**** BEGIN LOGGING AT Thu May 02 02:59:58 2013 May 02 07:17:54 good morning May 02 07:56:15 morning all, and mckoan May 02 07:57:52 morning all, mckoan, silviof May 02 08:05:01 we are invaded by silvios :-D May 02 08:06:28 :-P - YES! May 02 08:06:39 morning silvio_ May 02 08:06:50 mckoan, ...eheh we conquer the world :D May 02 08:08:56 silvio_: actually I hope not all of you ;-) May 02 08:11:33 morning all May 02 08:22:18 good morning all May 02 08:22:54 hi apelete May 02 08:29:57 hi bluelightning May 02 08:42:02 hi apelete , bluelightning May 02 08:46:16 hi silvio_ May 02 09:10:05 how come #oe is very quite these days ? is it a Workers' Day holidays period thing ? May 02 09:11:27 not sure... it does go through quiet periods sometimes May 02 09:11:41 I was away yesterday so I didn't see how busy it was then May 02 09:18:44 Apparently May 1 was a holiday May 02 09:19:01 yes, yesterday was more quiet than usual but that was predictable May 02 09:19:18 there is also this annoying division of technical discussion between #oe and #yocto May 02 09:22:00 I just sit on both channels... if I want to start something new about OE I always come here personally May 02 09:25:16 yeah, I do the same May 02 09:25:29 it is just confusing for some people May 02 09:25:37 and some people are only in one May 02 09:26:24 if people get their question answered I figure it's not too much of an issue May 02 09:27:20 splitting the community is bad May 02 09:27:23 true. I always got the help I needed here, and didn't even know there was a #yocto channel :) May 02 09:27:38 and we miss chances for people to see all the discussion May 02 09:27:47 :) May 02 09:29:59 I don't see how you could practically deprecate #yocto though May 02 09:30:28 there is much more split with the mailing lists May 02 09:31:00 dunno May 02 09:31:10 I'll blame jefro, since he is asleep :) May 02 09:31:20 as should I be May 02 09:31:43 * Crofton|work has a tree limb rubbing on his house next to the bedroom :( May 02 09:32:02 since yocto is a project of its own, it deserve its own channel I guess. people working with yocto would certainly like to have a specific channel May 02 09:32:07 also, how do you request a cherry pick for Dylan? May 02 09:32:33 but yocto is a project to make embedded dev easier and oe is the build systsem May 02 09:32:55 but people get confused about that and have oe technical discussion in #yocto May 02 09:33:19 Crofton|work: if you mean for OE-Core dylan branch, you can either ask me directly (I'm handling the dylan branch maintenance) or send a pull request with the cherry-pick on top of dylan, I don't mind either way May 02 09:33:44 did you see the fix for gcc 4.9 on the list? May 02 09:33:54 it came from a guy I know May 02 09:34:07 he is confused by khems request to cherry pick for Dylan :) May 02 09:34:45 do you need to wait for it to go into master? May 02 09:36:38 I'd prefer to do it that way May 02 09:36:55 I'll let Josh know May 02 09:37:00 nothing stopping me testing it on top of my own dylan-next branch though I guess May 02 09:37:14 yeah, I think it iwll save us alot of emails :) May 02 09:37:24 as distros move to 4.8 May 02 09:37:36 hm.. how it is we still have gcc 4.6 in meta-openembedded ? May 02 09:38:03 but it doesn't build unless you ask for it, right? May 02 09:38:18 people get attached to compilers and hate change May 02 09:38:20 ant_work: you mean in toolchain-layer? May 02 09:38:40 yes, I'd think it's deprecated layer, isn't ? May 02 09:39:23 not sure; it's hard to tell who is using things unless they pipe up and say so May 02 09:39:49 iI hope meta-openembedded will be soon purged of meta-systemd and tooclhain-layer May 02 09:40:25 the meta-openembedded layer it' still scares many Yocto dev May 02 09:40:46 it's a *lot* better now IMHO May 02 09:40:59 ant_work, they need to get over that and submit patches May 02 09:41:09 I'm a bit peeved that nobody has taken the time to eliminate the busybox bbappend though May 02 09:41:23 nobody seems willing to look into it :/ May 02 09:41:42 what is in the bbappend? May 02 09:41:52 as it looks to me, the development is going on in oe-core May 02 09:42:03 Crofton|work: it's something to do with busybox-syslog May 02 09:42:05 for systemd May 02 09:42:24 why it was not moved to meta-systemd earlier with the rest of the systemd stuff I don't know May 02 09:42:55 but now we have systemd in OE-Core it either needs to be merged there or just go away depending on whether it's still needed or not May 02 09:43:11 yep May 02 09:43:14 pasted that, we should really make an appealing layer with just 'other' recipes, not present in oe-core. May 02 09:43:16 I don't know enough about what it's supposed to be doing to evaluate it myself... May 02 09:43:26 that is the issue for me also May 02 09:44:02 I need to see how systemd from oe-core works May 02 09:44:07 but people get confused about that and have oe technical discussion in #yocto May 02 09:44:07 I don't see how to fix that. from my point of view, the only way to get the difference between yocto and oe is to start working with either one, and get redirected by people who know when you come on either channel asking for help (that's what I did, except I started using