**** BEGIN LOGGING AT Mon Jun 30 03:00:00 2014 Jun 30 07:10:55 good morning Jun 30 07:16:23 gm Jun 30 08:37:57 it seem libxxx.so will package into xxx-dev? how can I force it package into xxx (not -dev) Jun 30 08:38:11 it's just a symbol link to libxxx.so.0 Jun 30 08:38:26 mask for -dev package Jun 30 08:45:21 dlan: it's intended for development only Jun 30 08:45:45 dlan: linking your app to the symlink subverts the standard library versioning scheme Jun 30 09:16:32 bluelightning: the problem is we some third party prebuilt libraries ( no source) and they just hard link to *.so.. Jun 30 09:16:45 but anyway I found a solution which is using INSANE_SKIP_${PN} = "dev-so" Jun 30 09:16:58 to disable this checking Jun 30 09:27:30 dlan: the following should work: Jun 30 09:27:30 FILES_SOLIBSDEV = "" Jun 30 09:27:30 FILES_${PN} += "${libdir}/*.so" Jun 30 09:28:05 that will put the .so in the main package Jun 30 09:35:53 ndec: that's will trigger qa[dev-so] warning/error. Jun 30 09:36:44 oh. ok. sorry.. i read too fast. Jun 30 09:41:14 basta Jun 30 09:41:38 I am having a strange issue, my image fails to boot some of the services because my rootfs is broken somehow Jun 30 09:41:45 when I try to log in Jun 30 09:41:49 it says /bin/sh is missing Jun 30 09:42:02 I checked sh is not on the disk Jun 30 09:42:08 what could I ahve done to cause this Jun 30 09:42:32 I am currently messing with my image.bb Jun 30 09:43:33 the other strange fact is that it started happening on images produced by my own VM/build env... but the buildservers were fine... Jun 30 09:43:55 today I get exactly what I got last week monday, but now it is on my buildservers Jun 30 09:45:14 let me update everything again Jun 30 09:51:30 /bin/sh is a symlink to busybox Jun 30 09:52:04 or bash/other shell, but usually busybox Jun 30 09:52:15 maybe something broke with update-alternatives? Jun 30 09:52:21 check log.do_rootfs Jun 30 09:52:38 yes Jun 30 09:52:44 tthanks Jun 30 09:52:51 that could be it Jun 30 09:53:06 a week ago I added a recipy called busybox for some adnroid hacking Jun 30 09:53:14 then later realized there is such an recipy Jun 30 09:53:17 thanks man Jun 30 09:53:36 I did remove the clash but seems like it broke something in the cache Jun 30 09:54:25 hmm, that in itself shouldn't have caused such a problem Jun 30 09:54:34 basically Jun 30 09:54:41 this is what the output looked like Jun 30 09:54:44 I'd like to know how you triggered this situation if you can reproduce it Jun 30 09:56:08 yes Jun 30 09:56:12 let me find those logs Jun 30 09:56:16 since the log was strange too Jun 30 09:56:29 basically Jun 30 09:57:10 for now I would say create a layer with highest priority... add busybox_0.1.bb and manually bitbake busybox followed by bitbake fooimage Jun 30 09:57:20 then check the logs Jun 30 09:59:01 no error is preduced but the wrong files land on the device Jun 30 09:59:59 meh can t find that log... its hiding from me Jun 30 10:01:33 got it: Jun 30 10:01:37 [19:23:13][Step 4/5] recipe busybox-0.1-r0: task do_package: Started Jun 30 10:01:37 [19:23:14][Step 4/5] busybox-syslog does not appear in package list, please add it Jun 30 10:02:23 ebin.com/avufitevup.vbs Jun 30 10:03:03 that image went on to produce an artifact that was messed Jun 30 10:03:19 and then I guess the incorrect busybox entered the sstate Jun 30 10:04:27 also makes sense why I encoutered the problem on my dev machine first as I was doing the dev there and building that recipy manually while making it Jun 30 10:05:24 goting to do a manual bitbake -c cleansstate and see if that fixes things Jun 30 10:08:21 ...followed by bitbake fooimage -> then remove/rename the clashing recipy and make the image Jun 30 10:25:03 bluelightning Jun 30 10:25:04 disaster Jun 30 10:25:11 I cant get rid of this behaviour now Jun 30 10:25:22 cleansstate on busybox does not work Jun 30 10:25:32 the image is still broken Jun 30 10:26:13 gtg Jun 30 10:26:17 * pompomJuice lunch Jun 30 11:19:07 new theory... those old busybox ipk's were not removed from the feeds... lets see if that caused the jam Jun 30 11:46:29 bluelightning Jun 30 11:47:48 The ipk that was produced with the same recipy name was confusing the package index Jun 30 11:47:56 adfter deleting them things work again Jun 30 12:16:36 pompomJuice: ah right, I think I know how that might occur Jun 30 12:16:45 yea Jun 30 12:16:53 my feeds look like this Jun 30 12:17:24 -rw-r--r-- 1 teamcity teamcity 0 Jun 21 13:59 Packages Jun 30 12:17:24 lrwxrwxrwx 1 root root 20 Jun 27 10:56 armv7a -> armv7at2hf-vfp-neon/ Jun 30 12:17:24 drwxr-xr-x 2 teamcity teamcity 348160 Jun 30 14:11 armv7at2hf-vfp-neon Jun 30 12:17:24 drwxr-xr-x 2 teamcity teamcity 32768 Jun 30 14:11 impedo_duo Jun 30 12:17:24 drwxr-xr-x 2 teamcity teamcity 4096 Jun 30 14:11 all Jun 30 12:17:32 we have an impedo_duo Jun 30 12:18:10 that the incorrect busybox ipk was landing Jun 30 12:18:33 the other one was in armv7a