo> AnMaster, currently Task Manager says anywhere from "89%" used by hail_mary.exe to "95%" or "97%" or "99%" etc. (it varies)
23:27:18 <AnMaster> ais523, still. It can't do STDOUT easily. As it will go into the void
23:27:27 <AnMaster> ais523, you need to make your own window to draw it in
23:27:30 <ais523> you could freopen it to a file
23:27:32 <Rugxulo> okay, done again, now to see how many lines in that file I redirected ;-)
23:27:34 <AnMaster> ais523, or that
23:27:52 <Rugxulo> AnMaster, stuff like that has been done before, e.g. so-called Win16 extender (OpenWatcom)
23:28:01 <AnMaster> ais523, path separator on mac is :
23:28:04 <AnMaster> not / or \
23:28:16 <Rugxulo> or RSX's shell (rshell or whatever it was called)
23:28:20 <ais523> that isn't a problem if everything's in the same directory
23:28:49 <Rugxulo> yup, million lines, so it worked!
23:29:08 <Rugxulo> oy, 11 MB file ;-)
23:29:30 <Rugxulo> compresses to 21k .ZIP (unsurprisingly)
23:29:41 <AnMaster> ais523, oh and the MrC command has a completely weird syntax iirc. Will take a look at this tomorrow. Should be fun. I'm not even sure there is a full standard library there. Or if there is, how you link to it
23:29:42 <AnMaster> as in
23:29:51 <ais523> AnMaster: is it C89-compliant?
23:29:57 <AnMaster> ais523, *maaaaybe*
23:30:05 <Rugxulo> K&R, probably
23:30:18 <AnMaster> ais523, http://en.wikipedia.org/wiki/Macintosh_Programmer%27s_Workshop
23:30:29 <AnMaster> Rugxulo, it has some limited C++ support iirc. Oh and Pascal
23:30:59 <Rugxulo> ha! Interfunge prints output in Roman numerals ... that I didn't expect (and I knew Intercal loved 'em!)
23:31:01 <ais523> AnMaster: apparently it supports stdout if you redirected it
23:31:08 <AnMaster> hm
23:31:31 <Rugxulo> is there no GCC port for Macs?
23:31:40 <AnMaster> Rugxulo, for classic ones? I'd doubt it
23:31:49 <AnMaster> or discontinued if so
23:31:50 <ais523> apparently libiberty specifically had support for MPW
23:31:51 <Rugxulo> didn't GCC start out on 68000? also, Atari has one
23:31:52 <ais523> so there was for a while
23:31:57 <AnMaster> ais523, heh
23:32:16 <Rugxulo> there are people who have GCC 3.x running for Atari cpus, I think
23:32:41 <Rugxulo> they ported Quake to some Atari machine (Falcon 68060?)
23:33:27 <Rugxulo> (and Quake was originally cross-compiled from Alphas?? for DOS / DJGPP target, C + some FPU-heavy ASM)
23:34:00 <ais523> anyway, I really need to go home
23:34:03 -!- ais523 has quit (Remote closed the connection).
23:34:23 <Rugxulo> uh, ais523: thanks (post mortem) :-)
23:35:10 <AnMaster> :P
23:37:05 <AnMaster> Rugxulo, btw if you get the cfunge-intercal thing (IFFI) to work on that I will be extremely impressed. Considering that cfunge is a pain to get working even on cygwin (and even then some parts don't work)
23:37:30 <Rugxulo> well, I doubt it since you said it needed mmap, and that's unsupported in DJGPP
23:37:44 <Rugxulo> (without DPMI 1.0 fiddling, which I'm not intimately familiar with)
23:37:51 <AnMaster> Rugxulo, right. You could rewrite file loading code I guess.
23:37:54 <AnMaster> a lot
23:38:11 <AnMaster> (that won't go upstream though)
23:38:39 <Rugxulo> well, FBBI compiles in DJGPP
23:38:54 <Rugxulo> CCBI might work under HX (haven't tried)
23:38:58 <AnMaster> HX?
