IrcLog2010 09 21

17:03:08 <jason_at_intel> so greg that sample for the tar file builder 17:03:46 <jason_at_intel> are you suggesting that this should be special cased per builder? 17:04:05 <jason_at_intel> given that you have a special node wrapper in it 17:04:31 <jason_at_intel> that has special knowledge of the SCon internals 17:05:35 <GregNoel> I need to write up my notes on that; it requires some modifications to the Node class to create a special proxy, as well as some minor tweaks to the Taskmaster to deal with the proxy correctly. 17:06:03 <jason_at_intel> ahh ok... so we should fix up some stuff in SCons first then 17:05:56 <GregNoel> are we all here and ready to go? 17:05:59 Shall we start in? Are we waiting for anyone else? 17:06:03 :-) 17:06:28 i'd say let's go... 17:06:32 <GregNoel> 2674 consensus invalid 17:06:32 <GregNoel> 2677 consensus research p1 Steven 17:06:32 <GregNoel> 2678 consensus invalid 17:06:32 <GregNoel> 2679 consensus 3.x p3 (no owner required) 17:06:32 <GregNoel> 2680 needs priority, but otherwise consensus 17:07:02 <jason_at_intel> wow that was fast :-) 17:07:35 2680: p3? just by default? 17:07:52 <GregNoel> p3 or p4, I think. I don't really care which. 17:08:26 p3 seems fine to me (revisit in 3 weeks) 17:08:39 <GregNoel> six weeks 17:08:47 <GregNoel> (three meetings) 17:08:55 sorry, right you are 17:08:46 p3, get feedback to the OP a little sooner than p4 17:08:57 6wks +1 17:08:59 <GregNoel> p3 going once 17:09:05 <jason_at_intel> +1 17:09:11 <GregNoel> done 17:09:14 <GregNoel> 2681 We really should have installable packages for all three major platforms 17:09:37 yeah 17:10:03 If it comes to it, some platforms can be delayed slightly -- the new all-in-one Windows installer will only be buildable on Windows for instance (at least at first) 17:10:10 (not in 2.1) 17:10:14 <jason_at_intel> I would agree.. but i don't think package has long term viability outside open source only usage 17:10:32 Jason: don't understand? 17:10:59 <jason_at_intel> Package() does not have what is needed to make a real enterprise installer 17:11:17 Ah, I see. You mean to package SCons itself. 17:11:43 <jason_at_intel> ya.. 17:11:52 <jason_at_intel> well if we are to eat our own dog food that is 17:11:41 I thought this guy was proposing a separate script. 17:12:35 we're a little ways from being able to dogfood something like this, agreed 17:12:37 I think Package() has all the low-level bits, but of course will need work on high-level stuff. But that's irrelevant here. 17:12:42 but it's probably a step towards making things better 17:13:08 <jason_at_intel> but it would be nice to have a native option to install SCons other than "setup.py install" 17:13:42 Of course. We already have that for Windows, and will have a better option soon. And this guy's script gives us the "right thing" for Mac. All that's left is rpm/apt. 17:14:04 <jason_at_intel> it is a good start 17:14:24 so ... 2.2? 17:14:38 <GregNoel> sure; what priority? 17:14:40 All I'm saying is, rather than trying to do all the release builds on Mac, I'd be OK with doing the releases as we do now, but then notifying someone (like Richard) that it's time to do a Mac pkg. 17:14:42 is new win installer 2.2 also? 17:15:03 Hi Bill -- I think he's been keeping it up with trunk, so 2.2 seems good to me. 17:15:10 p3? or p2 because it's an actual patch and it's rude to make the OP wait? 17:15:58 I guess we could integrate his scripts into the bin/ dir, and ask him to start using them whenever he's ready. Even for 2.1. 17:16:49 (Then he or someone would upload his packages to SF of course) 17:17:12 I have mac will travel.. so can build mac pkgs' 17:17:18 Ditto. 