**** BEGIN LOGGING AT Tue Feb 07 02:59:58 2012 Feb 07 04:59:59 Tartarus: ping Feb 07 08:28:56 good morning Feb 07 10:07:29 morning all Feb 07 10:11:23 good morning Feb 07 10:17:37 hi all Feb 07 10:18:07 hey pb_ Feb 07 10:29:52 hi florian, pb_ Feb 07 10:31:40 bluelightning: hi Paul Feb 07 10:44:15 heh, that's funny nobody has privs to edit the outdated wiki frontpage (still no mention of Yocto, oe-core, layers, ..) Feb 07 10:45:13 ant_work: actually I wonder why front page has a different right policy Feb 07 10:49:01 hi mckoan, ant_work Feb 07 10:49:13 mckoan: paranoia I'm guessing Feb 07 10:51:45 heh Feb 07 10:52:04 I think it's probably an artifact from the previous security policy that we had. Feb 07 10:52:28 When it was easier to get a wiki account (i.e. just fill in a form) the idea was to make it harder to vandalise the front page by only allowing it to be edited by admins. Feb 07 10:52:49 but you're right, now that the signup process for accounts is more tightly policed, it's possible that this extra restiction is no longer required. Feb 07 10:55:02 bluelighting: try again, can you edit it now? Feb 07 10:55:19 bluelightning: sorry, my "n" key is sticky :-} Feb 07 10:55:34 back in a moment, gotta go feed the chickens Feb 07 10:56:08 pb_: ok can edit now, will do the fixes Feb 07 10:56:10 pb_: thanks! Feb 07 10:56:36 hi otavio Feb 07 10:58:59 ok, very good Feb 07 11:34:16 hi, does anyone have experience with the Zenithink ZT-180 tablet ? Feb 07 11:39:37 pwgen: I've only seen the press around the "Spark" version; it looks interesting Feb 07 11:42:07 Going from bitbake 1.12 to 1.14 my build seems to hang on "Parsing recipes". Oe-classic. Anything I should/could delete? Feb 07 11:48:57 tasslehoffwrk: probably best to stick with 1.12 in any case... Feb 07 11:49:24 tasslehoffwrk: you could use -DDD to determine where it's getting to Feb 07 11:59:07 : i just ordered one ...lets see how long it takes until it arrives. Feb 07 12:00:02 actually I was confused, it's the ZT280-C71 that Spark is using Feb 07 12:00:19 it's 7" as opposed to the ZT-180 which is 10.2" Feb 07 12:00:43 bluelightning: ok. Feb 07 12:09:58 anyone have Psyco working on Ubuntu 11.10 (w/Python 2.7)? Feb 07 12:15:47 tasslehoffwrk: psyco is not really needed anymore Feb 07 12:16:43 bluelightning: not in oe-classic with bitbake 1.12 either? Feb 07 12:17:57 tasslehoffwrk: it has always been optional Feb 07 12:18:34 ant_work: yeah, but since it NOTE's about it I figured it offered substansial speed improvements :) Feb 07 12:19:52 tasslehoffwrk: it doesn't, the speed improvement was minimal Feb 07 12:33:37 ack Feb 07 12:44:35 hmm anyone got any idea how to insert a file into the rfs before building it? Feb 07 12:45:07 hmm, something is generating the pango-module-basic-fc package, but the pango package only lists it in RRECOMMENDS, nothing in PACKAGES (also inspected with -e), a grep for pango-module-basic-fc in OE core/meta leayers did not reveal anything Feb 07 12:45:29 any ideas who is creating it? the package misses an RDEPENDS on pango, because it uses a utility that comes with pango in its postinst script Feb 07 12:45:56 see pango.inc Feb 07 12:46:33 you're right, it doesn't have a dependency on pango and that probably is a bug. Feb 07 12:46:59 seems moderately unlikely that pango-module-basic-fc would et installed without pango, though, so it probably is not a very serious problem Feb 07 12:47:15 ooh PACKAGES_DYNAMIC_.. I am not yet familiar with it Feb 07 12:47:39 pb_: well this unlikely thing happened here during updating :) opkg seemed to have messed up the install order Feb 07 12:47:53 because it did not know that