**** BEGIN LOGGING AT Mon Feb 13 02:59:58 2012 Feb 13 03:14:08 Thex1le, most folk are still on a bit of sunday here:D Feb 13 03:14:44 ka6sox: haha yeah i figured Feb 13 08:20:01 good morning Feb 13 08:20:43 morning Feb 13 09:13:15 is the oestats-client class gone in oe-core ? Feb 13 09:22:24 hi all Feb 13 09:22:43 pwgen: if it's not in classes/ then yes... Feb 13 09:24:35 hi Feb 13 09:29:57 : t copied an oestatsbbclass file from an old openembedded checkout to oe-core and it worked, but why is it removed Feb 13 09:30:22 pwgen: it's not really a case of it being removed Feb 13 09:30:39 pwgen: oe-core was based on poky, a cleaned up but earlier fork of OE Feb 13 09:31:09 pwgen: if we should have oestats-client back in oe-core then by all means send a patch... Feb 13 09:31:18 (I don't quite know what it does myself) Feb 13 09:31:40 hi JaMa Feb 13 09:32:23 . i am thinking on extending oestats to so something that let you build/ images with an webfrontend ( django based ) Feb 13 09:32:26 when that class was used for building on a large scale (regression) it clogged up the system something terrible. Feb 13 10:20:27 morning all Feb 13 10:20:36 hi pb_ Feb 13 10:52:29 bingo! broken UPS and broken Power supply at the same time, plus RAID working with only 1 disk :-( Feb 13 10:52:36 I hate mondays Feb 13 10:53:03 ouch :( Feb 13 11:55:32 hello Feb 13 11:55:46 I m configuring a u-boot Feb 13 11:56:15 I want to configure "smsc lan 9220" ethernet card... Feb 13 11:56:29 hw can I configure.... Feb 13 12:32:25 is there a way to forcefully rebuild a package in bitbake, without deleting the downloaded files? Feb 13 12:32:56 mertsas: bitbake -c clean Feb 13 12:33:20 or -c cleansstate Feb 13 12:34:44 kenws: does any of those not require a do_clean () function in the .bb files? I know, I'm lazy Feb 13 12:35:51 mertsas: so, you're only interested in re-packaging without rebuilding the source? Feb 13 12:36:51 hi, all! Feb 13 12:36:54 I'd check what `bitbake -c listtasks ` gives Feb 13 12:37:29 kenws: thanks, I'll look at it Feb 13 12:37:47 if package license is Public Domain and no license text in source, how should I setup LIC* fields? Feb 13 12:46:25 mertsas: do_clean is defined for all recipes Feb 13 12:46:28 FYI Feb 13 12:53:08 I found that I should set LICENSE = "PD", but what value for LIC_FILES_CHKSUM? Feb 13 12:56:29 We have our own glib which we wish to install, this have a lower version number than the one in meta-openembedded. Any way to say which version of glib to use. Shouldn't PREFERRED_VERSION_glib work? Feb 13 12:57:19 right now it picks the glib with largest version number Feb 13 12:57:27 What is the name of your .bb file for glib? Feb 13 12:58:56 glib-2.0_2.27.1.bb is our, while the one it picks is glib-2.0_2.31.1.bb Feb 13 12:59:07 In that case, no, PREFERRED_VERSION_glib will not work. Feb 13 12:59:40 ok, anything that will work then? PREFERRED_VERSION_glib-2.0 maybe? Feb 13 12:59:56 right, that might be better Feb 13 13:00:08 HI all Feb 13 13:00:08 it needs to match whatever PN is set to, which is almost certainly glib-2.0 in your case Feb 13 13:00:21 I ahve question regarding multiple providers Feb 13 13:00:27 is PN set automaticaly? Feb 13 13:00:30 I am working on openembedded Feb 13 13:00:55 yes, by bitbake.conf Feb 13 13:00:58 I have my own collections on top of it Feb 13 13:01:28 see the section that startes "Package default variables" Feb 13 13:02:04 multipel bb files are providing same thing does BBFILES_priority has any effect on selection of recipes or I have to set PREFERRED_PROVIDER for the package that they are providing Feb 13 13:02:40 the latter Feb 13 13:04:01 pb_even I have set DEFAULT_PREFERENCE = "-1" in recipes who has lower BBFILES_PRIORITY? Feb 13 13:04:07 pb_: so when the PN is glib-2.0, and we do PREFERRED_VERSION_glib-2.0 = "2.27.1" it will fetch glib-2.0_2.27.1.bb if that file exists? or do we have to set something in that file as well? Feb 13 13:04:31 yeah, that should be what you need Feb 13 13:05:12 ok, well try that then Feb 13 13:06:17 pb_ I was under the impression that there is not need if I have placed the recipe at higher priority and added DEFAULT_PREFERENCE = "-1" in lower Priority recipe file Feb 13 13:06:47 another question, what does BBFILE_PRIORITY in layer.conf do? As we figured, it should