17:17:14 <GregNoel> Should we add some issues to have us start using Package() to create our own packages? 17:17:42 GregNoel: good idea 17:17:42 Greg: that would help whip it into shape for sure! 3.x p3? 17:18:19 Anyway, I say for this issue let's go with 2.1 p2 just to check in his scripts. 17:18:30 <GregNoel> worksforme 17:18:41 But not commit to producing a pkg for 2.1 unless it is as easy as we hope 17:18:51 +1 17:18:56 <GregNoel> agree 17:19:07 <GregNoel> Assign to Steven? 17:19:38 sure 17:19:42 <GregNoel> done 17:19:45 <GregNoel> 2682 consensus 2.1 p2 Dirk (unless there's a violent objection) 17:19:45 <GregNoel> 2683 consensus 2.x p3 Bill 17:19:45 <GregNoel> 2684 17:20:22 2683: Bill, I suggest using a VM to set up mingw. You can't do it on a machine with cygwin. Just FYI. 17:21:17 garyo: Thanks! 17:20:40 2684: research p3 sk ? 17:20:57 <GregNoel> Sounds fair. 17:20:58 agreed. 17:21:06 <jason_at_intel> +1 17:21:17 <GregNoel> done 17:21:20 <GregNoel> 2686 consensus discuss in dev@scons; who shall lead discussion? 17:21:52 I can start it. 17:22:18 <GregNoel> Garyo, +1, thanks 17:22:36 <GregNoel> 2687 17:23:26 Jason, can you track it down a little further? 17:23:33 See where it's going wrong? 17:23:48 <jason_at_intel> Sure... It will be a little bit as i have a lot on my plate 17:24:05 It's your itch to scratch :-) :-) 17:24:04 <GregNoel> I'm pretty sure the prefix matching is a part of optparse, and it requires the shortest prefix first. 17:24:30 <jason_at_intel> I thought you where going to redo this stuff 17:24:46 <jason_at_intel> ie from some dev board postings 17:25:15 I want to, but it's a big job and not top of my list. Anyway, if it's a low-level parser issue that won't help. 17:25:21 <jason_at_intel> but ya... I can scratch it.. did not want to step on your toes 17:25:26 np 17:25:43 <GregNoel> milestone and priority? 17:25:50 <jason_at_intel> so i guess assign to me then p4 17:25:53 I put that behind toolchain and lots of other things (for me at least) 17:26:14 agree jason p4. Research? 17:26:20 <GregNoel> yes 17:26:31 <GregNoel> done? 17:26:34 done 17:26:38 <GregNoel> 2688 Lots of comments, no suggestions.... 17:27:21 Hmm. How about 2.x p3? 17:27:45 (I think we need to keep the .bat too, for older Windows that couldn't run .py directly) 17:28:15 garyo: which win would that be? 17:28:35 I think XP couldn't. 17:28:43 And I think 2000 also. 17:28:43 I 17:29:15 <GregNoel> sgk, II, too; anybody for III? {;-} 17:29:42 99% sure xp can.. 2k though perhaps not. 17:29:23 re: the arguments, I've also turned up reference to a registry setting that controls what arguments get passed to an invoked .py 17:29:48 but I don't know if it's across all Windows versions or not, so keeping the .bat file is safest 17:29:55 * jason_at_intel has quit (Ping timeout: 240 seconds) 17:30:09 * jason_at_intel (~[email protected]) has joined #SCONS 17:30:25 <jason_at_intel> ok back 17:30:59 <jason_at_intel> which one are we on? 17:31:09 still 2688 17:31:20 <jason_at_intel> thanks 17:30:13 I might be mistaken but I remember trying & failing a while ago, doing a lot of reading up on Windows Scripting Host. :-( 17:30:30 Anyway, setting %PATH% is a good idea no matter what. 17:30:59 agreed 17:31:45 Steven, I'm afraid you know the setup.py logic best; you'll have to start this one. I can clean up and test though. 17:32:16 fair enough 17:31:47 <jason_at_intel> adding \script 17:32:46 <jason_at_intel> so all 2.4 versions seemed good with .py on windows 17:33:09 jason_at_intel: with all arguments being passed to the script? 