Jun 30 12:19:29 so when you do a bitbake -c cleansstate busybox it does not nuke the old one, just the one in existance now Jun 30 12:19:57 old one -> incorrect one Jun 30 12:23:26 while do_install i get ERROR: cannot stat bb-example, no such file or directory, i guess something is missing, how can i get more infos about what is missing? Jun 30 12:29:34 Rick__, you need to go checkout the logfile Jun 30 12:30:33 if the logfile does not include the entire path of bb-example, you need to navigate to the actual build folder and look for the file there.... you install is looking for the file at the wrong place Jun 30 12:36:08 pompomJuice: thx my mistake my recipe was wrong Jun 30 12:36:20 ;) Jun 30 12:38:29 what does it mean when the build only generates dev.ipk and dbg.ipk ? Jun 30 12:44:33 Rick__: probably your do_install isn't installing the right stuff (or not to the right place perhaps) Jun 30 12:45:39 pompomJuice: ok, thanks for the info... it's a bit of a tricky problem in that we don't currently have a means to clean up after recipes that have been removed Jun 30 12:45:51 yea Jun 30 12:46:00 it makes sense why the problem occurred Jun 30 12:46:18 there's a bug open to track the issue: https://bugzilla.yoctoproject.org/show_bug.cgi?id=4102 Jun 30 12:46:46 I wonder if we can do better on the package end of things though Jun 30 12:47:59 opkg should do the work Jun 30 12:48:17 as it is failing to see the duplicate package Jun 30 12:48:51 it does not need to fix it, just say something about it... WARNING two packages with the name foo Jun 30 13:03:18 I am trying to add a postinst script to one of my recipies... but that script contains bash functions that look like "function foo {" <---- bitbake parse error Jun 30 13:03:38 bitbake does not like the "{" and escaping it with either "/" or "//" does not work... any ideas? Jun 30 13:03:51 "\" Jun 30 13:03:52 mmm Jun 30 13:08:01 pompomJuice: ensure there is a space at the start of the line... if } is not the first then it should be ok Jun 30 13:08:18 (on the closing) Jun 30 13:08:18 there's a note in the bitbake manual on this Jun 30 13:08:32 ok let me test it Jun 30 13:09:01 lol it works Jun 30 13:09:07 thanks Jun 30 13:19:03 JaMa, any more zeroc-ice failures? Jun 30 13:20:05 bluelightning: actually, the project im building does not use make install, in this case i dont need the do_install fonction right? Jun 30 13:24:00 Rick__: well you need to install something or there won't be any files to package Jun 30 13:24:07 Crofton|work: jenkins server is running test-dependencies build for last week, so nothing new Jun 30 13:24:20 Building recipe: libcroco (721/1885) Jun 30 13:24:26 it will run for 10 more days or so Jun 30 13:24:33 ok Jun 30 13:24:47 feels like we need more builders? Jun 30 13:24:53 Rick__: i.e. some files need to be copied after building (in ${B}, which is the current directory when in do_install) to ${D} Jun 30 13:25:32 Crofton|work: more people to read the logs then.. Jun 30 13:25:38 rofl Jun 30 13:25:42 fair enough Jun 30 13:35:12 bluelightning: do i have to specify something if there are source file in a subdir i need to build? Jun 30 13:37:10 Rick__: if do_compile doesn't already take care of that, yes you will need to add it Jun 30 13:40:02 bluelightning: the makefile is supposed to be doing this so i did not put any do_compile Jun 30 13:40:25 Rick__: right, that would normally be the case - but it depends entirely on what's in the Makefile Jun 30 13:40:54 bluelightning: when compiling on debian, i only use ./configure and make Jun 30 13:41:48 Rick__: so I'm not really following - is this about compile or install? since we seem to be talking about both Jun 30 13:42:48 im trying to compile a framework and build source examples provided in a subdir Jun 30 13:43:08 to use these examples on my target device Jun 30 13:44:35 bluelightning: wich is arm Jun 30 13:45:16 ok but what is actually happening vs. what you expect to happen? Jun 30 13:47:25 bluelightning: i guess im expecting a package with build files to install on my device Jun 30 13:49:04 bluelightning: For now, the framework seems to compile (at least it doesnt complain during the process) Jun 30 13:49:46 bluelightning: but nothing is built Jun 30 14:18:46 Rick__: you'll have to look in the workdir for the recipe to see what is being compiled, installed and then packaged Jun 30 14:19:02 consult the logs as well (also under the workdir, in the "temp" subdirectory) Jun 30 14:19:14 gotta go now I'm afraid Jun 30 14:19:52 ok thx for your help ill continue diggin Jun 30 14:43:06 is there a reference anywhere as to the SRC_URI 'cvs' type options? Jun 30 20:07:11 otavio, sometimes I think we are the only two people that read AB emails :) Jun 30 20:58:14 postgresql fails to build with: Jun 30 20:58:15 | configure: error: header file is required for XML support Jun 30 20:58:18 bbab **** ENDING LOGGING AT Tue Jul 01 03:00:01 2014