use files from the meta with the highest priority if you have glib-2.0 in two different layers? Feb 13 13:07:40 why PREFERRED_PROVIDER is necessary if we are setting DEFAULT_PREFERENCE = "-1" of lower priority recipe file Feb 13 13:09:50 mertsas: BBFILE_PRIORITY will affect the case where you have two different recipes with exactly the same name in different layers. Feb 13 13:10:24 pb_: and it will use the one with the highest priority? Feb 13 13:10:29 yes Feb 13 13:11:06 I'm not entirely sure what the semantics are right now if the versions are different across layers. In the original implementation of bb collections it would always use the one from the highest priority layer even if that was an older version. I'm not sure if that is still true today with collections. Feb 13 13:11:41 ok, doesn't look like it like this still, unfortunatelly Feb 13 13:12:39 btw, the toolchain that is built, where does it reside? we would like to take that toolchain and use it in other projects, with all headers and libs that is built by bitbake. Is this possible? Feb 13 13:12:57 sorry for all the noob questions, quite new to embedded systems Feb 13 13:31:15 does DEFAULT_PREFERENCE apply acrros different recipes or it only applies on recipes of the same version Feb 13 13:32:25 like if we have two different recipes providing the same thing ..... and one recipe has DEFAULT_PREFERENCE = -1 then do we have to set preferred_provider ? although we have set the DEFAULT_PREFERENCE to -1 for one recipe Feb 13 14:07:40 otavio: I looked at combo-layer btw; it seems to me that the config file won't be updated if the component revision hasn't changed... ? Feb 13 14:36:51 Hi, How can I have a default password for any user? I hav'nt find the full passwd file the recipes but only parts. Feb 13 14:57:09 DavidP_: I do not know if this is the "preferred way" of doing it or if OE provides anything special for this case, I simply added an own recipe that adds a user in its postinst script Feb 13 14:57:19 user + pw Feb 13 14:59:20 ok Jin^eLD, the only clue I've find is the Dropbear-backdoor... but it's the " dirty way " :) Feb 13 15:00:52 DavidP_: as I said, I do not know if this is the "way to do it", it just works for me so I did it that way: https://gitorious.digitalstrom.org/dss-oe/dss-oe/blobs/master/dS/meta-dss11-testing/recipes-core/dss11-users/dss11-users_1.0.bb Feb 13 15:01:56 thank's for the sources :) Feb 13 15:02:05 np Feb 13 15:05:47 Jin^eLD: hi, btw are you still working on mediatomb? Feb 13 15:06:05 bluelightning: well, "theoretically" :) we did not do any development for the last 2-3 years now Feb 13 15:06:07 simply had no time Feb 13 15:06:11 but yes, I am the one behind it Feb 13 15:06:51 Jin^eLD: ah ok, fair enough Feb 13 15:07:05 Jin^eLD: I was asking mainly as I submitted a patch I created in the yocto project context to the sourceforge patch tracker but AFAICT it has not been looked at... Feb 13 15:07:49 bluelightning: yeah that's what I meant by "we did not do any development for the last 2-3 years" ;( we still want to get ourselves together and go through all patches Feb 13 15:07:55 but no clue when we manage to do it Feb 13 15:07:58 what was your patch about? Feb 13 15:08:48 Jin^eLD: very trivial, basically don't warn about the youtube plugin being enabled if it's not actually enabled :) Feb 13 15:08:55 :) Feb 13 15:31:14 bluelightning: it does Feb 13 15:31:29 bluelightning: usually it changes the order on the output Feb 13 15:32:34 otavio: hmm, that's not good... hopefully I can reproduce that Feb 13 15:36:09 otavio: hmm, no luck... Feb 13 15:38:15 otavio: any clues on how I could make it happen here? Feb 13 15:54:31 bluelightning: i can give you mine .conf and you take a look if you spot something Feb 13 15:54:44 otavio: sure Feb 13 15:55:05 bluelightning: http://paste.debian.net/156115/ Feb 13 16:12:12 otavio: ok, I created a config with that content within the combo repo dir, ran an init, then several updates and I didn't get a changed config Feb 13 16:15:44 bluelightning: any idea when RP will be around to push that PR bump patchset? Feb 13 16:15:55 bluelightning: I've noticed that koen already pushed it to meta-oe.. Feb 13 16:16:23 JaMa: he's flying to the US today; hopefully he will check in later but I'm not sure Feb 13 16:16:34 ok, thanks