17:33:18 <jason_at_intel> yes 17:33:21 ... and what Windows version? 17:33:39 <jason_at_intel> NT line 17:33:45 wow, ok 17:33:28 <jason_at_intel> we have people here that tweak this 17:33:50 so they set up the registry to pass the arguments, then 17:33:57 that's not the default behavior 17:34:04 <jason_at_intel> registry? 17:34:21 <jason_at_intel> no they rename scons to scons.py 17:34:34 <jason_at_intel> so the PATHEXT will call it with out python 17:35:09 @jason_at_intel: http://www.lalitkapoor.com/blog/2008/12/21/python-wont-take-command-line-arguments-windows-vista/ 17:36:10 <jason_at_intel> never seen this myself.. we use python scripts all the time as programs at work 17:34:24 Jason, maybe you & Steven can take that issue offline and show him what's needed? 17:35:14 <jason_at_intel> fair enought .. as long as Steve answers the e-mails 17:35:19 :-) 17:36:09 so we'll take it off line 17:36:23 <jason_at_intel> ok... will you start the e-mail? 17:36:27 vanilla windows installs, or configured by your admins? 17:36:39 (wait, I just said we'd take this off line...) 17:36:42 onward 17:36:13 i should own the issue? 17:36:59 <GregNoel> Er, it needs a milestone and priority, if not an owner... 17:37:16 well, we have the other install-related issues 2.2, so how about 17:37:18 2.2 p3 sk? 17:37:44 <GregNoel> worksforme 17:37:40 Install cleanup: theme for 2.2. :-) 17:37:50 I'll notify Lukas. 17:37:59 cool 17:38:03 <GregNoel> done? 17:38:18 done 17:38:22 <GregNoel>
1170 bypass, no substantive comments 17:38:22 <GregNoel>
1176 bypass, no substantive comments 17:38:22 <GregNoel>
1182 bypass, no substantive comments 17:38:22 <GregNoel>
1406 Should we contact Russel to see if someone in the third-part crew can take it? 17:38:49 GregNoel: good idea 17:38:54
1406: that would be my recommendation too 17:39:23 <GregNoel> done 17:39:26 <GregNoel>
1418 bypass, no substantive comments 17:39:26 <GregNoel>
1708 17:39:46 <jason_at_intel> Looks like Steve needs some time 17:39:47 research p2 sk 17:40:03 time is definitely the precious commodity... 17:40:14 :-/ 17:40:24 <GregNoel> You don't know the half of it. 17:40:21 <jason_at_intel> ahem 17:40:53 <GregNoel> OK, I'll adjust the priority 17:40:46 Well, we're all here & that's impressive. 17:41:01 <GregNoel> Garyo, concur 17:41:09 <GregNoel> done? 17:41:18 yes 17:41:21 <GregNoel> 2249 bypass, no substantive comments 17:41:21 <GregNoel> 2521 bypass, no substantive comments 17:41:21 <GregNoel> 2575 17:41:38 <jason_at_intel> I think this needs to be taken offline 17:41:40 But wait, 2249 & 2521, any news from Bill? 17:41:53 <GregNoel> Ooohhh, 2575, that comment has changed since I looked at it earlier... 17:42:20 <jason_at_intel> it seems tied with node objects and task processing 17:42:25 (maybe Bill's gone) 17:42:30 ok, 2575 17:42:49 no updates from me been pretty swamped lately. 17:42:59 This is what you guys were talking about at the beginning, i think. 17:43:20 <jason_at_intel> ya... I just read his sample in the tracker 17:43:25 <jason_at_intel> hmm 17:43:36 <jason_at_intel> not tracker.. but bug report 17:44:04 ok, sounds like you guys are working on it. Just keep us up to date? 17:44:22 <jason_at_intel> I think Steve need to be in this talk 17:44:36 <jason_at_intel> as it seems to be related to stuff that been talked about needing some upgrades 17:45:34 <jason_at_intel> ie Node objects and TNG 17:44:30 <GregNoel> yes, I concur, and Gary, too 17:45:28 <GregNoel> I'll try to write up my notes, but my time is seriously compressed these days. I may have a bit of time to work on it in a week or two; I'll try very hard to do it then. 17:45:51 which issue(s)? 