IrcLog2010 04 27

16:35:36 * Jason_at_Intel (chatzilla@12.18.240.224) has joined #SCONS 16:54:01 * garyo (garyo@209-6-36-50.c3-0.smr-ubr1.sbo-smr.ma.cable.rcn.com) has joined #SCONS 16:54:41 * bdbaddog (bdeegan@adsl-71-131-8-164.dsl.sntc01.pacbell.net) has joined #SCONS 16:58:10 * GregNoel is here and getting set up, another couple of minutes... 16:58:26 * sgk (sgk@nat/google/x-vtdxqwybqnmycvwp) has joined #SCONS 16:58:28 * GregNoel is no longer marked as being away 16:59:36 All, Jason & I've been having some toolchain discussions offline. Hopefully I can write something up and/or Jason can implement something around it soon... based around iapat ideas mostly. 17:01:23 OK, shall we dive into the bug list? 17:01:25 <GregNoel> Hi, everybody... Looks like a quorum, shall we start? 17:01:41 sure, any sign of greg yet? (he just leaves himself logged in usually) 17:01:48 oh, there you are 17:01:49 ! 17:01:50 He's really here 17:01:55 cool 17:02:00 <Jason_at_Intel> he needed a minutes to setup 17:02:08 let's go then 17:02:11 <GregNoel> 2581 17:02:48 Seems like there's not a lot we can do about that in the near term, right? 17:02:55 <GregNoel> concur 17:02:58 i don't think it's a quick fix 17:03:03 so post-2.2, probably 17:03:16 <Jason_at_Intel> agreeded 17:03:34 2.x p3 sk then? (With option to punt further if needed) 17:03:39 <GregNoel> I'm even inclined to say it's post 2.x 17:03:40 2.x p4? i'm pretty sure re-ordering the lines is a valid workaround, lousy though that is 17:03:50 sgk: my thought too. 17:04:15 <GregNoel> I'll go along, even though 2.x is overfull 17:04:30 p4 makes it puntable 17:04:38 wouldn't surprise me if all 2.x p4 get pushed out when we re-evaluate them 17:04:55 but it at least makes sure we take a look when it comes time to reprioritize 17:04:59 <GregNoel> ok, 2.x p4 sk 17:05:05 done 17:05:06 good 17:05:12 <GregNoel> 2609 17:05:29 no response yet, defer 17:05:34 agree 17:05:35 <GregNoel> Are we still waiting for more info? 17:05:42 i think so, yes 17:05:45 Yes, I asked on 4/17 17:05:53 <GregNoel> ten days... 17:05:55 maybe it's too long to wait? 17:06:29 I'm sure there's some user error in there that maybe we could catch, but without the OP we can't do anything. 17:06:30 <GregNoel> kick him again, then close next time? 17:06:30 if we want to be super-nice, update it with a note re: we'll close it at the next bug party 17:06:39 agreed. 17:06:41 yes 17:06:44 <GregNoel> done 17:07:07 <GregNoel> 1610 17:07:14 <GregNoel> oops, 2610 17:07:41 <GregNoel> I don't know where this file is read, but would universal newlines help? 17:07:36 evil cygwin 17:07:40 evil, evil, evil cygwin 17:07:49 oh please. it's not that bad. 17:07:56 :-) 17:08:03 I use cygwin all the time. 17:08:06 Ditto. 17:08:12 as a user, it's great 17:08:13 for years and years and years. 17:08:14 It's dos that's evil in the first place. 17:08:24 many x-platform build systems with it too. 17:08:30 trying to develop for the cross-product of non-cygwin x cygwin? another thing altogether 17:08:36 anyway, I have no idea where to even start on this one. Where's that file list coming from? 17:08:51 right, i'm not sure myself 17:08:54 and where is scons parsing it? 17:09:23 we usually generate those, not pick them up from a file 17:09:29 I guess we look stupid if we ask the OP for this info. Who wrote the swig builder? 