Wiki

Clone wiki

SCons / BugParty / IrcLog2008-06-30

18:51:57 * stevenknight (n=stevenkn@c-69-181-234-155.hsd1.ca.comcast.net) has joined #scons 19:00:21 <stevenknight> hello, anyone else here for bugs? 19:01:28 * garyo-home (n=chatzill@209-6-158-38.c3-0.smr-ubr3.sbo-smr.ma.cable.rcn.com) has joined #scons 19:02:19 <garyo-home> Hi, folks. Thought I'd log in & do some of my bug homework, but now I see it's tonight! 19:02:42 <stevenknight> hi gary 19:02:45 <stevenknight> just you and me so far 19:02:55 <stevenknight> no thanks to my confusion about days... :-/ 19:02:56 <garyo-home> OK; let me get my windows set up. 19:03:15 <garyo-home> yah, I thought it was going to be tomorrow, oh well, in some ways this is better. 19:03:34 <stevenknight> hopefully greg will have seen the reply and show up as well 19:03:50 <stevenknight> if not we need to decide if we go ahead just us two or not 19:03:54 <garyo-home> yes, his msg was only 1.5 hrs ago 19:04:14 <garyo-home> I think two is not a quorum, though we could do some obvious ones anyway... 19:04:29 <stevenknight> true, just clear out the obvious consensus 19:04:32 <stevenknight> that's still valuable 19:05:04 <GregNoel> Hi, I'm here, but not set up yet; give me a minute 19:05:09 <garyo-home> Hi, Greg! 19:05:41 <stevenknight> np, take your time 19:09:14 <garyo-home> Sounds like the scons dinners have been fun. 19:09:18 <GregNoel> OK, I'm up 19:09:38 <garyo-home> OK, shall we dive into the current issues then? 19:09:47 <GregNoel> I'm ready 19:10:00 <stevenknight> okay, 2098: consensus 19:10:06 <garyo-home> 2098: who should integrate? Steven? 19:10:07 <stevenknight> 1.x p3 19:10:12 <stevenknight> yes, me 19:10:14 <GregNoel> done 19:10:18 <stevenknight> 2114: 19:10:37 <stevenknight> 1.0x p2 david 19:10:46 <garyo-home> sounds right. 19:10:52 <GregNoel> works for me 19:10:53 <stevenknight> done 19:10:59 <stevenknight> 2115: 19:11:17 <stevenknight> any objections to 1.x p3? 19:11:28 <garyo-home> for doing it as its own separate task? 19:11:42 <stevenknight> say more 19:11:45 <garyo-home> i.e. just making sconsign understand that special case 19:11:48 <GregNoel> it's the default; we'll have to triage those again, but it's fine. 19:12:03 <stevenknight> yes, by default 19:12:06 <garyo-home> ok, fine. 19:12:23 * garyo-home avoids long sconf discussion 19:12:25 <stevenknight> all right, 1.x p3 19:12:48 <stevenknight> ah, right -- i get it 19:13:00 <stevenknight> yes, not as part of the whole big SConf brouhaha on the MLs right now 19:12:58 <GregNoel> sigh, things get out of control when one is gone; there's a silverfish crawling across my desk... 19:13:20 <garyo-home> greg: gross! 19:13:30 <GregNoel> very 19:13:45 <garyo-home> ok, on to 2116? 2116: I agree w/ you guys. 19:13:52 <stevenknight> 2116: 1.0x p2 consensus 19:13:56 <stevenknight> Benoit 19:14:01 <GregNoel> done 19:14:16 <stevenknight> 2117: 19:14:32 <garyo-home> I don't think scons should delete anything read-only. 19:14:36 <stevenknight> i kind of like greg's classification, actually... :-) 19:14:40 <garyo-home> :-) 19:14:45 <GregNoel> er, it already does.... 19:14:52 <stevenknight> yes 19:14:54 <garyo-home> greg: yes, you're right. 19:15:05 <garyo-home> ... e.g. before building. 