**** BEGIN LOGGING AT Fri Jun 21 02:59:58 2013 Jun 21 05:20:16 *yawn* Jun 21 05:20:28 Is it okay to have files in tmp/deploy/images with AUTOINC? Jun 21 05:20:34 in the filename? Jun 21 07:53:50 good morning Jun 21 08:03:12 morning all Jun 21 10:23:37 zecke: isn't AUTOINC expanded in the packages? It's strange but it looks like it works like this... Jun 21 11:18:39 ant_work: yes, it is auto expanded but not all the time Jun 21 11:18:56 * zecke thinks kay sievers is a moron. :} Jun 21 11:34:03 zecke: I sent you email about AUTOINC which will hopefully fix this Jun 21 11:37:05 RP: thanks, I will have a look. I am kicking systemd a bit more (i finally want to upgrade from edison to dylan|master) Jun 21 11:37:26 this modern C stuff.. is so inefficient. And they feel like rockstars. :} Jun 21 11:39:00 zecke: I find it ironic that the optimised shell script boot process (with no fork and minimal execs) from poky written years ago is still faster than the current systemd implementation Jun 21 11:39:09 I'm told systemd needs to be optimised Jun 21 11:39:51 RP: echo "foo" | logger.. makes systemd mmap memory for 130MB for a 648kb ring buffer. :) Jun 21 11:40:22 RP: systemd is mostly open, malloc, strdup, memcpy. it can also start other applications but that appears secondary Jun 21 11:40:44 RP: the only reason I want to use it, is to replace some respawn shell scripts. :} Jun 21 11:43:25 RP: ah, i looked at kernel.bbclass but didn't know what to search for. I will add it to my bbappend but also send patches for u-boot Jun 21 11:47:09 zecke: thanks, if this works we can fix the u-boot.inc file, patches very welcome :) Jun 21 11:47:37 RP: I will try without PKGV,PKGR right now. Would you put it into deploy.bbclass? Jun 21 11:49:25 zecke: good question. Not all deploy functions use PKGR/PKGV which is why I didn't do that originally I think Jun 21 11:49:38 zecke: I suspect you need the PKGV/PKGR change too Jun 21 11:50:09 zecke: Basically we moved from trying to change the version before we'd fetched the recipe to after we'd fetched/compiled it Jun 21 11:50:30 zecke: trying to do so before you'd fetched it caused all kinds of nasty problems :/ Jun 21 11:52:17 RP: PR/PV doesn't work. Let's see about PKGR/PKGV now. :) Jun 21 11:55:35 RP: AUTOINC is replaced with '0' but after changing the SRCREV it is still '0' Jun 21 11:57:42 zecke: well, that would be a different problem. Did you enable the PR service? Jun 21 12:03:08 RP: no, I did not (i am not aware of it). I re-enabled my original PV (which contains SRCPV) and I will see Jun 21 12:03:09 hello, can you please help me, I need to set DL_DIR and SSTATE_DIR variables to a remote ssh directory ? Jun 21 12:05:27 zecke: see local.conf.sample.extended:# Set PRSERV_HOST to 'localhost:0' to automatically Jun 21 12:05:27 # start local PRService. Jun 21 12:05:27 # Set to other values to use remote PRService. Jun 21 12:05:27 #PRSERV_HOST = "localhost:0" Jun 21 12:05:41 mks: you can't do that Jun 21 12:06:00 mks: those are the directories files are locally copied to Jun 21 12:06:21 mks: you could probably make it pull files from a remote ssh location as a mirror though Jun 21 12:06:39 RP: okay, maybe you should pick a different name as it is not only influencing the PR? Jun 21 12:06:58 RP: so the code to store git revision (and id) in a local db is gone now? Jun 21 12:07:32 So someone not using a pr server, needs to bump the PR when changing the SRCREV now? Jun 21 12:08:23 zecke: the PR service became the local db Jun 21 12:08:34 zecke: it seemed crazy to have two ways of doing the same thing Jun 21 12:08:38 RP: ok so what do you suggest me to share to files from several machine ? Jun 21 12:09:08 to share