17:09:44 probably me, long ago enough to have forgotten the details 17:09:52 i think we probably slap my name on it by default 17:10:08 and there's this guy who just popped up on the ML with an itch to work on Java 17:10:11 with Russel Winder 17:10:14 It's probably a trivial fix once the right place is found... 17:10:31 sgk: this one might be a good one for him to start with. 17:10:36 sure 17:10:48 i'll take a quick look to see if it's obvious after a little digging 17:10:51 can you suggest it to him, with a hint or two to get him started? 17:11:00 (or what you said of course) 17:10:56 right 17:11:21 and contact him+Russel re: starting in earnest on refactoring Java support 17:11:51 <GregNoel> How should the issue be handled? 17:11:48 so... -research- sk (for now) 17:11:57 That sounds good. 17:12:00 <GregNoel> priority? 17:12:12 p4, only one person 17:12:30 sounds good 17:12:37 <GregNoel> ok, done 17:12:45 <GregNoel> 2572 17:12:49 invalid I think 17:12:53 agreed 17:12:55 <GregNoel> consensus 17:12:58 done 17:13:07 <GregNoel> 2576, consensus 17:13:20 2582, i closed 17:13:20 <GregNoel> 2587, garyo +1 17:13:26 sorry 2587 17:13:38 <GregNoel> 2588 17:13:39 yes, garyo +1 17:13:39 <Jason_at_Intel> already closed? 17:14:02 yes 17:14:11 sorry, not 2588 17:14:21 right, 2587 already closed 17:14:46 <GregNoel> 2588 17:14:52 2588: 2.x p3 +Easy ? 17:15:04 <GregNoel> Hmmm... 17:15:17 <GregNoel> Yeah, I can go with that. 17:15:17 sgk: ok, but I want a testcase first if you're ok w/ that 17:15:25 testcase++ 17:15:31 <GregNoel> agree w/testcase 17:15:49 so we wait for yann to give us a testcase, then it's 2.x p3 +Easy. 17:15:52 garyo: you mean, contact the OP for a testcase? 17:15:55 yeah 17:15:57 I already did 17:16:06 garyo++ again 17:16:12 <GregNoel> ++ 17:16:14 hmm, that was a long time ago now though 17:16:19 I'll ping him again. 17:16:31 <GregNoel> Put it on your plate for now? 17:16:36 ok 17:16:39 <GregNoel> research p2? 17:16:46 fine w/ me 17:16:49 <GregNoel> done 17:16:58 2589: consensus invalid 17:16:59 <GregNoel> 2589 17:17:04 <GregNoel> consensus 17:17:17 <GregNoel> 2590 close garyo++ 17:17:31 <GregNoel> 2599 17:18:09 Agree w/ Greg, we should do all this kind of thing in toolchain, but for now it's working as designed (though annoying) 17:18:25 Mark as invalid, assume workaround worked. 17:18:41 <GregNoel> Steven had some thoughts about doc? 17:18:55 yeah, if our doc example uses a string, that's misleading 17:19:09 it does, and it is. 17:19:21 ok, mark as doc w/ a note to fix example? 17:19:25 ++ 17:19:29 <GregNoel> who? 17:19:29 for our new tech writer? :-) 17:19:31 <Jason_at_Intel> not sure.. I always use list.. no issues 17:20:06 I raised the doc issue, i should probably own that 17:20:16 Jason: you can also use Append() etc., they always work. 17:20:28 just fyi. 17:20:32 <GregNoel> then when? and priority? 17:20:43 <Jason_at_Intel> I mean when i set a var .. i use a list not CLvar 17:20:51 list is fine too. 17:20:59 <Jason_at_Intel> I "don't get what value it has" 17:21:12 (heads up: 2-3 minutes until I get on the shuttle, I'll drop a minute or two until I reconnect) 17:21:21 sgk, 2.x, doc, p3? 17:21:25 done 17:21:29 <GregNoel> done 17:21:38 <GregNoel> 2600 17:21:51 any reasons MAXLINELENGTH isn't the workaround he wants? 17:22:13 I doubt his LINKCOM is using TEMPFILE, which is undocumented afaict 17:22:34 :-( 17:22:39 <Jason_at_Intel> I agree 17:22:59 back in a bit 17:23:00 * sgk has quit (Quit: sgk) 17:23:11 I think it's important to doc that. I'll take it for 2.something, p3. 