that the module was depending on pango Feb 07 12:48:07 so the postinst script failed Feb 07 12:48:09 ah, I see Feb 07 12:48:13 yeah, that sounds plausible Feb 07 12:48:54 so how does this dynamic thing work, will RDEPENDS_${PN}-module-basic-fc = "${PN}" Feb 07 12:48:57 be enough for me? Feb 07 12:49:40 not really, that would still fail for the other modules. Feb 07 12:50:02 it's a bit weird, do_split_packages() is meant to add the dependency on PN automatically Feb 07 12:50:24 hi, all! Feb 07 12:50:30 if you look at package.bbclass and search for "This is *really* broken", that's the code that's meant to be doing it. Feb 07 12:50:52 so I guess you need to figure out why it isn't doing the riht thing and fix it up. Feb 07 12:51:15 what is newer bitbake replacement for -DDD? I want to get full output of commands run by bitbake on screen Feb 07 12:51:20 I see, thanks for the hints Feb 07 12:51:22 hi slapin Feb 07 13:03:58 Jin^eLD: hi! Feb 07 13:13:31 RP__: ping Feb 07 13:32:36 slapin: pong? Feb 07 13:33:27 RP__: Hi! where could I see newer bitbake documentation online, for v1.15 and above? Feb 07 13:34:00 RP__: I want to see maximum debug output from software bitbake runs Feb 07 13:34:13 slapin: I'm not sure there is an online rendered version of it but it is in the doc directory of bitbake Feb 07 13:34:17 slapin: -DDD should work Feb 07 13:35:06 RP__: -DDD displays some bitbake debugging, but not what software (e.g. fetchers, git, etc.) outputs Feb 07 13:35:32 slapin: FWIW, debug info always goes to the task logs Feb 07 13:36:12 slapin: Look in WORKDIR/temp/log* Feb 07 13:37:02 hmm anyone know where i do not have /etc/network/interfaces in my rfs after building? Feb 07 13:37:05 RP__: some logs are not informatory enough, like with unappliable patches. Feb 07 13:37:36 RP__: so I can't get anything on stdout/stderr? Feb 07 13:37:58 slapin: The trouble was always the corruption you get with that and the mulitthreading Feb 07 13:38:10 slapin: so no, you need to look at the log files for the tasks Feb 07 13:38:25 slapin: Feel free to open bugs with examples on bugzilla.yoctoproject.org for cases you don't think there is enough information though Feb 07 13:39:14 RP__: I just do some build validation scripts, so I will have to scan tmp* for logs Feb 07 13:39:50 RP__: thanks for information, sorry for bothering you Feb 07 13:40:03 slapin: warnings/errors will always make it to the console Feb 07 13:43:20 RP__: bitbake doesn't always detect bad things which are critical for particular cases, e.g. bad configure, missing python modules, everything seems OK but we get buggy build and spend weeks looking for cause of trouble, so I want to filter-out some patterns in logs to fail build displaying the reasons. Feb 07 13:45:24 anyone knows which gcc-4.6 that is in meta-oe/recipes-devtools/gcc? is it 4.6.2? Feb 07 13:46:06 bluelightning: I though I have seen some patch from you to show buildhistory output (expecially long git status) only when bitbake is running in debug or verbose mode, I cannot find such patch now, have you sent something like that? Feb 07 13:47:20 slapin: I'd love to add those patterns to the main configure sanity checker ;-) Feb 07 13:47:21 bluelightning: buildhistory output is written between "NOTE: Tasks Summary: " and the rest of summary, which is not optimal if it has couple of pages Feb 07 13:47:39 JaMa: it's not supposed to output anything at that point... Feb 07 13:48:01 mertsas: Its 4.6.2 plus patches iirc Feb 07 13:49:15 kaylessa, /etc/network/interfaces comes from netbase Feb 07 13:49:19 RP__: nice, thanks Feb 07 13:49:21 RP__: is there a feature in bitbake/oe-core/whatever