**** BEGIN LOGGING AT Tue Nov 13 02:59:57 2012 Nov 13 03:25:03 help - trying to build the angstrom kernel for beaglebone and the build process is failing Nov 13 03:25:20 for some reason it can't git clone the meta-smartphone repo so the whole build fails. Nov 13 03:26:06 using the steps outlined @ http://www.angstrom-distribution.org/building-angstrom Nov 13 04:46:11 follow-up: I'm getting a 403:forbidden when trying to access http://git.shr-project.org/repo/meta-smartphone.git Nov 13 04:46:35 cloning from the gitorious mirror by hand works and my build is now progressing. Nov 13 04:47:02 wondering if my domain (cox.net) has been blacklisted by shr-project.org? Nov 13 07:29:50 hey, where does bitbake store stuff fetched by git? Nov 13 07:30:28 I'm getting an error fetching sources, and I think I can bypass the script altogether and place the sources there manually Nov 13 07:31:00 I know it's not ideal but I need something built, but for some unfathomable reason, the script fails to fetch my stuff Nov 13 07:36:18 gm Nov 13 07:36:32 linux_tom: downloads dir Nov 13 07:38:41 thanks Nov 13 07:46:22 how would I write a SRC_URI to fetch from a local git tree (not running a git server), git pull does allow this, but bitbake does not like the formats I've tried Nov 13 07:55:09 nevermind, found it, apparently I needed to say git:////home/.... Nov 13 07:55:12 eFfeM_work: I tried that yesterday - could not get it to work either ;( Nov 13 07:55:33 SorenHolm: works for me now :-) Nov 13 07:56:25 I used SRC_URI = "////home/whatever/.git;protocol=file" Nov 13 07:56:59 but from peeking at bitbake git.py it seems that (at least for the bitbake version I am using) that protocol=file is the default if there is no host Nov 13 08:00:07 SorenHolm: oops: used SRC_URI = "git:///home/whatever/.git;protocol=file" Nov 13 08:00:30 forgot the git part in my previous example Nov 13 08:03:31 eFfeM_work: sweet. Thanks for the hint. Realy usefull :) Nov 13 08:04:21 SorenHolm: yeah Nov 13 09:08:00 Hello everybody. I have a library which is called microxml and I am trying to port it to OpenEmbedded.. The strange thing that compilation works using my own toolchain (generated by crossdev) but when OpenEmbedded compiles it, I get this error "mxml-string.c:432:16: error: unknown type name 'va_list'".. Any help please ? Nov 13 09:12:06 vadmeste: maybe you are missing or or whoever declares va_list Nov 13 09:45:13 anyone knows why bitbake fails to fetch git urls? here's a paste of the error http://pastebin.com/k2Z9CHED Nov 13 09:53:07 morning all Nov 13 10:03:18 morning all Nov 13 10:04:37 hi pb_ Nov 13 10:04:50 hi bluelightning Nov 13 10:07:31 Hi Nov 13 10:09:51 is there a special trick to build meta-openembedded/meta-systemd ? Nov 13 10:25:20 what controls if my package gets a postinstall script or not? all my ptest packages get them but I'd like to remove it. Nov 13 10:44:32 hmm, my packages get opkg status "install user unpacked" instead of "install ok installed" Nov 13 10:47:36 How did you install them? Nov 13 10:49:34 I include them my image definition Nov 13 10:49:52 IMAGE_INSTALL = ... Nov 13 10:50:17 And are their postinsts runnable offline? Nov 13 10:50:34 well they don't have any... Nov 13 10:50:55 What does your log.do_rootfs look like? Nov 13 10:52:29 what should I look for? Nov 13 10:56:04 actually, they're not in IMAGE_INSTALL. they are added by the ptest-pkgs image feature Nov 13 10:57:53 lunch. back in 1h. Nov 13 11:19:03 Zagor: usually that indicates postinsts that weren't yet able to be run Nov 13 11:19:28 hmm Nov 13 11:19:51 or, they may have been installed after the piece of code that goes through and marks packages as installed Nov 13 11:45:13 bluelightning: something seems a bit off, though. there are no .postinst scripts in the rootfs for run-postinsts.awk to run. Nov 13 11:45:23 (for these packages) Nov 13 11:45:41 it must be the latter then.. when are you installing these additional packages? Nov 13 11:46:29 they are added by IMAGE_INSTALL_COMPLEMENTARY in image.bbclass Nov 13 11:46:52 so, rootfs_install_complementary then Nov 13 11:47:18 hmm, then they should be marked as installed Nov 13 11:47:28 nothing in the log.do_rootfs to indicate any issue? Nov 13 11:48:58 nothing that leaps out at me. but there's 52k lines and I'm not sure what to look for. Nov 13 11:51:35 hmm, here's something. the -ptest packages are installed after all the other packages are configured Nov 13 11:53:09 POPULATE_SDK_POST_TARGET_COMMAND += "rootfs_install_complementary populate_sdk; " Nov 13 