17:23:31 <GregNoel> 2.2? 17:23:38 sounds good. 17:23:42 <GregNoel> ok, done 17:24:11 <Jason_at_Intel> +1 17:23:49 <GregNoel> 2601 17:23:59 <GregNoel> consensus 17:24:12 <GregNoel> but needs milestone, priority 17:24:11 sgk needs to be here to decide :-) 17:24:31 I think 2.2 p3 17:24:53 <GregNoel> Hmmm... 2.1, I think 17:24:53 * sgk (sgk@67.218.110.220) has joined #SCONS 17:25:09 +1 17:25:14 Hi Steven, 2601, documenting new cpp scanner: how about 2.1 p3 you? 17:25:24 sold 17:25:27 <GregNoel> done 17:25:38 <GregNoel> 2602 17:26:26 I think we can close it; I tried to hook him into contributing because I think he cares about this stuff, maybe he'll respond. 17:26:38 Jason_at_Intel: how reusable do you think your subprocess work in parts is? 17:27:16 <Jason_at_Intel> hmm.. I plan to tweak it a little more.. it is bound to a reporting API i have for coloring and logging as well 17:27:34 <Jason_at_Intel> but unhooking that would not be hard 17:27:12 <GregNoel> I agree with Steven about identifying the big projects and at least listing them... 17:27:41 GregNoel: let's see how much time we have after the bugs 17:27:55 <GregNoel> sgk, agree 17:27:51 we could start by just brainstorming all the big things we know we'd like to do 17:27:59 and maybe prioritize / roadmap them next time? 17:28:16 ++ 17:28:31 <GregNoel> worksforme 17:27:19 I'm okay with closing 2602 in the meantime 17:27:37 <GregNoel> invalid it is 17:28:37 okay, then close 2602, and add SPAWN refactoring to a roadmap discussion 17:28:36 2604 seems like consensus 17:28:59 2604: rob is the man 17:29:04 <GregNoel> 2604, consensus 17:29:12 2606: 2.x p3 sk 17:29:32 <GregNoel> 2606, if Steven is volunteering... 17:29:45 yeah 17:29:48 <GregNoel> done 17:30:14 <GregNoel> 2607 17:30:23 <GregNoel> consensus on toolchain 17:30:28 yes 17:30:31 yes 17:30:36 <Jason_at_Intel> yes 17:30:50 <GregNoel> The rest of the toolchain issues are 3.x p3 17:30:58 <GregNoel> (except one) 17:31:03 and... that looks like it 17:31:06 quick work tonight 17:31:16 2608? 17:31:23 <GregNoel> er, 2608? 17:31:45 ? is that in the spreadsheet? 17:31:53 yes. progress #'s 17:32:01 <GregNoel> You even commented on it. 17:32:14 sgk: if you want to take a crack at it I see how it could be useful. I'd support that effort. 17:32:23 we sure get asked for it a lot. 17:32:27 okay, nm, i seem to have a shortened spreadhseet here 17:32:32 * sgk refreshes... 17:32:40 damn google... :-) :-) 17:33:11 no kidding... 17:33:16 stupid chrome 17:33:22 ah, there we go 17:33:35 hey, did you guys know there are more issue farther down the spreadsheet...? ::-) 17:33:59 <GregNoel> No, that's the last one. 17:34:07 222 is the last line in mine 17:34:17 ditto 17:34:17 <GregNoel> 220? 17:34:25 row # 17:34:30 <Jason_at_Intel> 2608 is the last? 17:34:33 yes. 17:34:34 looks like 17:34:36 <GregNoel> oops, yep, 222 17:34:46 * sgk slinks off and stops making bad jokes 17:34:46 <Jason_at_Intel> oh 222 row 17:33:46 <GregNoel> I like Gary's idea of a wiki page to figure out what we can do. I'd contribute to that... 17:35:18 GregNoel: do we have a keyword for TNG? 17:35:35 <GregNoel> Uh, I think so; if not, it's easy to add. 17:35:35 <Jason_at_Intel> TNG? 17:35:43 anything we do to the current infrastructure to support this is throwaway 17:35:45 sgk: why not start by putting it on the wiki, and if people like it we add it with a descriptive name that shows it's approximate. 