to add package-specific sanity tests? Feb 07 13:49:37 hi btw Feb 07 13:49:53 denis what u mean netbase? Feb 07 13:50:11 kaylessa, the netbase recipe Feb 07 13:50:26 slapin: see insane.bbclass Feb 07 13:50:50 slapin: They're not recipe specific but it could likely be extended to that Feb 07 13:50:55 btw you seem to know my name, do I know you? (maybe I forgott but your name doesn't ring a bell, sorry) Feb 07 13:50:57 bluelightning: http://paste.pocoo.org/show/547253/ it looks like this here Feb 07 13:51:08 slapin: well, it currently allows disabling tests on a per package basis Feb 07 13:51:27 btw, what browsers are there in core based oe beside midori,firefox,eve and evolution Feb 07 13:51:45 hi if i dont have /etc/network/interface can i still use nfs? Feb 07 13:51:57 JaMa: I don't get that, it almost looks like it's doing verbose printout of the commands... Feb 07 13:52:02 kaylessa, no idea, look at what the init script does Feb 07 13:52:34 JaMa: if we can figure out what triggers it I can definitely fix it, but at the moment I'm unsure Feb 07 13:52:42 kaylessa, also don't forget --numeric-owner tar option, and to extract as root,when extracting for NFS Feb 07 13:53:39 RP__: I want to check for existance of some .so after do_configure/do_compile and fail whole build if found, for several packages, but not for all. Feb 07 13:55:44 so inorder to have the netbase i can install it into my initramfs right? Feb 07 13:55:46 bluelightning: do you also have sh -e shebang in x86_64-oe-linux/bblayers-1.0-r0/temp/run.buildhistory_commit.* ? Feb 07 13:55:48 RP__: so having (disableable) do_sanity() in .bb should be nice Feb 07 13:59:10 bluelightning: and it also has sed +x inside, whole script: http://paste.pocoo.org/show/547260/ Feb 07 14:03:33 bluelightning: maybe it's not shown for you because poky's bitbake has this http://git.openembedded.org/bitbake/commit/?h=master-poky&id=ebaa9dc62246635efa634b7330d97606c6c02eb4 Feb 07 14:05:06 RP__: how come ^ is not in bitbake upstream? Feb 07 14:14:36 bluelightning: There was a big debate about that one and it was decided bitbake should always use set -x to have the most useful logs. I didn't like it in poky so I never took the change Feb 07 14:14:50 bluelightning: I've left it as I thought it should probably become a feature flag Feb 07 14:14:53 argh :( Feb 07 14:15:28 bluelightning: I've found the logs it generates to be less readable in many cases :/ Feb 07 14:15:29 RP__: at the moment it's resulting in buildhistory spewing output at the end when committing; what's the correct way to fix that for non-poky bitbake? Feb 07 14:17:12 bluelightning: Not sure, depends how that code is being executed I guess and why its ending up on the console Feb 07 14:17:54 RP__: bb.build.exec_func() Feb 07 14:18:28 bluelightning: hmm, you're running that without the logging redirection? Feb 07 14:18:38 (exec_task does that I guess) Feb 07 14:18:48 perhaps that was ill-advised Feb 07 14:19:10 on my part Feb 07 14:19:16 bluelightning: That set -x needs to become conditional Feb 07 14:19:26 upon? Feb 07 14:20:37 A parameter to the function. exec_task could make the choice (VerboseLogs || BB_VERBOSE_SCRIPTLOGS) and default it to off so your caller is fine Feb 07 14:21:39 ok Feb 07 14:44:22 * kergoth` isn't a big fan of set -x. it can always be set explicitly in the recipe or class when debugging, or manual echos Feb 07 16:00:05 RP__: I see that the TSC has discussed meta-oe. Have you discussed systemd yet? It seems like the threads are going nowhere to me. Right now meta-oe is unusable unless you use angstrom/systemd. I think we need the TSC to make a decision on