oe-classic and then oe-core, never needed yocto) May 02 09:44:08 ant_work: that's mostly what meta-oe (the layer) is now May 02 09:44:10 well, systemd was badly introduced an is taking time to merge May 02 09:48:39 morning all May 02 09:48:57 hi pb_ May 02 09:49:00 hi pb_ May 02 09:49:02 hi bluelightning May 02 09:49:54 * Crofton|work fails to see what is good about this morning :) May 02 09:50:21 I'll be pruning some tree limbs this afternoon May 02 09:51:33 bluelightning: of the initial 2 duplicates (udev and xserver-nodm-init) we got rid of udev in meta-oe. xserver-nodm-init is taking longer and in the meanwhile systemd has urged May 02 09:52:13 ant_work: AIUI a few people have been working on the xserver-nodm-init issue May 02 09:53:48 maybe we' need a lot of them ;) May 02 09:57:29 hi apelete May 02 09:57:33 hi ant_work May 02 09:58:14 gm pb_ May 02 10:08:01 * pb_ still on the trail of the mystery mode stomper May 02 10:08:13 very perplexing May 02 10:10:26 JaMa: that commit renamed the connman bbappend to .bbappen btw May 02 10:11:07 jackmitchell: (meta-oe 36d5c2f3736bc1121f700fd11130bd4675afba0a) May 02 10:11:10 er opps May 02 10:11:16 JaMa: ^ May 02 10:41:49 bluelightning, some guy on the gumstix list wants to disable just the meta-ope bbappend for qt4 from meta-oe May 02 10:42:00 is the a way to BBAMSK it? May 02 10:42:36 Crofton|work: sorry I can't decipher what you mean... ? May 02 10:42:50 gah, let me send you the email May 02 10:44:25 basically, can you mask specific bbappends May 02 10:45:36 bluelightning: huh.. fixing that now May 02 10:46:34 yes, I think BBMASK works with .bbappends May 02 10:54:00 Crofton|work: right, I also think it should work May 02 11:07:22 obviously it does sort of beg the question of whether the .bbappend ought to be there in the first place, if there are users who are trying to turn it off. May 02 11:08:55 pb_: hence why it's been removed in master (and dylan) May 02 11:09:51 ah, very good May 02 11:10:05 in that case it sounds as though crofton's dude ought to get with the programme. May 02 11:13:38 pb_: I guess they are sticking with danny at the moment May 02 11:16:21 yeah, I guess that's fair enough. May 02 11:17:09 I just updated our tree from danny-ish to dylan-ish and it did take me about a week of debugging to get everything working again afterwards so I can imagine that not everyone would want to rush into it. May 02 11:19:54 actually, I guess that's a slightly premature claim since I don't quite have everything working yet. nearly though :-) May 02 11:23:43 ah, it seems that 6775feb9fe935ab01fd9cae2b2d3fce5824a9a72 might be the cause of my mystery mode stomping May 02 11:28:55 possibly combined with something being rotten in oe.cachedpath and/or package.bbclass May 02 11:37:51 * pb_ experiments with reverting parts of 6021e309e69d823e1467648aee12a32182945569 May 02 11:38:49 it was some guy on the gumstix list May 02 11:39:11 good to know you can locally squash bbappends though May 02 11:51:26 http://www.emacswiki.org/pics/static/TabsSpacesBoth.png May 02 12:37:38 hrm, no, stomping still occurring there. May 02 13:01:24 ah, found it! May 02 13:01:39 pb_: what's the cause? May 02 13:07:10 it's an unfortunate interaction between fixup_perms() and the change that RP made to hardlink the files into /usr/src/debug rather than copying them. May 02 13:08:27 what's happening is that, in the case at hand, the files in ${S} are already hard links into a shared git checkout, and package.bbclass is creating additional hard links into packages-split. Then, fixup_perms() comes along and starts tinkering with the file modes (presumably under the impression that it has the files to itself) but, because they're hardlinked all the way back to the original repository, it ends up changing the modes in the git working tree as well. May 02 13:09:13 pb_: that does sound like an unfortunate set of interactions :/ May 02 13:09:41 I'm not entirely sure why fs-perms.txt has an entry for /usr/src/debug in the first place. There doesn't seem to be any totally compelling reason why the source files must be 0644 and not any other mode. May 02 13:10:07 pb_: it was more to get consistent owner information iirc May 02 13:10:53 ah, right. I think having fixup_perms() try to change the owner would be relatively harmless, since it will take effect inside pseudo's mind but fail to actually persist on the filesystem. May 02 13:11:16 so I could change the mode from 0644 to - in fs-perms.txt and that would probably be enough to stop the problem from happening. May 02 13:12:40 pb_: probably, yes May 02 13:16:41 ok, I'll try that May 02 13:27:20 I also found some useless-seeming calls to chmod and chown while I was debugging this issue. I've sent a patch to remove those. May 02 13:39:04 right, yes, changing 0644 to - in fs-perms.txt appears to be sufficient to prevent the stomping. May 02 16:29:05 moin May 02 17:51:29 god i hate the qt4-x11-free do_compile May 02 17:52:36 :) May 02 17:53:00 it's a pig to be sure May 02 20:46:10 if I want to use do_split_packages(), I must call it in a populate_packages_prepend() function, right? there is no other way, for example by using prefuncs etc.? **** ENDING LOGGING AT Fri May 03 02:59:58 2013