23:39:07 <Rugxulo> http://www.japheth.de/HX.html
23:39:13 <Rugxulo> not everything is supported (obviously)
23:39:17 <AnMaster> Rugxulo, cfunge is C99 + quite a lot POSIX
23:39:24 <Rugxulo> I know
23:39:35 <Rugxulo> DJGPP has GCC 4.4.1 and supports a fair subset of POSIX
23:39:42 <AnMaster> 4.4? not bad
23:39:52 <AnMaster> Rugxulo, wait, is it 32-bit?
23:39:58 <Rugxulo> yes
23:40:02 <AnMaster> ah good
23:40:09 <AnMaster> cfunge wouldn't be happy with 640k
23:40:10 <Rugxulo> GCC doesn't officially support anything less on x86
23:40:29 <Rugxulo> Quake needed 8 MB of RAM (1996), and it used DJGPP
23:40:44 <Rugxulo> heck, GCC itself would never ever fit in 640k
23:41:06 <Rugxulo> okay, I take it back ... maybe with HEAVY swapping (CWSDPMI)
23:41:18 <Rugxulo> DJ did start it in 1989 on his 386 w/ 2 MB of extended RAM
23:41:25 <Rugxulo> of course GCC was loads simpler then
23:41:49 <AnMaster> Rugxulo, iirc with loading *.so, setting up memory pools (large for performance reasons) and so on cfunge needs something like 15 MB virtual address space for the ulimit on my x86_64 system
23:41:56 <AnMaster> or was it more? I forgot
23:42:16 <AnMaster> Rugxulo, DJ?
23:42:30 <Rugxulo> DJ = founder of DJGPP (DJ's GNU Programming Platform)
23:42:36 <AnMaster> ah
23:43:05 <Rugxulo> and BTW, memory limits are imposed by DPMI host, not DJGPP (although I think 2 GB is somehow hardcoded somewhat in libc but can be worked around ... never completed because nobody had interest, according to CWS)
23:43:21 <AnMaster> not surprised
23:43:31 <Rugxulo> you can't build GCC anymore in less than 128 MB or such
23:43:43 <AnMaster> Rugxulo, not surprised either
23:43:46 <Rugxulo> used to be where 16 MB was enough (2.x)
23:44:07 <AnMaster> Rugxulo, anyway, good luck getting cmake working in there (alas, not needed for IFFI, IFFI builds it differently)
23:44:17 <Rugxulo> IIRC, DJ said (1996) he could build GCC in thirty minutes on his 486 DX/100
23:44:36 <Rugxulo> one guy wanted to port Cmake but stopped due to Bash bug
23:44:54 <Rugxulo> same bug cropped up when Autoconf released 2.64 (nobody tested, gah), now fixed ... but a bit too late (five years)
23:45:43 <Rugxulo> doesn't help that MS never fixed NT bugs in their DPMI server, etc.
23:45:59 <Rugxulo> nor that hardware acceleration almost never worked in DOS (does 3dfx count? barely ...), which distracted many a developer
23:46:11 <Rugxulo> e.g. Quake / DOS never ran in NT (only 9x)
23:46:15 <AnMaster> Rugxulo, why does anyone care
23:46:22 <AnMaster> dosbox, sure
23:46:28 <Rugxulo> heh, sorry, thought you might wanna know for trivia ;-)
23:46:36 <AnMaster> but DJGPP these days?
23:46:44 -!- augur has quit (Read error: 110 (Connection timed out)).
23:46:47 <Rugxulo> *shrug* why not, if it works?
23:47:05 <AnMaster> Rugxulo, you could use that to defend cobol too
23:47:14 <Rugxulo> ... or Befunge98 ;-)
23:47:26 <AnMaster> Rugxulo, or anything
23:47:27 <Rugxulo> *zing*
23:47:32 <Rugxulo> lol ;-)
23:47:35 <AnMaster> which was my point
23:47:43 <AnMaster> that that reason makes no sense
23:47:56 <Rugxulo> the point is ... DJGPP works well for what it does, so why not use it?
