**** BEGIN LOGGING AT Thu Oct 20 02:59:57 2011 Oct 20 03:16:20 otavio: you should explain whats not working Oct 20 03:16:55 It lacked PSEUDO environment var during rootfs and now I am doing a new try Oct 20 03:20:04 fuck Oct 20 03:20:11 it was a ordering issue Oct 20 03:24:35 khem: i am fixing systemd package to user useradd class but base-passwd need to be installed *before* it otherwise you lose the package Oct 20 03:26:29 khem: no ... didn't work Oct 20 03:29:00 how about native package for passwd Oct 20 03:32:43 khem: which? Oct 20 03:33:04 khem: shadow-native? Oct 20 03:34:04 khem: it is built Oct 20 03:40:02 khem: where the PATH env is set? Oct 20 03:40:15 khem: I fear groupadd is not being find in the env Oct 20 06:17:31 Does https://github.com/kergoth/oe-core/commit/2211989 seem relatively sane? Oct 20 06:18:04 had an undefined pattern from the metadata matching everything i fed it :P Oct 20 06:20:49 ouch, creating something that matches nothing on empty seems more sane to me. Oct 20 06:23:20 * kergoth nods Oct 20 06:23:33 okay, thanks, just wanted to make sure I wasn't out in left field Oct 20 06:23:37 * kergoth goes to bed Oct 20 06:25:58 nn Oct 20 07:10:40 good morning Oct 20 08:34:14 RP__: Hi, http://lists.linuxtogo.org/pipermail/openembedded-core/2011-October/011220.html still wasn't included (this version was without that do_install_append), was it overlooked or is there some problem with it? Oct 20 08:38:17 JaMa: no problem as such, I was just wondering about backwards compatibility Oct 20 08:38:46 JaMa: e.g. will images that currently contain libGL still include it. I think the answer is yes but I wasn't 100% sure Oct 20 08:50:30 hi JaMa, RP: btw adding xinput-calibrator ts calibration is working on core-image-sato Oct 20 08:51:02 seems the xcb issues have been solved Oct 20 08:54:06 RP__: this went in IIRC http://patches.openembedded.org/patch/13131/ so PR bump in packages depending on libgl would be usefull too Oct 20 08:55:34 JaMa: You know my feelings on PR bumps :/ Oct 20 09:03:38 RP__: yes I know Oct 20 09:09:12 JaMa: I'll take a patch but I want to get out of this situation of having to manually figure this stuff out Oct 20 11:33:14 Anyone used useradd class? I am trying to debug a failure and having problem with it Oct 20 11:50:52 otavio: Its used by a variety of recipes in oe-core iirc Oct 20 11:57:54 RP__: it seems to be not creating a group in rootfs Oct 20 11:58:11 otavio: It could be a bug in the class :/ Oct 20 11:58:27 RP__: in fact the class seems to be doing right Oct 20 11:58:46 RP__: and if I get the command, put the paths and pseudo byhand, it works Oct 20 11:59:01 RP__: i mean, if I run byhand it Oct 20 11:59:27 otavio: odd. I wonder why its not working then :/ Oct 20 11:59:46 RP__: http://paste.debian.net/138232/ Oct 20 11:59:49 RP__: me too Oct 20 12:00:05 RP__: i tried to find the reason yestarday until 2am heh Oct 20 12:01:17 otavio: offhand I don't know enough to be able to help, sorry :( Oct 20 12:01:26 RP__: right Oct 20 12:01:28 RP__: np Oct 20 12:01:34 otavio: zenlinux or fray might know more] Oct 20 12:01:35 RP__: suggest someone? Oct 20 12:01:38 k Oct 20 12:01:42 (when they're awake) Oct 20 12:01:45 heh Oct 20 12:05:21 only openssh uses it atm Oct 20 12:05:41 otavio: We need to fix that :) Oct 20 12:05:56 Will improve initial boot time Oct 20 12:08:02 RP__: yes Oct 20 12:08:16 RP__: and for readonly rootfs support Oct 20 12:08:24 right Oct 20 12:08:30 RP__: did you see the duplcated commit? Oct 20 12:08:33 RP__: in -next Oct 20 12:08:45 otavio: Its just because I haven't refreshed it yet Oct 20 12:08:50 RP__: ok Oct 20 12:08:54 RP__: good Oct 20 12:09:06 zenlinux: fray: please ping me when around. Oct 20 12:30:47 I am still having build errors using the the git branch edision - I can't get it past building sysprof error as follow: Oct 20 12:31:43 http://pastebin.com/j5vf6wzj Oct 20 12:32:28 I was tackling this yesterday and I rebuild glib-2.0 as the error seemed to indicate an issue with g_datalist_get_data but it didn't help Oct 20 12:35:12 Humm .. seems sysroot usage is working Oct 20 12:35:41 but in rootfs it is not working Oct 20 14:12:50 ugh, if layer name happens to match an override, have to get BBFILE_PATTERN rather