files* Jun 21 12:09:18 mks: NFS share for those directories? Jun 21 12:10:08 zecke: I guess "revision server" would be a better way to name it but its a bit late to change now Jun 21 12:10:36 RP: thanks I will try that Jun 21 12:11:21 RP: sshfs could be a good way to make it Jun 21 12:11:47 RP: As a plain user, I would appreciate a warning that I have a "reduced" service now. Maybe just forbid and enforce people to pick not have '0' or a PRserver? Jun 21 12:12:23 RP: in terms of 'late'. In WebKit I learned the philosophy that the future of the project is still ahead. Jun 21 12:22:20 zecke: its a balance, I don't the benefit in renaming the PR server would be worth it at this point Jun 21 12:23:17 zecke: this issue only affects people installing from package feeds, normal builds are now unaffected due to the way the hashes are used in the stamps Jun 21 12:25:35 RP: Naming: I would not have guessed that a PR server influences the PV. And in general I start to dislike "silent" choices, specially when they are different to the old behavior. These are my two cents. :) Jun 21 12:49:52 zecke, have you tried making a recipe for gr-osmosdr? Jun 21 13:13:20 yay! I got stickers of Yocto and Intel Open Source Technology Center from my friend Jun 21 13:14:05 nice :) Jun 21 13:14:18 There are also OE stickers around :) Jun 21 13:14:40 yes, if you come to an ELCE or FOSDEM you can probably get all of those Jun 21 13:15:07 yocto t-shirts at ELCE as well :) Jun 21 13:15:21 (well, usually... not sure what's planned for the upcoming ELCE) Jun 21 13:16:34 pants, finally Jun 21 13:17:26 Crofton|work: no, I have not Jun 21 13:29:25 bluelightning: he was in America for embedded linux conference, I guess Jun 21 13:29:37 I couldn't find a familiar name in the conference though Jun 21 13:29:53 I expected some guys from OE and Yocto community :) Jun 21 13:30:36 ah, ELCE, embedded linux conference europe Jun 21 13:31:21 * eren needs immediate quit, see you later! Jun 21 16:36:04 Hello. Jun 21 16:36:10 I have a question. Jun 21 16:36:25 hi agherzan Jun 21 16:36:25 ask Jun 21 16:37:05 if i have a library with SONAME myname.so and the myname.so is already in the rootfs, when the build system wil run ldconfig the installed file will be erased right? Jun 21 16:37:10 is this normal? Jun 21 16:37:20 agherzan: Did you see my email about the two meta-rpi patches I sent that never merged? Jun 21 16:37:21 how can i avoid this Jun 21 16:37:29 RP: nope Jun 21 16:37:31 will check Jun 21 16:38:10 will* Jun 21 16:38:25 agherzan: http://lists.openembedded.org/pipermail/openembedded-devel/2013-June/091124.html Jun 21 16:39:04 agherzan: I'm not sure why the file would get erased, that doesn't sound right Jun 21 16:39:11 RP: Why did i miss these? Jun 21 16:39:14 Strange Jun 21 16:39:37 agherzan: actually, I think it might erase them thinking they're symlinks to a versioned .so Jun 21 16:40:01 RP: this is soooo wrong Jun 21 16:40:23 agherzan: I would have thought it should check they are symlinks first... Jun 21 16:40:58 here's a think. i have 3 libraries with the same soname. and i create this soname before to make sure is one of these libraries (example graphics libraries with different backends). guess what? Jun 21 16:41:04 agherzan: I've never seen it do this, I'm just guessing about what might be happening since the x.so is usually a symlink and ldconfig can create them Jun 21 16:41:48 in my case the erased library is a real binary Jun 21 16:42:10 and it gets erased. Jun 21 16:42:26 agherzan: probably need to check into the ldconfig code and only allow it to remove symlinks Jun 21 16:42:50 RP: yeah... that would have been my guess to. Jun 21 16:42:56 RP: thanks for hint Jun 21 16:43:09 (will take a log on your patches this weekend - cross my heart) Jun 21 16:44:11 RP: here's some breaking news. Jun 21 16:44:43 any idea why would ldconfig rm these on a host and leave it on another host (same sha's - same layers - same code) Jun 21 16:46:33 agherzan: ldconfig is doing some kind of check against the native system and getting a different result. Might be interesting to strace both systems and compare them Jun 21 16:46:45 (I'm just guessing) Jun 21 17:23:08 RP: Do you have a sec (same ldconfig stuff Jun 21 17:23:12 ? Jun 21 17:53:40 WHat happends if two rpms provide the same file Jun 21 17:53:48 ? Jun 21 17:54:51 how is the req determined? Jun 21 18:09:10 agherzan depends on the contents.. Jun 21 18:09:23 if the contents are the same (md5/sha256) no error occurs.. Jun 21 18:09:46 if contents are different, then if they are ELF, and we're in a multilib configuration.. the 'best' match is used.. Jun 21 18:09:58 (currently that is usually the 64-bit version) Jun 21 18:17:30 fray: yes you're right thanks Jun 21 18:17:36 one more question Jun 21 18:17:44 when is the buildhistory exported Jun 21 18:17:48 @rootfs time? Jun 21 18:18:11 i mean it is erased and recreated everytime @do_rootfs time? Jun 21 21:34:14 hello guys Jun 21 21:34:22 what is native packege ? Jun 21 21:34:28 ex: python-native? Jun 21 21:41:38 a native package is built as a normal "native" install for the build host Jun 21 21:42:20 it would typically be installed to the native sysroot, usually x86 or x86_64 Jun 21 21:43:57 look under tmp/sysroots/ and you should see several install trees, at least one native, one machine-specific, one target-cpu specific Jun 21 21:44:36 umm Jun 21 21:44:45 'tree' is a good application for viewing directories Jun 21 21:44:49 on my x86 build host for oe-classic, i have these: armv7a-none-linux-gnueabi dm816x-evm-none-linux-gnueabi i686-linux Jun 21 21:44:57 apt-get install it if you don't have one :) Jun 21 21:45:21 so .i have a x86_64 machine that is build code for arm Jun 21 21:46:15 i have 2 sysroots in tmp Jun 21 21:46:21 x86_64 Jun 21 21:46:28 then you should have one arm-foo sysroot, one x86_64 sysroot and one for your "machine" whatever that is Jun 21 21:46:43 yup Jun 21 21:46:54 now which one is called "native" Jun 21 21:47:05 x86_64 Jun 21 21:47:10 i see Jun 21 21:47:16 i think i got it Jun 21 21:47:26 ie, whatever your build host is Jun 21 21:47:35 so native packet is one that is needed to build the target sysroot Jun 21 21:47:46 the build host is the machine that does the building .. Jun 21 21:47:52 yes Jun 21 21:48:10 and a native packet goes into sysroot for the build host Jun 21 21:48:17 for example gcc will go there Jun 21 21:48:19 so native packages are typically needed to run during the build Jun 21 21:48:28 like gcc Jun 21 21:49:19 so .. is it possible to create a recipe that has different build dependecies at build time that run time? Jun 21 21:49:31 oh yeah Jun 21 21:49:48 but that's just RDEPEND vs. DEPEND Jun 21 21:49:50 long story short .. i need a build some software qith qt .. but i don't want qt installed on the target .. but only 3 qt libs Jun 21 21:50:02 i see Jun 21 21:50:13 RDEPEND vs DEPEND in the main recipe Jun 21 21:50:19 the recipes should more or less handle that Jun 21 21:50:34 the recipes that build qt libs and applications Jun 21 21:57:23 feydrautha80: this may help: https://wiki.yoctoproject.org/wiki/Creating_a_recipe_for_a_Qt_application Jun 21 22:34:35 thank you Jun 21 22:34:45 sorry for that Jun 21 22:34:52 i got abserbed by the link Jun 21 22:34:57 i forgot to say thank you Jun 21 22:34:58 :) Jun 21 22:46:57 feydrautha80: no problem :) **** ENDING LOGGING AT Sat Jun 22 02:59:58 2013