23:48:10 <AnMaster> Rugxulo, also, tell me when you rewrote fungot in befunge93
23:48:10 <fungot> AnMaster: and it was to late yeah it is it you don't have anything and was being assesed on my face he.d be in movies like this
23:48:20 <Rugxulo> heh
23:48:22 <AnMaster> ^style
23:48:22 <fungot> Available: agora alice c64 ct darwin discworld europarl ff7 fisher ic irc jargon lovecraft nethack pa speeches ss wp youtube*
23:48:31 <AnMaster> ^style nethack
23:48:31 <fungot> Selected style: nethack (NetHack 3.4.3 data.base, rumors.tru, rumors.fal)
23:48:37 <AnMaster> fungot, .
23:48:37 <fungot> AnMaster: eating a yeti. ceram)
23:48:44 <AnMaster> fungot, not a good idea...
23:48:49 <Rugxulo> fungot Dungeon Crawl: Stone Soup for the win!!
23:48:50 <fungot> Rugxulo: c*ckatrice: once you know what that is?)
23:48:57 <AnMaster> Rugxulo, no way
23:49:02 * Rugxulo throws a cockatrice at fungot
23:49:02 <fungot> Rugxulo: they say that blue stones are radioactive, beware. " the eyes!" " how the indian deity quetzalcoatl came from the fiends, that david hasted, and he had no wallet and no pockets, so that they might challenge his own sword, lion's skin, and damage, both spell and physical, is about the time that paris was full of carlists who had brought. for a price list.
23:49:14 * Rugxulo while wearing gloves, of course
23:49:22 <AnMaster> Rugxulo, too late to mention that now
23:49:36 <Rugxulo> to him, maybe ;-)
23:49:41 <AnMaster> Rugxulo, to you
23:49:49 <AnMaster> fungot, good thing you are alive
23:49:49 <fungot> AnMaster: zruty: the name _sasquatch_ doesn't really become important in canada until the 1930s, when because of the truly great unknown animals of the giant farbauti and of themselves, their ragged, unfurry skin is as white as a god of light who was next me; and in the motion of the twilight they could find it, on those marvelous birthday mornings, he would peel back a tiny bit of a horse. ( conan the usurper, by david gerrold
23:49:51 <Rugxulo> why would I mention it to myself? :-)
23:49:52 <fungot> A|t{u~v[X7=Ψ:Pzno5xp}y8/f1liated/anmaster: they say that applying a unicorn horn rule is if it was a serviette he had no wallet and no pockets, so you can be almost instantly fatal. its streaming mane kindled, and is looked upon as the wolf runs," mom said. " it can catch.
23:50:11 <AnMaster> Rugxulo, because you are insane due to crawl?
23:50:12 <AnMaster> ;P
23:50:22 <AnMaster> wait
23:50:23 <Rugxulo> nah, Crawl is fun
23:50:25 <AnMaster> fizzie, !!!!!
23:50:30 <AnMaster> fizzie, fungot bug above
23:50:30 <fungot> AnMaster: you swallowed the fortune cookie is the son of the sun god of fire.
23:50:35 <Rugxulo> BTW, NetHack compiles under DJGPP (as does SLASH 'EM and Falcon's Eye)
23:50:37 <AnMaster> fizzie, dual highlight thing again
23:50:50 <AnMaster> fizzie, with the corrupted data
23:50:59 <AnMaster> happened once before as you may remember
23:51:08 <Rugxulo> as does Dungeon Crawl: Stone Soup (and the original)
23:51:21 <AnMaster> Rugxulo, what about X?
23:51:22 <AnMaster> ;P
23:51:30 <AnMaster> X runs under cygwin
23:51:33 <Rugxulo> there was a Xlibemu at one time, but v1 only
23:51:36 <AnMaster> well, a ported X I assume
23:51:41 <AnMaster> Rugxulo, v1 of?
23:51:44 <Rugxulo> actually, I forgot about Desqview/X
23:51:51 <Rugxulo> but that's proprietary, old, and v1 only also
23:51:54 <Rugxulo> v1 of DJGPP
23:51:55 <Rugxulo> (pre 1996)
23:51:57 <AnMaster> ah
23:52:05 <AnMaster> Rugxulo, ABI breakage?
