**** BEGIN LOGGING AT Mon Aug 15 02:59:57 2011 Aug 15 08:25:02 RP__: ping Aug 15 08:25:10 yuke: pong Aug 15 08:25:48 RP__: for commit oe-core commit 0086e2e7f776b71378eb9cb69c0e59db4a5ffea7 Aug 15 08:26:00 commit e8755c782fb1640c3a24f512f7c7bfe210d466c1 Aug 15 08:26:00 Author: Richard Purdie Aug 15 08:26:00 Date: Tue Jul 5 22:33:08 2011 +0100 Aug 15 08:26:00 qt4: Package debug source files Aug 15 08:26:00 (From OE-Core rev: 0086e2e7f776b71378eb9cb69c0e59db4a5ffea7) Aug 15 08:26:57 the , "${QT_BASE_NAME}-dbg" actually does not exist Aug 15 08:27:23 so i wonder what the purpose of this code, and how i can revise it Aug 15 08:27:40 yuke: It did exist when I tried that :/ Aug 15 08:27:41 i happen to find this issue when fixing the qt4 multilib issue Aug 15 08:28:32 RP__: so it is removed later? or i miss something? Aug 15 08:29:02 RP__: do you know which code generate this package? Aug 15 08:29:18 yuke: which qt4 are you building to test? Aug 15 08:29:36 RP__: qt4-x11-free Aug 15 08:29:55 yuke: ah, I think I tested with qt4-embedded where PN = QT_BASE_NAME Aug 15 08:30:24 RP__: that make sense. PN-dbg does exist Aug 15 08:30:58 RP__: so it should be ${PN}-dbg? Aug 15 08:31:14 yuke: yes Aug 15 08:31:30 RP__: ok, thanks, i will revise it Aug 15 08:31:42 yuke: thanks Aug 15 08:33:22 RP__: BTW, for the qt4 multilib issue, i found make package unexpand has other issue, say, PACKAGE= "... ${OTHER_PACKAGE}" OTHER_PACKAGE="xxx yyy", then PACKAGE will map to lib64-${OTHERPACKAGE} which is not correct Aug 15 08:34:22 RP__: so i use another approch, i.e. revise the qt4.inc, and expand the package first Aug 15 08:34:36 yuke: ok, sounds reasonable Aug 15 08:35:00 yuke: I did wonder if that "don't expand" tweak would work Aug 15 08:37:03 RP__: i can not go further to test the don't expand tweak, because there is dependency failure caused by the issue i mentioned. but in theory, i think it should fix the qt4 multilib bug Aug 15 08:37:26 RP__: but will introduce other issue Aug 15 08:38:00 yuke: Right, we can't introduce other problems Aug 15 08:40:54 morning all Aug 15 09:06:14 hi bluelightning Aug 15 15:47:39 Good Morning All ! Aug 15 17:03:57 RP__: if I set FILES_${PN}-dev += "something" and it ends up setting the value to None according to bitbake -e, would you call that a bug? _append works... Aug 15 17:07:00 scratch that, I'm being an idiot Aug 15 17:49:57 pidge: ping Aug 15 17:52:28 sgw: pong Aug 15 17:52:53 pidge: it looks like the autobuilder is stuck and not updating git on checkout Aug 15 17:53:21 sgw: looking Aug 15 17:54:11 sgw: which target is this? Aug 15 17:54:43 both fuzzys Aug 15 17:57:18 sgw: I see fuzzy mut building 2e7aa7886ec2be13659e31e35f97c95966f0c358 Aug 15 17:58:46 Ok, I am confused then, that;s right, but wrong. Aug 15 17:58:56 I might have missed a push! Aug 15 17:59:10 Master is 4 days out of date though Aug 15 18:00:14 ok, looking now Aug 15 18:00:37 * pidge is going to crush this problem once and for all. one moment while I smack the ab. Aug 15 18:59:04 ugh, the sort of bbfiles by priority is done before glob expansion / traversal, which means there are any number of configurations which can result in bbappend files not being applied in layer priority order Aug 15 18:59:08 lovely Aug 15 19:50:21 RP__: I'm booting core-image-sato on 32MB poodle ;) Aug 15 19:50:43 is slow as hell... Aug 15 21:49:36 ant__: any idea of the cause? Aug 15 21:49:53 seems touchscreen Aug 15 21:50:02 I can log from serial console Aug 15 21:50:22 happens also on 64MB c7x0 Aug 15 21:51:47 RP__: situation is a bit troubled because kdrive in oe-core is almost deprecated and other imnages would need xorg-xserver from meta-oe Aug 15 22:13:17 has anyone seen git ls-remote failing quite often? Aug 15 22:15:36 RP__: http://paste.debian.net/126345/ Aug 15 22:39:13 denix: which repo? Aug 15 22:40:33 ant__: I was meaning why is it slow and do you mean in comparison to some other image or just commenting on poodle in general Aug 15 22:40:36 ? Aug 15 22:41:49 no, i'ts not slow. Just the gui somehow freezes after touching the touchscreen Aug 15 22:42:16 