than the proper individual layer version. someday we really need to move to a different overrides separator Oct 20 14:12:56 * kergoth mutters and gets coffee Oct 20 15:09:18 kergoth: agreed Oct 20 16:00:56 useradd class works for sysroot but fails to rootfs Oct 20 16:00:58 humm Oct 20 16:23:16 fray: hi Oct 20 16:23:28 fray: got your mail about useradd class Oct 20 16:23:34 fray: and it didn't work Oct 20 16:42:30 sgw: Oct 20 16:43:06 Can you also look at the patchwork.openembedded.org when consolidated pull request is applied ? Oct 20 16:44:00 sgw: although it will update the patches but v0 and v1 wont get updated if v2 gets applied those have to change to ACCEPTED and ARCHIVED manually Oct 20 16:52:09 khem, I have enough to do with patches, I am not going to be using patchworks at this time, we are looking at gerrit and reviewboard also. Oct 20 16:53:05 i never used gerrit myself but it does seems quite nice Oct 20 16:53:47 reviewboard seems not as complete as gerrit from what I researched Oct 20 16:54:22 gerrit might be too heavy weight and it ties directly to git and requires you use gerrit tools, that was my first impression, where as reviewboard works with git but not tied directly to the git structure. Oct 20 16:54:54 sgw: but what the problem of it being tied to git if we use git? Oct 20 16:59:13 otavio: it ties to repo, not git Oct 20 16:59:54 EPARSE Oct 20 16:59:55 heh Oct 20 16:59:57 otavio: there was some issues for people Oct 20 16:59:57 RP__: ? Oct 20 17:00:11 oops Oct 20 17:00:30 There are problems with moving between repos as RP was trying to say. Oct 20 17:00:45 sgw: moving? Oct 20 17:01:08 otavio: repo is a git like tool that android uses Oct 20 17:01:14 sgw: you mean moving between oe-core and poky? for example? Oct 20 17:01:16 (approimately) Oct 20 17:01:20 RP__: ahhhh Oct 20 17:01:43 RP__: but it is mostly a porcelain above git, afaik Oct 20 17:02:06 otavio: eventually we will stop moving stuff back and forth and layers will be in place properly, at least that's my understanding. Oct 20 17:02:37 sgw: between poky and oe-core? Oct 20 17:02:50 sgw: that would be awesome .. Oct 20 17:03:33 sgw: no, we need to talk about that Oct 20 17:04:01 RP__: if distrodata ends up moving from oe-core this will be harder Oct 20 17:05:03 * otavio still fighting againt useradd grr Oct 20 17:08:42 otavio: I think you might have mis-understood what I was talking about, I was referring to using the Yocto layering tools with poky and oe-core, not gerrit and gerrit repo's Oct 20 17:09:01 sgw: combo-layer? Oct 20 17:09:04 sgw: I am using it Oct 20 17:09:26 sgw: it works well ... even thought i miss some features on it Oct 20 17:13:49 otavio: feel free to request these, I should have some time to improve it during the 1.2 cycle Oct 20 17:14:28 otavio: bluelightning beat me to the punch, please let us know any suggestions you have. I just wanted to make sure I did not leave you with the wrong idea. Oct 20 17:15:00 the android repo tool is quite nice, for people who want a submodule-like setup rather than a combo-layer/git-subtree like setup. course that doesn't help yocto. https://github.com/kergoth/oe-repo-manifest - been playing around with it a bit Oct 20 17:16:22 bluelightning: a way to restrict repository to a branch and n-auto mode Oct 20 17:16:45 bluelightning: and a way to it to commit the sync message automatically if it is only the combo-layer.conf file Oct 20 17:17:14 otavio: ok, these sound reasonable Oct 20 17:17:35 unfortunately I'm on my way out but we should discuss these further and file some enhancement bugs Oct 20 17:18:38 bluelightning: sure; let me know when you want to do that and we do it Oct 20 17:19:50 otavio: I'll be online tomorrow, we can talk further then if you're around Oct 20 17:19:59 otavio: btw will you be at ELCE? Oct 20 17:20:13 RP__: https://github.com/kergoth/oe-core/commit/2211989 seem sane to you? Oct 20 17:22:18 bluelightning: sure, tomorrow works fine. Oct 20 17:22:27 bluelightning: elce no; linuxcon brazil yes Oct 20 17:22:59 otavio: ah ok, fair enough Oct 20 17:25:28 bluelightning: will you come to linuxcon brazil? Oct 20 17:26:38 otavio: I don't think I personally will be able to but I think we are hoping someone from the Yocto