19:15:10 <stevenknight> i don't agree w/his solution (make it writable silently) 19:15:21 <stevenknight> but would want some configurability / option that permits it 19:15:23 <garyo-home> steven: I agree, it's rude & could have bad consequences. 19:15:33 <GregNoel> actually, I was surprised that SCons deletes files before rebuilding them, but that's another discussion 19:16:03 <garyo-home> so can we just say wontfix? 19:16:19 <stevenknight> i'd rather turn it into a feature request for the configurability 19:16:36 <garyo-home> OK, 2.x p3 feature req would be OK by me 19:16:42 <stevenknight> i can go with 2.x 19:16:48 <GregNoel> done 19:16:55 <stevenknight> on reflection, it is definitely lower priority than other 1.x stuff 19:17:00 <stevenknight> 2119: 19:17:20 <stevenknight> consensus 1.0.x p2 19:17:30 <garyo-home> sure. 19:17:28 <stevenknight> i'm definitely going to fix this soon for my own purposes 19:17:43 <stevenknight> like, tomorrow 19:17:58 <stevenknight> (but not check it into branches/core yet) 19:17:58 <GregNoel> works for me 19:18:00 <garyo-home> OK. 19:18:11 <stevenknight> okay, on to 2006h2? 19:18:21 <garyo-home> ok, I'm there. 19:18:27 * GregNoel has visitors at the door 19:18:34 <garyo-home> 1437, consensus 19:18:49 <stevenknight> yes, 1437 dup 19:18:52 <stevenknight> done 19:18:57 * garyo-home needs a drink, brb 19:19:27 <stevenknight> 1438: 19:19:39 <stevenknight> consensus 1.x p3 me 19:19:46 <stevenknight> i can go w/you guys on the time frame 19:19:58 * stevenknight whistles aimlessly while waiting for everyone else to return... 19:20:06 <garyo-home> hi, I'm back 19:20:18 <garyo-home> 1438 1.x p3 is fine w/ me. 19:20:26 <garyo-home> There's plenty to do before then. 19:20:50 <stevenknight> done 19:21:01 <stevenknight> 1439: i can go w/invalid 19:21:23 <bdbaddog> which spreadsheet are you guys on now? 19:21:24 <stevenknight> i was thinking research because it is kind of a pain to hook up new builders 19:21:28 <stevenknight> hey bill 19:21:30 <garyo-home> 1439: and tell him to use src_builder? I guess after this long he probably doesn't care anymore... 19:21:30 <stevenknight> 2006h2 19:21:47 <stevenknight> sure, as a courtesy for closing it out 19:21:48 <bdbaddog> Hey. I'll just be here a few, but I'll add what I can. 19:21:55 <garyo-home> Hi Bill. 19:21:57 <stevenknight> bdbaddog: cool 19:22:19 <stevenknight> so 1439: invalid, point him to src_builder 19:22:35 <garyo-home> re 1439: I'd like a new ticket for making adding src builders easier. 19:22:38 <stevenknight> i'd still like another issue for some feature (API extension?) to make it easier to hook up your own builders to our existing ones 19:22:42 <garyo-home> +1 19:22:47 <stevenknight> +1 19:22:58 <bdbaddog> Like AddToCBuilder? 19:23:20 * stevenknight applauds garyo-home's ability to put things much more economically 19:23:31 <garyo-home> bdbaddog: Maybe, but let's not design it now, just make a ticket for later. 19:23:33 <stevenknight> bdbaddog: something like that 19:23:54 <stevenknight> i'd genericize it somehow (wave hands mumble mumble) 19:24:03 <garyo-home> yes. 19:24:12 <bdbaddog> :) yeah. that's probably why it's not done already. 19:24:43 <garyo-home> 1442, folks? 19:24:46 <stevenknight> 1442: sounds like greg's right 19:25:02 <garyo-home> In that case, maybe it's already better due to David's stuff? 