12:13:33 the order is definitely wrong. here's what happens to dropbear: http://pastebin.com/3NpBmD2Z Nov 13 12:14:14 I don't quite see in rootfs_ipk_do_rootfs() why this happens though. how come the "Configuring" step runs before rootfs_install_complementary()? Nov 13 12:26:24 hey. is there a specific way to override functions in a bbclass? Nov 13 12:26:48 like update-rc.d - i want to substitute it with functions doing nothing Nov 13 12:27:07 and i was wondering if you already had a process for doing such things Nov 13 12:49:56 while developing a recipe, is there a way to execute do_install by hand? I'm currently at the installstep, no need to recompile Nov 13 13:05:24 devzero_, bitbake -c install Nov 13 13:05:45 i think... Nov 13 13:15:17 CMoH: thx, -c install is valid, but it has already run.. is there a way to clean the install step? Nov 13 13:16:06 i don't know that; we use a srctree.class customized from oe-classic for minimal rebuilds Nov 13 13:18:20 use -f to force it to run the step even though it doesn't think it' needs to Nov 13 13:22:09 I'll try, thx Nov 13 13:42:46 CMoH: I can't remember if we discussed this already but have you looked at externalsrc.bbclass ? Nov 13 13:43:39 bluelightning, yes we did, but i think at that time there was no solution on oe-core yet Nov 13 13:46:29 so i used what was available at the time Nov 13 13:48:53 i think i also submitted my srctree.bbclass to oe-core, but well... Nov 13 13:52:38 CMoH: externalsrc.bbclass should cover the same use cases as srctree, if it does not then we should fix that... Nov 13 13:54:30 heh - when other pressing matters with our oe-based project get out of the way... Nov 13 13:54:33 :) Nov 13 13:58:08 sure, of course :) Nov 13 14:29:24 i was wondering about something: most daemon-related recipes expose initscripts and afferent functions (update-rc.d etc.); now with meta-systemd it feels like an orthogonal approach would be to strip oe-core of initscripts and add a meta-sysvinit layer Nov 13 14:29:41 or perhaps other layers as well for upstart, openrc or what else Nov 13 14:29:53 is there any work being done in such a direction? Nov 13 14:30:16 CMoH: I think there is work planned in this direction for the 1.4 release Nov 13 14:30:53 ty jackmitchell; however, as an outsider i have no clue of the oe release plan :) Nov 13 14:30:56 CMoH: I would expect (another) large discussion to happen on the mailing list before any work is started though Nov 13 14:31:28 CMoH: indeed, so am I, the release plan is all laid out somewhere, let me have a quick look Nov 13 14:32:21 the thing is i've forked away from angstrom because it is now too deeply tied to systemd, and i've started a new distro with openrc Nov 13 14:32:47 CMoH: https://wiki.yoctoproject.org/wiki/Yocto_1.4_Features Nov 13 14:35:02 and much of the work is to overwrite stuff being done by the built-in sysvinit-related code Nov 13 14:35:08 like update-rc.d inheritances and such Nov 13 14:37:53 anyway... glad to hear the same thoughts are in yocto Nov 13 14:41:49 well... i hope the world of linux won't revolve around systemd-based systems only... though reading these threads got me a bit worried Nov 13 14:43:43 hmm... a generic question: if our organization were to buy professional support for consulting with our OE setup, to what organization should we address? Nov 13 14:47:21 depends; there are many organisations and smaller consultancies in the OE/Yocto Project community that offer such services Nov 13 14:50:35 so... who would be my go-to guy? :D Nov 13 14:51:53 i guess a search on google could reveal some hints... Nov 13 14:53:07 CMoH: sure, or you could post on the mailing list asking for interested parties to reply in private perhaps Nov 13 15:04:37 thanks... perhaps in the future if we hit a problem similar to this systemd upgrade mess-up... Nov 13 15:07:15 i mean the mess-up that happened with the upgrade for us Nov 13 15:15:57 good morning Nov 13 16:04:08 oh son of a bitch. i have a distro specific interfaces file in my distro layer. and poky and other layer shave machine specific interfaces files, as you'd expect. but due to my layer's priority being higher than poky, the distro config file is used in preference to the machine config files :| Nov 13 16:04:34 * kergoth sighs and grumbles Nov 13 16:05:04 guess i could drop my layer's priority, but then i wouldnt' be able to override things in other layers bbappends if i need to.. Nov 13 16:06:34 the question is why do you have a distro-specific interfaces file? Nov 13 16:07:59 this