project team will be able to go Oct 20 17:28:10 bluelightning: good Oct 20 18:00:33 nitink: I've just read your perl update and before I test it, is there something new about python-2.7? Oct 20 18:01:01 nitink: I've also noticed that your python patches are no longer in nitink/upgrade is there any other branch with it? Oct 20 18:01:08 hi JaMa: I could not make much progress on python yet Oct 20 18:01:26 JaMa: so far the python works all but on arm Oct 20 18:01:38 I will push it in a nitin/python branch on contrib Oct 20 18:01:48 thanks.. Oct 20 18:02:00 have you tried also qemux86-64? for me it failed too Oct 20 18:16:32 JaMa: I did not try qemux86-64, will try it Oct 20 18:22:52 JaMa: nitin/python branch is on contrib now Oct 20 18:26:38 only for poky right? Oct 20 18:32:49 nitink: the version in poky-contrib is worse then what you've sent to oe-core ML Oct 20 18:33:17 nitink: ie this looks wrong http://paste.pocoo.org/show/495694/ Oct 20 18:36:55 JaMa: thanks for catching that, looks like I lost some of the commits I sent out from my branch, the diff you sent needs to be there. other than that the commit should be same, I am testing these commits here to make sure. Oct 20 18:38:26 also you said that DEFAULT_PREFERENCE = "-27" was removed in your branch Oct 20 18:38:30 and this still have it Oct 20 18:45:43 jaMa: looks like these bits are coming from old commits, I will verify them one more time Oct 20 18:47:31 thank you for pointing out the stale pieces Oct 20 18:48:24 I am seeing these errors Oct 20 18:48:25 | Collected errors: Oct 20 18:48:25 | * extract_archive: Cannot create symlink from ./var/run to 'volatile/run': File exists. Oct 20 18:48:28 | * extract_archive: Cannot create symlink from ./var/log to 'volatile/log': File exists. Oct 20 18:48:32 anyone else seen them ? Oct 20 18:50:00 I see that this is probably being done by meta/files/fs-perms.txt Oct 20 18:54:35 sgw: regarding patchwork I think submitter should close the obsoleted ones. Once he/she sends a new version of patches Oct 20 18:54:53 sgw: I have read about reviewboard and gerrit Oct 20 18:55:08 I am still not convinced we need them for oe-core Oct 20 18:55:21 reviewboard is still nicer Oct 20 18:55:31 if compared to gerrit Oct 20 18:57:04 I did not like the directly commit feature of gerrit Oct 20 19:03:59 khem: this is what I liked heh Oct 20 19:04:02 khem: why? Oct 20 20:23:11 is there any reason, why we're using git-native but not subversion-native? with current subversion-1.7.0 it would be nice to have native version in oe-core Oct 20 20:23:23 I have already partial patches for this Oct 20 20:42:10 quick sanity check: Has anyone recently built pseudo on Ubuntu 11.10? I got a report about an older version failing to compile, and I don't know of anything I've done that would fix it. Oct 20 20:42:22 Error has to do with the always-inline definition of open() in fcntl2.h. Oct 20 20:53:25 people here are definitely running poky master atop Ubuntu 11.10 Oct 20 21:10:10 seebs: I'm running it Oct 20 21:12:42 FUCK! I FOUND IT Oct 20 21:12:43 heh Oct 20 21:13:37 RP__: useradd, for rootfs, depends on base-passwd to change the postinst to preinst as /etc/group needs to be available at rootfs build time Oct 20 21:19:32 RP__: Thanks. Maybe I'm lucky and this was fixed entirely by accident since the release the reporter was running. Oct 20 21:19:52 I am really afraid of that one, because it is pretty obvious to me that it SHOULD fail. :) Oct 20 21:21:42 RP__: I just talked to shadow upstream (he is my friend and coworker at Debian) and it seems we will integrate most patches on next release of it Oct 20 21:28:24 otavio, I'm finally available now if you'd like to discuss this issues you ran into with useradd.bbclass Oct 20 21:28:50 sorry for the delay, as I've been in meetings and other commitments all day Oct 20 21:30:11 zenlinux_laptop: I just fixed it Oct 20 21:30:17 zenlinux_laptop: hold on Oct 20 21:30:57 zenlinux_laptop: https://github.com/OSSystems/oe-core/commit/9c3cff6e514e8ccc69b9462197c7bceeb117a55e Oct 20 21:31:09 zenlinux_laptop: https://github.com/OSSystems/oe-core/commit/d0e2424c7510f09a6733c16199e86c158cbfbb7c Oct 20 21:31:14 zenlinux_laptop: those two Oct 20 21:33:37 otavio, dang - sorry I