17:35:59 tng=taskmaster next gen 17:36:46 <Jason_at_Intel> the next generation star trek goes across my mind everytime i see that 17:36:07 <GregNoel> ApproximatePercentage 17:36:16 yeah, something like that 17:36:27 but it needs to be a callback, let's not design it here. 17:36:38 <GregNoel> ApproximatePercentageThatMayGoBackward 17:36:44 :-) 17:36:36 I'm not following you... put it on the wiki? you mean a discussion about whether people want this feature? 17:37:00 sgk: no, put the code itself on the wiki for people to try. 17:37:13 <GregNoel> No, a discussion on how we can implement it, and how approximate the options would be. 17:37:17 well, that was my original suggestion anyway. 17:37:27 ulp. what i had in mind would probably be pretty invasive 17:37:46 invasive as in changes, or invasive as in using undocumented apis? 17:37:54 <Jason_at_Intel> any more so than the buildNow tool? 17:37:54 i wasn't thinking about the walk-the-tree-once-to-count idea 17:38:01 <GregNoel> No surprise, but invasive or not, it's not obvious what the tradeoffs are. 17:38:03 buildNow tool? 17:38:31 invasive as in I was thinking avoid the duplicate tree walk by counting Nodes as they're added 17:38:32 <Jason_at_Intel> I might have teh wrong name... but someone made a tool to build a target 17:38:34 <Jason_at_Intel> RIghtNow 17:38:41 <Jason_at_Intel> that was it i think 17:38:56 never heard of it 17:39:05 Jason_at_Intel: send me a pointer / link? I haven't heard of that 17:39:06 <Jason_at_Intel> so it calls the taskmaster and stuff to build a target right then 17:39:20 Hm, there it is in the wiki. Will have to check it out. 17:39:22 <GregNoel> sgk, yeah, but a LOT of nodes do nothing: rfile duplicates, ... 17:39:31 wow, sounds only slightly less gnarly than the SConf stuff... :-/ 17:39:52 GregNoel: I'm doing a lot of hand-waving, yeah 17:39:58 <Jason_at_Intel> http://www.scons.org/wiki/RightNow 17:40:28 it just wouldn't be an easily-patchable, self-contained bit of code behind an if-test, say 17:40:56 <Jason_at_Intel> thought it would be useful to do something like this in Parts as well to speed up build times for large incremental builds 17:41:14 <GregNoel> What's wrong with "Execute()"? 17:41:31 Execute() runs an action, no dependency checking 17:41:55 right, and doesn't set the target as uptodate 17:42:15 <Jason_at_Intel> ideally i can read other Parts files while i start build leaf components 17:42:18 <GregNoel> Um, what runs configure checks, then? 17:42:36 heh. that's pretty interesting 17:42:54 bet he's not using Configure() at all 17:43:12 Jason_at_intel: are you using RightNow() in Parts ? 17:43:19 <Jason_at_Intel> not yet 17:43:33 <Jason_at_Intel> I was thinking about it for the next drop 17:44:06 <Jason_at_Intel> not directly... but build it in to Parts ... 17:44:18 RightNow code isn't very big. A page or less. 17:44:29 it's pretty fresh, his initial (only) checkin was 17 March 17:44:52 <Jason_at_Intel> If i allow user to call right now .... the read phase would take forever 17:45:16 anyway, sgk, this ApproximatePercentageThatMightGoBackwards sounds like an interesting bg task if you get to it, but maybe we can design TNG to make it easier? 17:45:24 <Jason_at_Intel> However it uses the internal code.. a don't know if this would bad for TNG 17:45:42 <GregNoel> In any event, we're drifting from the topic of issue 2608, the progress indicator. 