**** BEGIN LOGGING AT Fri Nov 13 02:59:59 2015 Nov 13 03:43:21 Hmm, I'm still seeing * extract_archive: Cannot create symlink from ./var/log to 'volatile/log': File exists. Nov 13 03:43:28 when I try to build my image Nov 13 03:43:38 any pointers on tracking down the package that is causing that? Nov 13 04:00:56 should be able to use oe-pkgdata-util to search the pkgdata by target filesystem path to find it Nov 13 04:01:30 i'd also recommend confirming that the fs-perms.txt override is working Nov 13 04:02:18 kergoth: how would I confirm that it is working? Nov 13 04:49:01 hrm, it appears that I've gotten it to work. Unclear what the preceise cause was, but it looks like having base-files provide /var/log/journal was the issue. Nov 13 04:49:39 Of course, it wasn't creating a symlink (I had done 'volatiles_remove="log"'), so the error doesn't seem like what I'd expect Nov 13 05:19:51 So I’m trying to compile elibc and using it with a pre-compiled dist, and that’s pretty much working, but it now looks like I have access to 2.18-r0 while the target is using 2.18-r1. Not sure where to obtain/how to find the [patch??] that changes r0 ——> r1. Help? Nov 13 05:26:31 In fact here’s some other guy that seems to be doing the same Nov 13 05:26:32 http://pastebin.com/BpxU4V8S Nov 13 09:30:26 another case of them: "hey I read this in the Yocto Kernel manual and it doesn't work" me: "sorry, $vendor doesn't use kernel-yocto" Nov 13 09:30:32 we need a kernel manual lite ;-) Nov 13 09:40:14 kergoth: I'd like to deal with the sambe-depends in the way discussed yesterday with JaMa Nov 13 09:40:24 objections? Nov 13 12:37:26 sno: found the libusb race Nov 13 12:43:37 rburton: impressive Nov 13 13:13:32 sno: the recipe-specific autotools thing is in my staging branch btw Nov 13 13:15:30 rburton: that's good - otavio want's the openjdk-8 merge, and it needs it ... Nov 13 14:01:58 congrats for the release folks! Nov 13 14:57:10 is there a way to do "-c devshell" before patches are applied? Nov 13 14:57:44 sadly not, we should really add another task for that Nov 13 14:58:03 kergoth: ok, thanks Nov 13 15:06:58 When I have one patch in a series that has a typo, should I re-submit the series to the list, including the redundant patches, or just update the branch and indicate this in a reply to the cover letter email? Nov 13 15:09:54 kergoth: I try to reproduce your "attr" situation - seems there seems to be another dependency Nov 13 15:10:16 I build with installed attr but don't get the QA issue Nov 13 15:29:13 kergoth: for oe-core i'm happy with a mail saying "fixed a typo in this patch on the branch" Nov 13 15:32:26 rburton: Damn! I had re-send a complete series for just one typo Nov 13 15:54:05 JaMa, kergoth - I sent an example how I would imaging the conditionally depends in samba could be handled Nov 13 16:18:46 anybody run "bitbake -g -u depexp" on Ununtu 14.04? I am getting 'ImportError: No module named gobject' and nothing in doc about this Nov 13 16:21:15 nm.. env setup issue Nov 13 16:38:07 argh so many upstream still use AM_MAINTAINER_MODE Nov 13 16:39:56 i don't really see what that buys you, other than removing a few rules at the risk of using outdated files, unless it was intended to deal with scms and the fact that they don't always reproduce file times correctly on checkout or something Nov 13 16:40:04 hmm Nov 13 16:40:09 * kergoth shakes head at autoconf Nov 13 18:02:58 WARNING: QA Issue: dbus-glib-native: configure disables maintainer mode (add --enable-maintainer-mode to EXTRA_OECONF) [autoconf-maintainer] Nov 