for info Feb 13 16:20:27 evening Feb 13 16:20:37 (morning in UGT) :) Feb 13 16:20:55 what combination of MACHINE/DISTRO/IMAGE is stable for current oe-core? Feb 13 16:21:13 (if any DISTRO exists there) Feb 13 16:21:36 Jay7: qemu* without distro Feb 13 16:21:51 JaMa: but which image? Feb 13 16:22:44 Jay7: what kind of image do you want...? Feb 13 16:22:59 I want test PARALLEL_MAKE :) Feb 13 16:23:27 so.. I would prefer biggest possible image Feb 13 16:23:44 but oe-core-only Feb 13 16:23:58 bluelightning: humm confusing ... Feb 13 16:24:17 bluelightning: drop the java one, this one i added today Feb 13 16:24:17 Jay7: probably core-image-sato in that case... Feb 13 16:24:30 bluelightning: i'll try to reproduce it here Feb 13 16:24:48 bluelightning: thanks Feb 13 16:27:05 bluelightning: http://paste.debian.net/156122/ just reproduced it here Feb 13 16:29:24 otavio: ok, I guess this is python's configparser doing this... I'm not sure how to fix it atm :/ Feb 13 16:29:57 well.. started first build with empty sources Feb 13 16:30:07 kind of cache filling :) Feb 13 16:41:04 bluelightning: simple; reload and compare the hashes, if all match, doesn't save the file Feb 13 16:41:22 otavio: yeah I think that will be the easiest workaround Feb 13 16:46:45 Does anyone know where the OE-core qemuarm kernel config comes from? Feb 13 16:47:04 It looks like the kernel source gets pulled from a certain branch Feb 13 16:47:44 the meta/recipes-kernel/linux/linux-yocto_3.2.bb recipe for example uses KBRANCH_qemuarm="standard/default/arm-versatile-926ejs" Feb 13 16:49:56 otavio: this should help: http://git.yoctoproject.org/cgit/cgit.cgi/poky-contrib/commit/?h=paule/fixes12&id=6c85a13b334cd974f19923a0918344b625f7c9df Feb 13 16:50:49 otavio: ah, heh... maybe not Feb 13 16:51:34 heh Feb 13 16:51:43 bluelightning: want me to test this one or not? Feb 13 16:52:00 otavio: no, I fixed it in the wrong place... hang on a sec Feb 13 16:54:59 otavio: try this: http://git.yoctoproject.org/cgit/cgit.cgi/poky-contrib/commit/?h=paule/fixes12&id=95683b50507d96eb6b9a649ab7892bc615b1db71 Feb 13 16:57:12 Hm, the kernel.bbclass:kernel_do_configure checks for a defconfig, copies it to .config if it doesn't exist and runs oldconfig. But where does the config come from? Feb 13 16:57:59 kenws: the config is produced from config fragments Feb 13 16:58:35 kenws: http://www.yoctoproject.org/docs/latest/kernel-manual/kernel-manual.html#kernel-configuration Feb 13 16:59:06 Ah, some docs - thanks! Feb 13 16:59:22 I was reading http://www.openembedded.org/wiki/Kernel_Building but it not for oe-core (or outdated) Feb 13 16:59:39 s/it/it's/ Feb 13 17:57:39 bluelightning: Do you know where these fragments come from? The arm-versatile-926ejs.scc for example (the one I suspect to be responsible for having a kernel for the versatile board in case of MACHINE=qemuarm) seems not to be part of the oe-core git tree nor part of the arm-versatile-926ejs branch of the yocto kernel tree Feb 13 17:58:02 maybe they are generated, but I havn't found any code that indicates that Feb 13 17:58:09 hmmm Feb 13 17:58:20 .scc doesn't seem like device tree or something like that Feb 13 18:00:18 GNUtoo: no, they seem to pull in cfg files Feb 13 18:02:50 what's inside theses cfg files? Feb 13 18:04:03 kenws: I'm no expert on this, you'd probably want to ask on the mailing list for clarification Feb 13 18:04:16 kenws: though bear in mind a lot of people are at ELC atm Feb 13 18:05:08 GNUtoo: cfg files are just lists of kernel config options as they might appear in a defconfig, except they are meant to be a small set related to enabling some piece of functionality Feb 13 18:05:15 * Jay7 want bbtop - tool to view currently running bitbake tasks :) Feb 13 18:05:32 like top/atop/pftop/etc Feb 13 18:06:22 ok Feb 13 18:07:07 Jay7: richard has a prototype patch for the knotty bitbake UI that shows a list of running tasks at the bottom of the screen at all times via curses Feb 13 18:07:10 heh Feb 13 18:07:49 kergoth: btw, do some GUI have such view? Feb 13 18:08:26 Jay7: goggle shows running tasks. pretty sure hob does too Feb 13 18:09:02 hm.. may be I should consider to install x11 libs into my build container.. Feb 13 18:11:40 kenws: they are part of yocto-kernel tooling Feb 13 18:12:25 kenws: yocto-kernel build does