19:25:29 <stevenknight> maybe 19:25:33 <garyo-home> Anyone have a mingw env? 19:26:01 <garyo-home> ok, guess not. 19:26:08 <stevenknight> not me 19:26:23 <stevenknight> i really want to set up buildbots with the more common windows configs 19:26:28 <GregNoel> 1439: concur (I'm back, BTW) 19:26:30 <stevenknight> one for MinGW, one for Cygwin, etc. 19:27:36 <bdbaddog> looks like I have cygwin with mingw-g77 installed. 19:27:48 <stevenknight> the more i think about it, the more 1442 seems like a really interesting case 19:27:57 <GregNoel> how so? 19:28:04 <stevenknight> conceptually i agree w/Greg's analysis that .f is clearly an error 19:28:21 <stevenknight> but if so, how would you specify the weird corner case where you really did want to archive .f files? 19:28:46 <stevenknight> after all, there's no reason why you should be prohibited from doing that 19:29:00 <stevenknight> just because there's a more common use case of .f files generating .o files 19:29:13 <GregNoel> You need an "ar" builder, not a library builder 19:29:13 <garyo-home> ... or .c files for that matter. Maybe File nodes would do it? 19:29:33 <stevenknight> hmm, interesting distinction 19:29:36 <bdbaddog> doesn mingw builder setup fortran at all? 19:29:44 <GregNoel> The archive builders take any file suffix 19:30:26 <stevenknight> hmm, i think Gary's right -- File nodes circumvent the suffix checking 19:30:41 <bdbaddog> nope. mingw sets up the following: 19:30:41 <bdbaddog> gnu_tools = ['gcc', 'g++', 'gnulink', 'ar', 'gas', 'm4'] 19:30:42 <GregNoel> you sure? 19:30:49 <stevenknight> nope 19:31:18 <stevenknight> but i am worried that the distinction between a "library" (a .a file with objects) and an "archive" (the same suffix but with different contents) would be really subtle and easily lost 19:32:02 <garyo-home> It's a pretty atypical case though. 19:32:05 <GregNoel> Uh, "ar" archives don't have a .a suffix; that's only for libraries. 19:32:30 <GregNoel> The suffix is usually .ar or none at all 19:32:58 <garyo-home> greg: never seen such a thing myself. 19:33:01 <stevenknight> yeah, the case is atypical 19:33:10 <stevenknight> but i think the potential for confusion remains 19:33:19 <stevenknight> if there is more than one builder that causes "ar" to be invoked 19:33:48 <bdbaddog> has anyone run into a build which used ar for things other than static libraries? 19:33:50 <garyo-home> They could always use Command() if Library() doesn't do what they want. 19:33:51 <GregNoel> Really? More than one builder causes 'gcc' to be invoked... 19:34:11 <stevenknight> hmm, fair point. i'm probably worrying needlessly 19:34:21 <garyo-home> I think so :-) 19:34:25 <GregNoel> bdbaddog, you're not old enough; the evolution was the other way around 19:35:04 <GregNoel> 'ar' was used to build archives; eventually, archives of .o files were acceptable to the linker 19:35:15 <garyo-home> So where does that leave 1442? 19:35:41 <stevenknight> dup 19:35:45 <GregNoel> dup 19:35:48 <garyo-home> ok. 19:35:56 <bdbaddog> dup of ? 19:36:04 <bdbaddog> mingw builder doesn't setup g77... 19:36:09 <garyo-home> 1437 says the ssheet. 