would happen for any file, not just this one. distro should never override machine Nov 13 16:08:03 machine is always more specific information Nov 13 16:08:43 i wonder what the impact would be of flipping filespath construction, so it's assembled override, then path, rather than path, then override Nov 13 16:09:04 that'd let more specific overrides be used in preference to less specific overrides in higher priority layers Nov 13 16:09:13 * kergoth ponders Nov 13 16:10:05 * kergoth wanders off Nov 13 16:42:47 Crofton|work: looks like something is moving... https://bugzilla.yoctoproject.org/show_bug.cgi?id=2308 Nov 13 16:57:24 How do I make a package go away in a new version Nov 13 17:00:28 Crofton|work: what do you mean? Nov 13 17:01:34 well, I split the uhd recipe to make several packages Nov 13 17:01:48 one of those packages no longer exisst Nov 13 17:02:12 so when I upgrade, I need the old package to be removed and not replaced Nov 13 17:02:22 JaMa: ^ Nov 13 17:02:37 I can't remember where we got to with that last time we needed it Nov 13 17:02:48 it is not the end of the world Nov 13 17:02:54 I am just trying to get smarter Nov 13 17:03:09 basically, we had some test utils we used to need, by the set is now gone Nov 13 17:03:14 I think it's supposed to be just a matter of setting RCONFLICTS but then you may have to be specific about versions Nov 13 17:03:47 I may use the appraoch, wait until someone notices and tell them to remove the old ones :) Nov 13 17:30:24 i think you probably want RREPLACES, either on its own, or with RCONFLICTS Nov 13 17:30:32 * kergoth never remembers the particulars of those Nov 13 17:30:43 best to double check the debian packaging/policy manual Nov 13 17:30:46 heh Nov 13 17:32:41 actually I think this situation may not be the broken one - I think it's when you split out a new package that's problematic Nov 13 21:59:16 Crofton|work: some bugger did one of my segments uphill at an average of 26MPH, I need to get some leg muscles :-D Nov 13 21:59:45 rofl Nov 13 21:59:57 I only managed 16MPH peak average 10MPH Nov 13 22:00:11 i need a new bike... Nov 13 22:00:17 yeah, he is also on a better bike I bet Nov 13 22:00:28 from the pictures yeah 2-3K bike Nov 13 22:00:31 mine 200 GBP Nov 13 22:02:37 Crofton|work: I shall have to chase you round CA next time Im there :-D Nov 13 22:02:44 heh Nov 13 22:02:49 I have a fancy bike there :) Nov 13 22:02:58 Crofton|work: road or mountain? Nov 13 22:03:01 road Nov 13 22:03:11 to far to ride to get off road Nov 13 22:03:13 Crofton|work: I have fancy mountain bike here :-D Nov 13 22:03:34 I have an old one here, but my wife says I should get a better one so I do not hurt myself Nov 13 22:03:40 hehe Nov 13 22:03:44 her logic is poor Nov 13 22:03:51 I will just go faster and further Nov 13 22:03:53 sounds good logic to me Nov 13 22:03:59 maybe she likes her peace Nov 13 22:04:07 heh Nov 13 22:04:15 and electricty bill goes down when your out the house ;-) Nov 13 22:04:57 hopefully the weather for ELCE next year will permit riding Nov 13 22:05:11 its Scotland, we go wind/rain or snow Nov 13 22:05:22 masochists Nov 13 22:05:28 or more likely all threee + hail on same day :-D Nov 13 22:05:54 if we can find a spare car driver we can cheat and they drive us to top of mountain :-D Nov 13 22:06:04 then you only need to do the downhills Nov 13 22:06:41 I did a downhill race, but we "rode" up to the start Nov 13 22:06:46 some people drove Nov 13 22:07:06 I really need a bike with disk brakes Nov 13 22:07:06 the one we normally go to is singletrack Nov 13 22:07:16 yes, this one was Nov 13 22:07:24 ah you were time trialing? Nov 13 22:07:31 yeah Nov 13 22:07:37 I was not serious Nov 13 22:07:43 just tried to get down alive Nov 13 22:07:50 nor me, need much more practice Nov 13 22:08:01 its been a few years since I last downhilled seriously Nov 13 22:08:09 like in the range of 20-25 Nov 13 22:08:12 heh Nov 13 22:08:17 ok, must go shopping Nov 13 22:08:24 need new pads for the brakes and dinner Nov 13 22:08:25 bbl Nov 13 22:08:27 :D Nov 13 22:08:36 * XorA|gone will be asleep later have fun Nov 13 22:42:36 Hello everbody. I hope someone can explain me this error "ERROR: QA Issue: ubus-doc rdepends on libubox-dev", I added RDEPENDS="libubox-dev" in my ubux_1.0.bb file and PROVIDES = "libubox-dev" FILES_${PN} = "${libdir}" FILES_${PN}-dev = "${includedir}" in my libubox_1.0.bb.... Am I missing something ? Nov 13 22:45:15 I get the QA Issue when I run bitbake ubux **** ENDING LOGGING AT Wed Nov 14 02:59:59 2012