missed that. Oct 20 21:33:53 zenlinux_laptop: seems right no? Oct 20 21:34:21 otavio, yes. you've confirmed it fixes your issues? Oct 20 21:34:22 zenlinux_laptop: i just found it because i changed the groupadd matching and then grep failed Oct 20 21:35:01 zenlinux_laptop: seems so Oct 20 21:35:14 zenlinux_laptop: I will do more tests and will post it tomorrow Oct 20 21:35:29 zenlinux_laptop: it took 1 1/2 day to find this stupid thing heheh hack! Oct 20 21:36:06 well I'm quite relieved you found it Oct 20 21:36:28 otavio: great, patches upstream is great news :) Oct 20 21:36:55 RP__: Nicolas is quite nice. He will review the patches on WE Oct 20 21:37:01 otavio: and great you've found the problems Oct 20 21:37:58 RP__: did you look at the patches? Oct 20 21:39:29 otavio: not yet, just reading he backlog Oct 20 21:53:12 Well folks, I will go out ... get a bit o sun and then I will be back in 4hs or so Oct 20 21:53:22 cya Oct 20 21:53:55 Looks as though the ports changes fixed the 11.10 thing entirely by accident. Woo! Oct 20 22:01:37 seebs: nice :) Oct 20 22:01:52 fray: where were we at with the unload thing btw? Oct 20 22:02:07 I got preoccupied with something else and didn't get it done.. Oct 20 22:02:15 fray: ok, np Oct 20 22:02:16 I can do it now if you want, or first thing in the morning Oct 20 22:02:26 * RP__ was just hoping he wasn't blocking it Oct 20 22:02:35 spoke w/ seebs and we're going to change what I originally implemented.. Oct 20 22:02:54 we're going to drop PSEUDO_RELOAD completely.. and replace it with PSEUDO_UNLOAD and make sure we've got a conssitent path through the system Oct 20 22:03:03 fray: ok, sounds good :) Oct 20 22:03:06 (shouldn't be difficult.. I just have to spend the time to do it and write up the test case(s)) Oct 20 22:03:30 there was one case before where PSEUDO_RELOADED didn't do what you thought it was.. Oct 20 22:03:37 the PSEUDO_UNLOADED should bes lightly quicker.. Oct 20 22:03:54 PSEUDO_RELOADED wasn't always paid attention to on the -first- fork/exec.. but was on the second... Oct 20 22:04:05 PSEUDO_UNLOAD is honored on the first fork/exec Oct 20 22:04:14 great :) Oct 20 22:04:20 * RP__ likes speedups Oct 20 22:04:35 need all we can get atm :/ Oct 20 22:04:38 speed wise, I would expect exactly what you go witht eh pseudo-unloaded tests you did before Oct 20 22:05:26 fray: The impact of one more fork overhead might be interesting actually Oct 20 22:05:48 run the code I send before.. and change the bitbake side to use PSEUDO_RELOADED vs PSEUDO_UNLOADED.. Oct 20 22:06:01 'er.. PSEUDO_UNLOAD that would allow you to see the difference, if any Oct 20 22:06:09 I suspect it's not much Oct 20 22:06:13 fray: ah, right Oct 20 22:06:24 fray: I thought you meant this new unload would be different from that one Oct 20 22:06:42 nope.. it's one of the things I "fixed" with the PSEUDO_UNLOAD patch Oct 20 22:06:44 The _RELOADED thing is an artefact of epic and hilarious stupidity. Oct 20 22:07:18 The reason _RELOADED exists is that it took me OVER A YEAR to figure out why unsetting environment variables didn't consistently affect the environment of something being executed by a shell. Oct 20 22:07:32 Hint: What's the letter at the end of "execve" stand for? Oct 20 22:07:54 ;) Oct 20 22:07:59 So it was originally purely an ugly internal hack to deal with something I hadn't been able to figure out, and it wasn't really designed coherently. Oct 20 22:08:03 heh :) Oct 20 22:08:12 Turning it into a defined and correctly-designed feature with a meaningful name is a big win. Oct 20 22:08:43 seebs: At least it gave me a nice way to prove we had a problem :) Oct 20 22:08:47 I am still amazed at how long it took me to figure that out. Oct 20 22:09:04 * fray is still annoyed that the lzy init didn't speed things up more Oct 20 22:09:06 It's not as though I was unaware of execve. Oct 20 22:13:13 seebs: it happens, I think we've all done our share of such things :) Oct 20 22:15:17 I do more than mine, usually. I have *epic* ADHD, so easy stuff is really hard for me. Oct 20 22:15:51 It's funny, though, so I don't entirely mind. Oct 20 22:21:07 * fray goes to lay down.. leg hurting Oct 21 01:22:02 * otavio is back **** ENDING LOGGING AT Fri Oct 21 02:59:57 2011