a bit behind the scenes to setup .config by default Feb 13 18:12:52 hm, someone should make an rpm-lite recipe that doesn't depend on openssl and all that garbage. Feb 13 18:12:58 * pb_ waits patiently for his build to start Feb 13 18:18:14 otavio: did that change work for you? Feb 13 18:20:56 khem: btw, thanks for joining linaro connect :-) Feb 13 18:27:16 rsalveti: I was happy to squeeze in few days Feb 13 21:10:07 how would I run something like ldd on an application I built in OE? Feb 13 21:10:16 (license compliance time ...) Feb 13 21:10:49 run readelf -d Feb 13 21:15:03 cbrake: license compliance ? so you want to chck the licenses of files it will want at runtime ? Feb 13 21:19:56 khem: yes, need to check linking Feb 13 21:21:44 so readelf -d will give a list of libraries that directly linked to Feb 13 21:23:01 thinking if I need to know about libraries that libraries link to Feb 13 21:24:06 it seems ldd gives a much longer list that readelf -d Feb 13 21:24:56 sounds like a good application for a script on the target system Feb 13 21:25:45 yeah ldd will show indirect deps too Feb 13 21:26:06 but if you have all packages covered with readelf -d then you have all covered Feb 13 21:28:45 khem: I guess I probably don't need to worry about secondary libs, because if they were GPL, the primary lib would need to be GPL as well Feb 13 21:29:18 as long as I trusted that all primary libs were license compliant Feb 13 21:29:28 yes Feb 13 21:29:40 plus since you checked readelf -d of whole root file system anyway Feb 13 21:30:18 khem: so run a readelf -d on every binary in the rootfs? Feb 13 21:41:00 ack...bitbake git fetcher is apparently downloading the entire rev history of everything... does it not work to just grab a specific revision? About to hit 500 MB for omap-linux-psp :( Feb 13 21:41:15 err, linux-omap-psp Feb 13 21:44:04 recipe in question: http://cgit.openembedded.org/openembedded/tree/recipes/linux/linux-omap-psp_2.6.32.bb Feb 13 21:46:23 hbeck: it does a clone first Feb 13 21:46:32 * cbrake finds the license of sqlite amusing: http://pastebin.com/hMTsc5GY Feb 13 21:46:41 khem: seems like overkill? Feb 13 21:46:50 (perhaps inspiring is a better description) :-) Feb 13 21:50:08 hbeck: so how do you suggest it should it with git Feb 13 21:50:40 khem: I don't know git all that well so I don't have an idea :( Feb 13 21:51:49 hbeck: ok, then unfortunately you have to suck it up :) Feb 13 21:51:53 haha Feb 13 21:51:57 cbrake: thats interesting Feb 13 21:52:32 cbrake: I doubt those 3 blessings are true :) Feb 13 21:53:07 khem: while I'm sucking it up, any way to know how much in total it's going to grab? Feb 13 21:53:16 khem: :-) Feb 13 21:56:21 hbeck: my lnx git tree with master checked out is 1.8G Feb 13 21:57:21 khem: bleh. My suggestion would be for recipe writers to find a way to grab snapshot tarball if they're working from a specific SRCREV ;) Feb 13 21:58:54 hbeck: angstrom does provide snapshot tars Feb 13 22:47:12 any ideas how to convert a link like http://arago-project.org/git/projects/?p=linux-omap3.git;a=snapshot;h=5fc29e7b2a76a64a739f857858ef0b98294aa155;sf=tgz to SRC_UI usable? Feb 13 23:20:39 in theory, you should be able to use it as is, if not it's arguably a bug in the fetcher. any parameters which aren't handled by the fetcher itself, iirc, should be reassembled with the url and passed down as is Feb 13 23:22:09 I'll give it a shot kergoth Feb 13 23:22:18 good luck :) Feb 13 23:22:38 fetch2 improved matters somewhat, but our fetching is still a mess imo :\ Feb 13 23:22:52 kergoth: any idea if I interrupt the git fetcher if it will continue from how much it has downloaded thus far? Feb 13 23:23:05 I'm past 1GB now and really don't want to redo that... Feb 13 23:23:15 doubtful. it just runs git, and I'm pretty sure git can't continue a partial clone Feb 13 23:23:26 ouch. Ok then Feb 13 23:23:36 I'll wait, and THEN try the tarball. Feb 13 23:24:08 another option that might be worth looking into in the longer term would be to have the fetcher support shallow clones, specify the depth in the URI, or something. heh Feb 13 23:24:41 * kergoth wanders off for a while Feb 13 23:43:44 kergoth: fetcher does not handle that url. I suspect it has to do with the ?= portion... **** ENDING LOGGING AT Tue Feb 14 02:59:56 2012