17:45:44 <Jason_at_Intel> but suggests a dev level API woudl be a nice addition with TNG for doing stuff like this 17:45:44 garyo: yes, that's why i was asking about TNG 17:46:10 yup, just agreeing w/ you and trying to return to the topic at hand. 17:46:22 yeah 17:46:44 Hey so Did u guys see my email about the tech writer? 17:46:51 <Jason_at_Intel> so progress bar is an impl for people to try to invasive? 17:47:13 I did -- look up a few hundred lines & I mentioned them :-) 17:47:36 yup. saw that. 17:47:45 bdbaddog: tech writer++ 17:47:56 what would be a good next step to explore the fit w/her? 17:48:01 Jason: too ugly for a wiki implementation, sgk may try it in the bg but no promises (did I get that right?) 17:48:01 any low haning fruit for her to take a wack at? and/or howto's she should go through? 17:48:18 <GregNoel> Steven, could you update the issue with what you think might be possible, now and TNG? 17:48:25 garyo: i agree 17:48:36 <Jason_at_Intel> +1 greg 17:48:37 GregNoel: yes, give it to me for updating 17:48:39 Is she up for just fixing a few of the easy doc bugs (not the ones that require detailed impl knowledge)? 17:48:46 yes. 17:49:01 That seems like a great start. 17:49:02 <GregNoel> done, now we can go off-topic. And think about starting a wiki page. 17:49:18 bdbaddog: let's you and I sync up off-line re: doc tasks 17:49:22 wiki page for roadmap/projects? 17:49:46 sgk: sounds good. 17:50:03 <GregNoel> I was still on wiki page for progress percent, but a wiki page for big projects would be a good thing, too. 17:50:56 re: progress percent, sounds like that's on my plate, yes? 17:50:56 Maybe sk's comments on 2608 form the basis of the wiki page, if we're lucky 17:51:09 that's what i was thinking 17:51:13 +1 17:51:31 <GregNoel> And while everybody is throwing in topics, how about a 2.0 checkpoint? I think it's ready. 17:51:59 cool 17:51:59 I can help w/ it this weekend, not before. 17:52:11 <GregNoel> sgk, yes progress percent is on your plate; I'll have to figure out how. 17:52:15 (Well, I can start Fri night) 17:52:43 fyi, i'll be out of town and probably mostly off-line this Thursday through next Tuesday 17:53:19 No prob for the ckpoint; if it's terrible, we'll just take it down. :-) 17:53:38 (Not that it would be of course.) 17:53:44 <GregNoel> Well, I've got one more thing I can work on until Fri, so I guess we'll start then. 17:53:46 of course! 17:54:10 are we ready for 2.0 checkpoint build? 17:54:23 right, that's what GregNoel and garyo are discussing 17:54:25 That's what Greg's saying, yes. 17:54:51 Anyone actually tried it in real life yet? 17:54:55 ugh long day.. 17:55:02 :-) 17:55:24 not that i know of 17:55:26 I can do 1.3.1 checkpoint and 2.0 this week if you like. 17:55:32 I'll try it on my Windows 7 box. 17:55:53 bdbaddog: both? I'll give you a hand of course! 17:56:16 yeah no problemo. 17:56:23 awesome, you're on. 17:56:40 I'll at least help w/ release announcement text etc. 17:56:46 i was starting on 1.3.1 ckpoint on sunday, ran out of steam. 17:57:07 Sure. That'd be great. We can coordinate via mail. 17:57:15 perfect. 17:57:45 So, project list? 17:57:56 <GregNoel> TNG 17:58:03 subst 17:58:04 <GregNoel> Toolchain 17:58:13 GSoC windows installer 17:58:34 <GregNoel> (is that a big project or just a merge?) 17:58:35 Node refactor 17:58:44 Greg: hopefully just merge 17:59:05 is the installer Wix or NSIS ? 