13 18:03:01 mwhaha Nov 13 18:06:07 nice Nov 13 18:06:46 it'd be nice to take things like that and add them to recipetool create when we add the check, so new recipes automatically get the necessary bit Nov 13 18:06:49 as a later task Nov 13 18:06:52 * kergoth ponders Nov 13 18:20:48 look what I found justanotherboy: Nov 13 18:20:48 https://mender.io/blog/build-info-yocto-2 Nov 13 18:25:01 benjamirc: Nice find! Personally I would set the var in my local.conf file instead of the bbclass Nov 13 18:27:37 I saw Jefro published it Nov 13 19:33:08 so I decided to do a crazy thing, which is to try to read some documentation Nov 13 19:33:14 which required me to render some documentation Nov 13 19:33:27 which leads to the discovery that texinfo 6.0 appears to choke on the binutils manual for gas Nov 13 19:34:17 Apparently, there is a change in how indexes are built, and the net result is that there's a \initial {\\} in as.cps, which creates a very strange diagnostic, no actual display of that initial character... and an exit status that aborts a build. Nov 13 19:34:36 (This after using the ASSUME_REQUIRED, etc., changes mentioned in the texinfo bbclass to actually build a texinfo-/native in order to render docs.) Nov 13 19:34:55 I have no real request here, I just wanted to gripe about it. That said, if anyone happens to have resolved this, I"d love to know how. Nov 13 19:42:34 Solution: Use texinfo 4.x. Oh, wait. That fails with "egrep: invalid range end" because someone spelled a range as [A-z], which only works in some locales, apparently. But if you force locale to C, that builds cleanly. Nov 13 20:05:16 kergoth: 118 warnings from core-image-sato build, makes me think maybe we should pass —enable-maintainer-mode and force —disable where it breaks stuff (binutils failed with it) Nov 13 20:05:45 sheesh Nov 13 20:06:10 either that or it could add the arg based on AM_MAINTAINER_MODE existing, dynamically, at runtime, but that'd add additional overhead to do_configure, which wouldnt' be ideal Nov 13 20:06:56 well, that would still need manual tweaking for binutils, which has AM_MAINTAINER_MODE but fails to build if you enable it Nov 13 20:08:28 ah Nov 13 20:08:29 indeed Nov 13 20:28:08 i have a question about the purpose of the devshell Nov 13 20:28:28 lets assume i run bitbake -c devshell xyz Nov 13 20:28:53 my local.conf is setup for a raspberrypi Nov 13 20:29:09 should i be able to test a "configure" within the devshell? Nov 13 20:29:21 devshell gives you a shell in the same environment that our tasks run. it doesn't make the task functions available to you directly Nov 13 20:29:33 you can bitbake -c configure first, then -c devshell, and run the ../temp/run.do_compile shell script Nov 13 20:29:59 kergoth: thanks Nov 13 20:30:03 np Nov 13 20:30:25 i always wondered what the run.... files are for Nov 13 20:30:27 ;-) Nov 13 20:30:40 its the script bitbake runs to run the task :) Nov 13 20:30:48 if you want to make changes to the sources quickly and rebuild, e.g. for a build/debug/test cycle, you could also make use of devtool Nov 13 20:31:06 devtool modify -x busybox src/busybox; will extract/patch the sources into src/busybox/ and arrange for the recipe to build from there Nov 13 20:31:24 will try that. Nov 13 20:31:27 then you could modify the sources and just bitbake -C compile busybox for example to rebuild it on demand Nov 13 20:37:26 hi all Nov 13 20:37:44 wondering, is there support for ar9331 for yocto somewhere? Nov 13 20:59:27 hmm Nov 13 20:59:38 i am developing a new