17:46:10 <jason_at_intel> Keep this as research lower priority? 17:46:28 <GregNoel> yeah, p3 or p4 17:46:32 2575? 17:46:34 2575 17:48:02 <jason_at_intel> so the idea Greg proposed and has a sample of if to allow a tuples to be used 17:48:37 <jason_at_intel> this would be useful for copy builder, or any builder that the user might want to control tree structure a little 17:50:26 I don't have a conceptual problem with that; devil's in the details, of course 17:50:54 <jason_at_intel> what is missing in the sample is the ability for the targets returned from the builder to have this tuples in it, in case structure needs to be passed form builder to builder 17:51:31 <GregNoel> Hmmm... I don't see that... 17:51:32 hmm, for passing from builder to builder arbitrary tuples start to feel messy 17:51:41 <jason_at_intel> yep the details... 17:51:38 what about attributes on an object 17:51:50 (not that there isn't already enough in a Node as it is) 17:52:03 <jason_at_intel> Well i was playing with the idea of tagging the nodes 17:52:11 <jason_at_intel> but i have to prototype it yet myself 17:52:13 <GregNoel> sgk, problem is if same node is passed with multiple names (or to multiple destinations). 17:52:04 so that's what you guys were talking about earlier re: proxies? 17:53:29 <GregNoel> sgk, yes, tuple needs to be turned into a proxy, which looks like a Node for everything except scheduling. 17:53:48 that sounds workable 17:54:47 <GregNoel> (Garyo, that's why arg2nodes() works; proxy acts on underlying node.) 17:54:04 K. I'm gonna run unless there's other release items. are we looking at 2.1 checkpoint anytime soon? 17:54:44 I was just going to ask. I want to get full Intel 11 support in, then it's just a bunch of piddly stuff from me. 17:55:03 <jason_at_intel> what about 12 :-) 17:55:32 Jason: get me a copy or at least an ls-lR of Linux and a reg dump on Windows. 17:55:38 <GregNoel> wait, did we decide on p3 v. p4? 17:55:44 cutting edge! 17:55:57 <jason_at_intel> the 12 XE is rc1 at the moment 17:56:14 <jason_at_intel> but i will be adding it to Parts soon myself 17:56:38 <jason_at_intel> windows should be working today ( ie composer 2011) 17:56:35 I'll follow up w/ you and see if we can do something. 17:56:45 <jason_at_intel> sure 17:57:04
11 doesn't work on Windows today 17:57:21 <jason_at_intel> in SCon native... in with Parts it does 17:57:12 <GregNoel> wait, did we decide on p3 v. p4? 17:57:27 for 2575? I like p3, but it's up to you folks 17:57:43 <GregNoel> Hearing no other objections, p3 it is. 17:58:05 <jason_at_intel> ok 17:58:22 p3 17:59:14 <GregNoel> That's all the issues for this meeting; good work. I've got to go; I'll leave my session running to catch the rest of the comments. 17:59:06 Anyway, as far as I personally am concerned, as soon as I can get intelc 11 support in (and Linux is done now), I'm ready for a checkpoint. 17:59:25 i'll take a look at what's on my plate and whittle it to the most important 17:59:47 offhand, i don't think anything I have is worth holding up a checkpoint for, though 18:00:02 so intelc 11 is a good milestone from my standpoint 18:00:17 (or gating item, actually) 18:00:25 OK, I'll try very hard to get it in soon (this week maybe) 18:01:17 <jason_at_intel> so I will wait for you e-mail Steve 18:01:42 <jason_at_intel> I also have a question about actions and task... I will e-mail offline tomorrow 18:02:05 <jason_at_intel> hopefully you can take a moment to answer :-) 18:02:29 okay, i'll try to get at it later tonight (working late) 18:01:46 ok guys, thanks a lot -- talk again soon. 18:02:32 gnight all then. :) 18:02:36 'night 18:02:40 <jason_at_intel> night!