17:59:20 nsis if I remember correctly 17:59:38 INSTALLER: k. I have some experience with NSIS 17:59:00 <GregNoel> sgk, refactor what about Nodes? 17:59:08 <Jason_at_Intel> Nodes 17:59:34 <Jason_at_Intel> and API.. should be easier to use 17:59:09 Node hierarchy 17:59:20 use composition instead of inheritance 17:59:24 <GregNoel> Hmmm... sgk, interacts with TNG. 17:59:45 <GregNoel> probably minor, though... 17:59:49 very possibly 17:59:47 How about switching tests to py.test ? 17:59:58 componentization model / Parts integration 18:00:08 tests as dirs 18:00:27 <GregNoel> expose test strings 18:00:41 greg: what's that mean? 18:00:46 yeah, tests as dirs + expose test strings + unittest 18:00:57 it's a side effect of tests as dirs 18:01:00 <GregNoel> Test strings go into directories. 18:01:00 ok 18:01:09 all the Python code that's in in-line strings get put into files 18:01:11 as files rather than strings in the test files. 18:01:21 so the Python 3.x fixers can operate on them 18:01:16 yah, got it. 18:02:08 take many tools out of scons core, make them plug-ins 18:01:53 <GregNoel> That's a pretty good list. Someone want to start a wiki page? 18:02:22 Greg: I'll start the page based on this list. 18:02:31 <GregNoel> garyo++ 18:02:57 right, most likely coordinating w/Russel Winder re: his ideas for separate Tool development 18:03:17 oh, use a DVCS to front the SVN repository for devlopment? 18:03:26 sgk: definitely. And adding system site_scons dirs, all that stuff. And interacts w/ toolchain too. 18:03:27 <GregNoel> I'm not seeing any more brainstorming; shall we deem the list complete for now? 18:03:42 complete enough 18:03:45 Yes! 18:03:50 sgk: anyone can front svn with a dvcs today. 18:03:57 send out a link, we can think and add more for two weeks 18:04:06 and then put them in some rough priority order 18:04:14 ok, will do. SConsFutureProjects or something. 18:04:34 <GregNoel> garyo, call it BigProjects; we can start a separate page for Roadmap. 18:04:44 true enough re: front-end development 18:05:02 BigProjects it is. 18:05:22 * sgk has another 3 minutes or so 18:05:26 <Jason_at_Intel> I can use Bazaar with SCOns and Parts... but i can't use it at work ( crashes) ( same with GIT) 18:05:28 <GregNoel> SConsBigProjects would work, too. 18:05:29 anything else to cover? 18:05:43 No, I don't like BigProjects, that's confusable with "how to do a big project with SCons." Anyway I'll think about it. 18:05:52 I've found that I like Mercurial, myself 18:06:18 I front a svn repo with git and hg daily. hg is easy, git takes a little more work but no biggie. 18:06:34 <GregNoel> techtonik has mentioned on the mailing list that he wants to use Hg with SCons SVN. 18:06:44 * sgk needs to get more modern 18:07:04 I prefer git because I'm hardcore, but hg is nice & pillowy :-) :-) 18:07:15 I'm fine with either hg or git. 18:07:36 <Jason_at_Intel> does hg work with non standard SVN layouts? 18:07:49 not easily, they both suck at that 18:08:03 okay, i'm gone -- thanks guys 18:08:09 g'night Steven! 18:08:14 * sgk (sgk@67.218.110.220) has left #SCONS 18:08:16 <GregNoel> Me, too; dinner is called.... 18:08:24 time for me to go too, homework time for kids 18:08:26 * GregNoel has been marked as being away 18:08:34 <Jason_at_Intel> ok later all 18:08:37 l8r 18:09:01 * garyo has quit (Quit: Leaving.) 18:09:14 * Jason_at_Intel has quit (Quit: ChatZilla 0.9.86 [Firefox 3.5.3/20090824101458])