recipe for a piece of software Nov 13 20:59:46 it is autotools based Nov 13 21:00:03 so i inherit the autotools Nov 13 21:00:32 configure fails with "cannot run C compiled programs. If you meant to cross compile, use `--host'." Nov 13 21:00:56 if i pass --host=arm-poky-linux-gnueabi configure will pass Nov 13 21:01:24 i wonder what i am doing wrong. why is my autotools based recipe not working? Nov 13 21:01:57 autotools.bbclass passes --host already Nov 13 21:02:02 unless you're overriding do_configure Nov 13 21:02:10 https://paste.kde.org/pvulsmhgw Nov 13 21:02:20 kergoth: here is the recipe Nov 13 21:02:35 i have a configure_prepend Nov 13 21:02:45 does this count? Nov 13 21:02:48 that should be working fine Nov 13 21:02:54 as i said, we pass --host by default Nov 13 21:02:57 along with a ton of other args Nov 13 21:02:59 see the do_configure log Nov 13 21:04:30 damnit, i think file-checksums is failing for me. anyone else run into this, where the file mentioned inf ile-checksums is modified yet it doesn't cause a rebuild? Nov 13 21:04:31 hmm Nov 13 21:06:26 kergoth: can you elaborate, I think I may have seen something like this. Nov 13 21:06:31 kergoth: SRC_URI checksums ? Nov 13 21:06:53 no, file-checksums. it's a bitbake flag that makes it checksum the mentioned file on disk, so changes to it cause a change to that var's checksums Nov 13 21:07:14 kergoth: i think their autogen.sh is the problem. the last line of this script is "./configure". its not passing anything Nov 13 21:07:21 fixing IMAGE_CMD_wic to re-run do_rootfs when teh wks file changes, but it's not behaving Nov 13 21:07:42 fmeerkoetter: ah, indeed. also, autotools.bbclass will automatically re-run autcoonf/automake/etc via autoreconf, so if that's all their autogen wants to do, there's no need to run it at all Nov 13 21:09:37 kergoth: thanks a lot. removing my custom prepend fixed it! Nov 13 21:09:49 nice Nov 13 21:09:59 kergoth: not used it Nov 13 21:10:02 yet Nov 13 21:10:03 get_checksum_file_list Nov 13 21:10:13 thats what should get the list of files ? Nov 13 21:10:21 and how do we spcify it in recipe ? Nov 13 21:12:08 oh! i think it's only supported for tasks Nov 13 21:12:12 not generically for all variables Nov 13 21:12:14 thats my problem Nov 13 21:12:21 khem: it's just a list of filesystem paths.. Nov 13 21:12:55 * kergoth tests Nov 13 21:24:48 make minor change to configure options .. random package starts to fail at with gcc errors Nov 13 21:29:22 heh, lovely Nov 13 21:29:32 gotat love unrelated problems blocking progress Nov 13 21:29:35 gotat, even Nov 13 21:29:37 gotta, blah Nov 13 21:29:39 need more caffeine Nov 13 21:30:56 ugh Nov 13 21:31:09 i just noticed do_rootfs doesn't rerun when building a wic image and changing IMAGE_BOOT_FILES Nov 13 21:31:19 changing the vars in WICVARS doens't change the IMAGE_CMD_wic checksum Nov 13 21:31:46 should've caught that when reviewing the original patch series, it should never have gotten in without correct vardeps Nov 13 21:35:58 * kergoth opens bug 8693 to deal with it at some point in the future Nov 13 21:35:59 Bug https://bugzilla.yoctoproject.org/show_bug.cgi?id=8693 normal, Undecided, ---, richard.purdie, NEW , do_rootfs isn't re-run when wic variables (e.g. IMAGE_BOOT_FILES) change Nov 13 21:40:07 * paulg wonders what the bot will think of bug 31337 Nov 13 21:40:09 Bug https://bugzilla.yoctoproject.org/show_bug.cgi?id=31337 was not found. Nov 13 21:49:24 paulg: the bot is smart... have a good AI. Nov 13 21:49:35 **** ENDING LOGGING AT Sat Nov 14 02:59:59 2015