surely extra 32MB give a small boost to corgi Aug 15 22:47:16 RP__: first noticed on our own git server, then also on kernel.org Aug 15 22:47:35 RP__: I don't think it's repo-specific Aug 15 22:48:37 denix: could it be a network issue? Aug 15 22:48:41 was rt-tests switched to actual commit ID instead of tag due to git ls-remote needed for every parsing? Aug 15 22:48:51 I've noticed servers with issues before but I wouldn't expect it from kernel.org Aug 15 22:49:09 denix: no, commit ids would remove the need for ls-remote Aug 15 22:49:32 RP__: that's what I'm trying to debug now, was wondering if anyone else experienced it Aug 15 22:50:01 that was my question - was the change to avoid ls-remote? Aug 15 22:52:09 we use tags extensively instead of commit IDs, plus several AUTOREV-ed recipes. all of them use ls-remote and it often fails with "The remote end hung up unexpectedly" Aug 15 22:54:41 denix: master shouldn't hit the network Aug 15 22:56:05 denix: I've seen that quite a bit too, intermittant, not sure what's going on with it Aug 15 22:57:49 RP__: by "us" I meant our recipes on top of oe-core/yocto :) Aug 15 22:58:58 kergoth_: hmm, interesting. it is intermittent. it's annoying when it works during the day when I try to isolate and debug the issue, but then fails at night on autobuilder... Aug 15 22:59:05 denix/kergoth_: Server side I've seen git-daemon run out of connections on ports Aug 15 23:00:14 RP__: I guess I can convert all tags into commit IDs, but would be harder to fix AUTOREV-ed recipes Aug 15 23:00:51 denix: IIRC one thing to improve the siutation is to use inetd instead of sv to run git-daemon Aug 15 23:02:05 RP__: that's what I use (xinetd), but still had to put a limit to conserve resources Aug 15 23:02:40 denix: don't know then :/ Aug 15 23:03:02 I'll try to raise the limit for now, see if it improves the situation... Aug 15 23:03:03 thanks Aug 15 23:03:43 but it won't help with other git servers, like kernel.org... Aug 15 23:04:33 denix: I'm surprised/concerned you're seeing it there Aug 15 23:04:40 RP__: one more question - with recent logger overhaul, should I see all the process outputs on the console? Aug 15 23:04:49 I wonder if bitbake is somehow making too many repeated requests Aug 15 23:04:59 * Tartarus waves Aug 15 23:05:09 Tartarus: hey Aug 15 23:05:09 denix: That is one area we need to get better at, we don't output enough logs Aug 15 23:05:12 my scrollback implies someone said my name, but it's a lot of scrollback Aug 15 23:05:14 hi denix Aug 15 23:05:16 * RP__ -> Zzzz Aug 15 23:05:23 * RP__ has an early flight Aug 15 23:05:27 later RP__ Aug 15 23:05:34 RP__: bye Aug 15 23:07:08 well, I have early flight myself... Aug 15 23:07:29 denix: Do I need to look for you in Dallas tomorrow? :) Aug 15 23:07:31 git:// has limitations, I avoid it wherever possible nowadays, even given the hassles of 'dumb' transports :\ Aug 15 23:08:05 Tartarus: yep :) Aug 15 23:08:16 denix: Noted :) Aug 15 23:08:50 Tartarus: will be there tuesday through thursday Aug 15 23:09:40 kergoth_: unfortunately http:// for git is not better, plus ls-remote uses what SRC_URI has... Aug 15 23:10:01 denix: not better in what way? this context? Aug 15 23:10:15 I'm sure apache can manage to handle a ton of collections a lot better Aug 15 23:10:16 heh Aug 15 23:10:51 kergoth: how efficient is it with packed objects etc? Aug 15 23:14:10 I thought that was a big part of what the git update info script does, dumps data about indexes into the packs, but I'm really not sure Aug 15 23:15:04 Yeah, but my vague recollection is you can't, or couldn't at one point, do all of that over http Aug 15 23:15:05 and, afk, food Aug 15 23:15:27 http is the main reason the update info script exists, because a smarter protocol doesn't need that info emitted at all Aug 15 23:15:45 or do you mean generating that info as part of "do all of that"? Aug 15 23:23:00 is do_clean supposed to clean sstate as well? Aug 15 23:25:41 nope, there's a cleansstate task for that Aug 15 23:31:41 thanks Aug 15 23:32:03 np **** ENDING LOGGING AT Tue Aug 16 02:59:57 2011