this, or more generally, on meta-oe usability with non-angstrom distros, if it hasn't already, and make that policy somewhere public and obvious. as it stands today, I'm ready Feb 07 16:00:08 to stop using the layer and suck the bits I need into my own instead, as it's unuseable Feb 07 16:00:24 though this recent init/systemd patchset looks interesting, will have to look that over Feb 07 16:00:43 kergoth`: Its not been raised to the TSC until now Feb 07 16:00:59 kergoth`: We can put that on the agenda for the next meeting though Feb 07 16:01:04 maybe it'll work itself out, but the current state just won't do Feb 07 16:01:09 fwiw: it's usable for SHR with and without systemd Feb 07 16:01:15 that'd be great, thanks Feb 07 16:01:46 in particular, busybox-httpd depends on systemd, and the busybox recipe is appended to inherit systemd. so building busybox sucks in systemd now :| Feb 07 16:03:26 but not runtime dependency.. that's pulled only by polkit if you don't bbappned it correctly Feb 07 16:03:47 so for images without systemd I have to build systemd but it's not in image Feb 07 16:03:59 yeah, true Feb 07 16:04:16 sorry, my build times are bad enough, I'll ditch meta-oe before I let systemd come into my builds Feb 07 16:04:40 * JaMa would like to ditch gcc-* to get better build times :) Feb 07 16:04:50 hehe Feb 07 16:04:53 if only Feb 07 16:05:38 JaMa: oh, I never managed to repro the hang failure in the shr chroot. did manage to hit it once or twice at random during my 40 or so builds yesterday, though, on my own box. think I'll have to continue to try to fix it via code inspection Feb 07 16:05:40 heh Feb 07 16:06:23 kergoth`: on slow host I had it more often Feb 07 16:06:48 kergoth`: so if you're trying to reproduce it on very fast machine then maybe it wont show Feb 07 16:07:09 ah, okay, that's interesting Feb 07 16:07:12 how many cores? Feb 07 16:08:01 on my machine (quad phenom) I got it maybe once in 20 builds, on virtual machine with 2 cores It was something like once in 5 builds maybe Feb 07 16:08:19 ah, okay, thanks Feb 07 16:08:22 I'll give that a shot Feb 07 16:09:00 and on slow host with: Feb 07 16:09:00 PARALLEL_MAKE = "-j 2" Feb 07 16:09:00 BB_NUMBER_THREADS = "1" Feb 07 16:09:49 k, hopefully a slow vm helps. I've been fighting this issue on and off for days as a background task, today it's my focus Feb 07 16:24:20 hi otavio Feb 07 16:25:01 I'm reviving the chromium recipe which depend on some mozilla stuff(at least it did in oe-classic) Feb 07 16:25:29 so you marked some libs as PRIVATE_LIBS Feb 07 16:27:18 what's PRIVATE_LIBS for exactly, it's not mentioned in the doc and package.bbclass isn't very clear on the intent of PRIVATE_LIBS Feb 07 16:27:41 I guess I'll contiune to do nss, nspr etc... recipe while waiting for a response Feb 07 16:27:53 I guess the way to go is to make firefox depend on that Feb 07 16:27:56 right?> Feb 07 16:27:58 s/>// Feb 07 17:34:08 kergoth: yeah, I also found meta-oe unusable (for various reasons) and have resorted to just pulling individual recipes into my local layers. it is a bit of a shame. Feb 07 17:36:50 pb_: in the end it was not a missing RDEPENDS, I was wrong, it was a missing /etc/pango, so when postinst scripts were doing pango-querymodules >/etc/pango/pango.modules - it was clearly failing Feb 07 17:38:05 pb_: I guess you saw the discussion about splitting up meta-oe? Feb 07 17:38:21 pb_: there is some momentum to get this fixed Feb 07 17:39:08 bluelightning: even more splitting is going on? Feb 07 17:39:26 pb_: I wish koen would think about someone other than angstrom once in a while. I realize that's his priority, but the fact