23:52:06 <Rugxulo> v2 is DPMI only, LFNs out-of-the-box
23:52:14 <AnMaster> LFN?
23:52:18 <Rugxulo> v1 was GO32.EXE extender, supported VCPI or DPMI etc., no LFNs
23:52:24 <Rugxulo> LFN = more than 8.3
23:52:26 <AnMaster> VCPI?
23:52:30 <AnMaster> Rugxulo, 8.3 of?
23:52:34 <AnMaster> oh filenames
23:52:35 <Rugxulo> VCPI = DOS extenders before DPMI existed
23:52:56 <Rugxulo> which Windows stopped supporting (directly) in Win 3.0
23:53:22 <AnMaster> Rugxulo, and all my systems are 64-bit. So I don't need to bother about even seeing that mess any more
23:53:24 <Rugxulo> only Win 3.x /S ("standard" / 286) mode supported VCPI (which was ironically an extension of EMS and needed a 386)
23:53:29 * immibis notices anmaster is sleep-IRC'ing
23:53:33 <Rugxulo> what mess?
23:53:40 <Rugxulo> you don't have to see anything if you close your eyes ;-)
23:53:43 <AnMaster> immibis, failing to get away from the computer when I should
23:53:46 <AnMaster> :P
23:54:01 <AnMaster> Rugxulo, mess? Well. DPMI. The yield stuff. Lots of more?
23:54:11 <Rugxulo> BTW, DOSEMU works pretty well under x86-64 (for most DJGPP apps, anyways)
23:54:36 <Rugxulo> AnMaster, it's supposed to be transparent, you shouldn't have to think of the low-level details
23:54:41 <AnMaster> Rugxulo, most of them probably work natively under linux anyway. So I fail to see the point.
23:54:43 <Rugxulo> ... unless you want to
23:54:57 <Rugxulo> natively IF somebody compiled them for you
23:55:01 <AnMaster> Rugxulo, no proper memory protection. No fork(). Lots of more
23:55:09 <Rugxulo> it has memory protection
23:55:11 <AnMaster> Rugxulo, um. I have make :P
23:55:20 <AnMaster> Rugxulo, oh really? NX too?
23:55:31 <Rugxulo> no
23:55:50 <AnMaster> Rugxulo, can you make an app not able to break out?
23:55:58 <AnMaster> like, even proper *nix like security
23:56:25 <Rugxulo> maybe, maybe not
23:56:35 <AnMaster> Rugxulo, what about file permissions?
23:56:35 <Rugxulo> I don't know for sure, honestly
23:57:04 <Rugxulo> DR-DOS supports passwords and weak partition encryption, but I assume that's not what you mean
23:57:18 <AnMaster> Rugxulo, I meant chmod
23:57:20 <AnMaster> and chown
23:57:29 <AnMaster> multiuser system basically
23:57:35 <Rugxulo> DOS inherently doesn't natively support anything more than +a +r +s +h
23:57:46 <Rugxulo> there are multiuser DOSes, but I've never used 'em (very rare)
23:57:48 <AnMaster> Rugxulo, I'm not familiar with those
23:57:56 <Rugxulo> +a = archive bit
23:57:58 <Rugxulo> +s = system bit
23:58:00 <Rugxulo> + h = hidden
23:58:03 <Rugxulo> +r = read only
23:58:11 <AnMaster> mhm
23:58:18 <AnMaster> why +a?
23:58:27 <AnMaster> wouldn't comparing timestamp work
23:58:37 <Rugxulo> timestamps can change, I guess
23:58:45 <Rugxulo> I don't know the official reason, honestly, and nobody uses it anymore
23:58:50 <AnMaster> Rugxulo, so can that bit, otherwise useless
23:59:40 <immibis> if a backup is interrupted halfway through, the next run will still backup the other half because their A bits are still set?
←2009-11-11 2009-11-12 2009-11-13β†’ ↑2009 ↑all