19:36:34 <GregNoel> That's "better error messages when tool not configured" 19:37:10 <stevenknight> ah, but bill's point is good: our default doesn't even make this possible 19:37:17 <bdbaddog> ahh. o.k. well it's a dup and also it's mingw doesn't setup g77 though. 19:38:07 <bdbaddog> o.k. gents. I may have a patch for that. I've gotta head out for a while though. I'll shoot an email later with pach if I get it working. 19:38:20 <garyo-home> There's also 1895, g77 and gfortran not detected on windows which is about mingw. 19:39:21 <garyo-home> Seems like 1442 could be a dup of 1895. 19:39:49 <GregNoel> (other way around; use the earlier issue as the basis) 19:40:08 <garyo-home> greg: you're right. 19:40:59 <garyo-home> 1895 is research, p3, david. So mark 1895 as dup of 1442, and make 1442 research, p3, david. 19:41:18 <GregNoel> done; 1443? 19:41:58 <garyo-home> There is now a SHFORTRANFLAGS, so I presume it could get set to /fPIC if appropriate. 19:42:13 <GregNoel> oops, brb 19:42:36 <garyo-home> But it's clearly David's if it's still broken. 1.0.x p3 David? 19:45:03 <garyo-home> Can we just give 1449 and 1450 to Jim, 1.x p3, keyword "quoting"? 19:45:34 <garyo-home> h'lo? 19:46:06 * sgk_ (n=stevenkn@c-69-181-234-155.hsd1.ca.comcast.net) has joined #scons 19:46:23 <sgk_> hmm, looks like the server i was connected to died 19:46:29 <garyo-home> Hi again, didn't even see you drop out. 19:46:31 <sgk_> what was the last you got from me? 19:46:41 <garyo-home> "our default doesn't even make this possible" 19:47:05 <sgk_> so maybe it's a one liner of 'g77' (or more likely 'gfortran') to the mingw tool list 19:47:26 <sgk_> anyone object to that solution for... 1.x p3? 19:47:33 <garyo-home> Yes, just after you dropped out bdbaddog said: "o.k. gents. I may have a patch for that. I've gotta head out for a while though. I'll shoot an email later with pach if I get it working." 19:47:53 <garyo-home> If it's trivial, could be in 1.0.x, right? 19:48:00 <sgk_> yeah 19:48:06 <sgk_> 1.0.x p3? 19:48:09 <garyo-home> OK. 19:48:12 <sgk_> done 19:48:19 <garyo-home> Can we just give 1449 and 1450 to Jim, 1.x p3, keyword "quoting"? 19:48:34 <sgk_> +1 19:48:39 <garyo-home> sorry, "research" ? 19:49:02 <garyo-home> research was the ssheet consensus but I kind of think it's 1.x timeframe stuff. 19:49:06 <sgk_> sure, research 19:49:14 <sgk_> research p3 "quoting" 19:49:20 <sgk_> done 19:49:22 <GregNoel> dkjak 19:49:23 <garyo-home> good. 19:49:30 <sgk_> 1452: 19:49:46 <sgk_> 1.x p3 me 19:49:48 <garyo-home> sounds like that one's yours, Steven. 19:49:58 <garyo-home> ok, done. 19:50:06 <GregNoel> yes 19:50:15 <sgk_> 1456: research, me "VisualStudio" 19:50:55 <garyo-home> maybe, but I like "invalid" -- can't just have msvc as the only tool. 19:51:21 <GregNoel> true, but a better message would help 19:51:47 <garyo-home> fine, in that case it's dup of the "better errors" one, not VisualStudio. 19:51:51 <sgk_> oh, hey, even better 19:52:04 <sgk_> that makes it a toolchain issue and i can give it to you guys... :-) 19:52:18 <GregNoel> that's why I said dup 1437, better messages 19:52:20 <sgk_> invalid is good, though 19:52:31 <sgk_> either one is fine w/me 19:52:41 <GregNoel> Let's go with dup 19:52:53 <garyo-home> Yes, that gives the OP more info. 19:53:00 <sgk_> ok, dup 1437 19:53:03 <GregNoel> done 19:53:23 <sgk_> 1458: dup 1437 as well 19:53:24 <garyo-home> 1458, same. 19:53:27 <GregNoel> yes 19:53:54 <garyo-home> 1459, does Ludwig have a prototype of this already? 