remains that oe has always intended to remain flexible and distro agnostic, not paying attention to that is a problem Feb 07 17:39:28 heh Feb 07 17:39:28 florian: it's being discussed: http://comments.gmane.org/gmane.comp.handhelds.openembedded/50131 Feb 07 17:40:26 kergoth`: yeah, indeed. right now, as far as I can tell, meta-oe is basically just an extension of meta-angstrom by another name. Feb 07 17:40:44 bluelightning: yes, that sounds like progress Feb 07 17:40:46 * kergoth` nods, exactly Feb 07 17:41:12 to be fair, SHR uses meta-oe successfully as well Feb 07 17:41:26 but that could be because it just happens to align well with angstrom Feb 07 17:43:57 ah yes, true Feb 07 17:44:06 heh, I see saul does not approve of patches being signed off with kisses Feb 07 17:44:18 I thought that was rather sweet Feb 07 17:44:58 anyone can tell what is the recipe that install tree command ? Feb 07 17:45:30 searchs on google only return info about multiple tree of open embedded .... Feb 07 17:48:00 nine_pt google for oe-classic and oe-core and than decide yourself Feb 07 17:48:44 I don't think his question is regarding oe-classic vs oe-core, he wants to know which recipe will give him /usr/bin/tree, afaict Feb 07 17:49:39 kergoth`: yep, what I wan't is to have /usr/bin/tree Feb 07 17:49:58 there is a tree_1.5.2.bb in oe-classic Feb 07 17:50:00 sorry for the misleading explanation ... Feb 07 17:50:38 also treecc_0.3.6.bb if you don't mind a few extra letters Feb 07 17:51:04 pb_: thanks Feb 07 18:07:26 hmm, mailing list server seems to be struggling today Feb 07 18:10:25 what's wrong with it? Feb 07 18:10:45 I'm seeing replies to replies but not the original reply Feb 07 18:11:14 ah, hm. Feb 07 18:18:50 when creating an SDK (meta-toolchain), what would the recommended approach be for allowing the SDK to build for multiple machine targets which are using the same TARGET_ARCH ? Feb 07 18:19:08 ^ I'm using oe-classic if it makes a difference Feb 07 18:22:32 can you give me an example of a mail that didn't arrive? Feb 07 18:22:47 I haven't noticed that effect myself but maybe I haven't been paying enough attention Feb 07 18:23:41 pb_: Koen's reply to my patch that rightly pointed out PACKAGES shouldn't be sorted, for example Feb 07 18:23:52 on the oe-core list Feb 07 18:24:01 I got Otavio's reply to that reply Feb 07 18:24:20 of course the bottleneck could be at my end, it's hard to tell Feb 07 18:24:47 http://lists.linuxtogo.org/pipermail/openembedded-core/2012-February/017427.html Feb 07 18:24:49 that one? Feb 07 18:25:04 hi! Feb 07 18:25:26 I'm just about to leave the office now but I'll have a look at the logs later this evening and see if anything seems to be wrong. Feb 07 18:25:29 pb_: that's the one Feb 07 18:25:44 is it possible to put several systemd's .service in one SYSTEMD_SERVICE_${PN}? Feb 07 18:25:48 I did receive that email, fwiw, so it did go out to at least some recipients. Feb 07 18:25:57 pb_: heh, as we spoke it just appeared in ny inbox Feb 07 18:26:10 heh Feb 07 18:26:17 any clue from the headers where it got delayed? Feb 07 18:26:40 discovery isn't running any particularly high load, 0.18 last time I checked, so it shouldn't have been introducing any significant slowdown Feb 07 19:10:27 re Feb 07 19:49:44 http://www.linuxfoundation.org/news-media/blogs/browse/2012/02/embedded-and-mobile-linux-trends-shaping-2012 Feb 07 21:11:01 Crofton, Crofton|work ping? Feb 07 21:32:26 ka6sox, pong Feb 07 21:32:34 he crofton Feb 07 21:34:21 hi Crofton|work Feb 07 21:36:04 are you able to go to 2 back to back shows...ie will I see you at ELC? **** ENDING LOGGING AT Wed Feb 08 02:59:57 2012