19:53:55 <sgk_> 1459: Ludwig 19:54:10 <garyo-home> +1 19:54:18 <GregNoel> He was looking at it; I think he has an idea 19:54:44 <garyo-home> Good. 19:54:44 <GregNoel> His experiments showed very little impact due to the size of buffer 19:54:54 * stevenknight has quit (Read error: 110 (Connection timed out)) 19:55:12 <sgk_> Ludwig, any appropriate target milestone + priority 19:55:18 <GregNoel> done 19:55:59 <GregNoel> 1460 19:56:00 <sgk_> 1460: i'm agnostic 19:56:20 <GregNoel> interesting choice of word... 19:56:50 <garyo-home> I can take it. Any time; 1.0.x p3? 19:56:56 <sgk_> works for me 19:57:07 <GregNoel> works 19:57:09 <garyo-home> ok. 19:57:22 <sgk_> 1462: worksforme 19:57:26 <sgk_> i'm a dual core 19:57:39 <sgk_> it can be re-opened if it's still a problem elsewhere 19:57:40 <garyo-home> Sounds like you tried pretty hard to repro it. 19:57:48 <garyo-home> worksforme works for me. 19:57:50 <sgk_> but my guess is some of Benoit's Taskmaster changes have fixed it 19:57:53 <sgk_> :-) 19:58:11 <GregNoel> That would be my guess as well 19:58:26 <garyo-home> 1464: agree, wontfix. 19:58:33 <sgk_> done 19:58:46 <GregNoel> done 19:58:58 <sgk_> 1466: me, research, VisualStudio 19:59:15 <sgk_> agree w/Greg that it might end up in toolchain, but i'm happy to be stuck with it in the meantime 19:59:14 <GregNoel> OK, worst case you toss it to us. 19:59:21 <sgk_> yes 19:59:33 <garyo-home> ok. 19:59:40 <sgk_> 1468: 1.0.x p2? 19:59:57 <GregNoel> at least 20:00:02 <sgk_> sounds pretty serious 20:00:15 <garyo-home> People must be working around it. 20:00:26 <GregNoel> working by blind luck is another way of saying pretty serious... 20:00:33 <sgk_> yeah 20:00:37 <sgk_> 1.0.x p1? 20:00:43 <garyo-home> ok w/ me. 20:00:48 <GregNoel> yes 20:00:52 <sgk_> done 20:01:21 <sgk_> 1469: d'oh! wonfix 20:01:24 <sgk_> wontfix 20:01:25 <sgk_> greg is right 20:01:32 <garyo-home> agreed. 20:01:36 <GregNoel> done 20:01:41 <sgk_> 1471: closed 20:01:43 <sgk_> 1476: 20:01:57 <sgk_> research, me 20:02:13 <sgk_> (sorry, trying to get through these quick, I have to start winding down) 20:02:30 <sgk_> 1478: research, me, VisualStudio 20:02:34 <garyo-home> Me too. OK, 1476 is yours. 20:02:48 <sgk_> 1478: gary, fixed? 20:02:48 <GregNoel> 1476: you're welcome to it 20:02:53 <sgk_> 1483: gary, fixed? 20:03:05 <garyo-home> 1478: hopefully will be overtaken by vsvars.bat stuff 20:03:31 <sgk_> yes re: 1478 & vsvars.bat 20:03:51 <sgk_> 1488: 1.x p3 me 20:04:07 <garyo-home> 1483: yes, I consider that fixed. 20:04:44 <sgk_> cool 20:04:49 <GregNoel> Now would be a good time for me to quit; our guests are watching the football game... 20:05:01 <garyo-home> I should go too. Next week? 20:05:03 <sgk_> and i have a dog that needs walking and won't wait 20:05:09 <sgk_> same time? 20:05:13 <garyo-home> OK for me. 20:05:17 <GregNoel> done 20:05:23 <GregNoel> later, all... 20:05:24 <garyo-home> good, bye for now! 20:05:28 <sgk_> done (and i'll remember Monday night this time...) 20:05:29 <sgk_> later.. 20:05:36 * sgk_ has quit ("Leaving") 20:05:45 * garyo-home has quit ("ChatZilla 0.9.83 [Firefox 3.0/2008052906]")

Updated