**** BEGIN LOGGING AT Thu Aug 02 02:59:59 2012 Aug 02 07:48:03 morning. Aug 02 08:57:28 hi Aug 02 08:57:56 i have a question regarding replacing/removing certain values of a variable in a .bbappend Aug 02 08:58:38 for example the base recipe sets a variable as follows VARIABLE += "value_a value_b value_c" Aug 02 08:58:58 how can i remove one of the values in a .bbappend? Aug 02 09:08:08 MaxPowers: it's a bit tricky to do that Aug 02 09:08:36 MaxPowers: you can use := in combination with oe_filter_out, that's a bit nasty though Aug 02 09:09:10 MaxPowers: it may indicate the recipe above needs to be more flexible... are you able to give more context? Aug 02 09:24:03 did no-one fix the oe-members list yet :-( Aug 02 09:24:15 XorA: I was unaware that it was broken.... Aug 02 09:24:33 bluelightning: I am still listed as a admin for that list years after I left Aug 02 09:24:58 bluelightning: I figure if I moan about it once every couple of months someone will eventually fix it Aug 02 09:25:46 looking at the emails I get I suspect we can no longer function as a eV Aug 02 09:26:47 XorA: probably time to raise the issue with the board directly... Aug 02 09:27:04 ka6sox: ^^^ Aug 02 09:28:01 I have no access to that one.. Aug 02 09:28:02 :( Aug 02 09:28:10 ka6sox: ah ok, do you know who does? Aug 02 09:28:28 not really anymore... Aug 02 09:29:03 hmm :( Aug 02 09:29:59 we can ask the current board members who does Aug 02 09:30:35 * XorA has annoyed the board but like normal that does not acheive anything in OE Aug 02 09:30:59 XorA: sigh... Aug 02 09:31:40 I guess if we have lost direct control over infrastructure that may be something for the TSC to resolve Aug 02 09:31:54 I'll raise it in that forum as well anyway Aug 02 09:32:07 XorA: sorry for the hassles Aug 02 09:32:34 bluelightning: its not a hassle, its 2-3 emails a day, but it does mean no-one is maintaining that list Aug 02 09:32:52 right... we definitely need to get onto it Aug 02 09:32:54 bluelightning: and I cant as my password was removed Aug 02 09:33:09 * XorA would otherwise have removed him from the admin list Aug 02 09:36:59 XorA, what list is this? Aug 02 09:38:04 oh, I see Aug 02 09:40:02 XorA: maybe I'm misunderstunding, which kind of mails are you receiving? Aug 02 09:40:30 ant_work: the moderator emails Aug 02 09:40:53 ah, I'm subscribed to the list and I don't see any traffic since months Aug 02 09:41:29 so you're probably getting spam Aug 02 09:42:01 what has that to do with E.V. impeachments? Aug 02 09:43:12 ant_work: I know some people have fallen off the list due to email bounces, this means votes are not valid, if no-one can admin the list we cannot have new members etc etc Aug 02 09:43:41 ouch Aug 02 09:47:12 XorA: I have sent out an email to tsc/florian/philip, hopefully we can resolve this swiftly Aug 02 10:07:43 Regarding boost.... it seems like it is compiled using the build-hosts gcc-version. The recipe resenbles the OE-classico one. And I'm not that good friends with BJAM. I get QA-error telling me that the architecture is wrong. Aug 02 10:19:10 morning all Aug 02 10:22:06 hi flo_lap Aug 02 10:23:14 bluelightning: I exchanged emails with Philip a couple of months ago about the oe-members situation. I never heard back from him so I assumed it was all ok, but if that is not the case then I am happy to sort it out. Aug 02 10:23:21 I'll forward you a copy of the last mail I sent to him. Aug 02 10:25:19 pb_: do you have access? Aug 02 10:25:24 (morning btw) Aug 02 10:25:25 yes Aug 02 10:25:36 I've forwarded you that mail now. Aug 02 10:27:01 no-one has performed moderator actions on that list since Feb 26th Aug 02 10:27:28 :-D Aug 02 10:27:56 that is a little bit sad. Aug 02 10:45:20 XorA: iirc whole lists.openembedded.org is at linuxtogo which is more or less unmaintained due lack of time Aug 02 10:49:50 RP__: ping? Aug 02 10:50:25 pb_: pong Aug 02 10:53:27 you can change it to whatever you want from the admin page Aug 02 10:58:26 how does one handle different MACHINES in oe-core ? as far as i understand , oe-core only builds quemu images ?! Aug 02 10:58:57 rob_w: you add a BSP layer supporting the machine you need Aug 02 10:59:08 hmm ok Aug 02 10:59:23 rob_w: there are a number listed here: http://www.openembedded.org/wiki/LayerIndex Aug 02 10:59:31 thx Aug 02 11:04:55 pretty cool Aug 02 11:05:25 uhhh ;-) WARNING: Host distribution "Linux Mint 13 Maya" has not been validated Aug 02 11:05:51 rob_w: that's correct... if it works for you, great, but if it breaks, no guarantees... Aug 02 11:06:01 no worrys .. Aug 02 11:06:18 we'll try to help to the best of our abilities though of course Aug 02 11:06:57 even "mainstream" distros introduce nasty problems that cause us pain :/ Aug 02 11:07:05 ok, tell me this .. how far is poky now away from oe-core ? Aug 02 11:07:11 rob_w: closer than ever Aug 02 11:08:06 rob_w: changes go directly into OE-Core and then get pulled back into Poky from there (in a scripted manner) Aug 02 11:08:37 same applies to bitbake Aug 02 11:09:05 is there a way to recreate sstate-cache from tmpdir? Aug 02 11:09:16 so i stick with oe-core and i am good Aug 02 11:09:54 rob_w: sure, you can use whichever you feel comfortable with, ultimately it's the same metadata... Aug 02 11:10:17 rob_w: poky just provides an out-of-the box tested pairing of bitbake and OE-Core, and a distro config Aug 02 11:10:45 hrw: not really, no, it's usually the other way around... Aug 02 11:10:47 gotya Aug 02 11:11:07 bluelightning: I know but by mistake I removed share sstate cache Aug 02 11:11:22 bluelightning: whole build would take another night Aug 02 11:11:45 hrw: that shouldn't cause any actual failure though, we don't rely upon sstate-cache being there... Aug 02 11:12:00 bluelightning: sure, but incremental builds will suffer ;D) Aug 02 11:12:02 how is the order of parsing of those layers ? like last parsed layer will overwrite em all ? Aug 02 11:12:22 order in bblayer.conf i mean Aug 02 11:12:41 rob_w: depends; each layer has a layer priority which controls which layer's recipes take precedence Aug 02 11:13:03 rob_w: for classes and conf files it's more or less the order in which the layers appear in bblayers.conf Aug 02 11:13:27 say for normal recipes of apps n stuff Aug 02 11:14:04 rob_w: right, in which case it's layer priority order; if priorities are the same between two layers then it's the order found in bblayers.conf Aug 02 11:14:19 ah Aug 02 11:14:23 rob_w: you can get some introspection into a layer config by using the bitbake-layers utility FYI Aug 02 11:14:37 just noticed fe. meta-igep/conf/layer.conf Aug 02 11:15:23 bitbake-layers can tell you the layers enabled and their priorities, which recipes & appends exist, which version of a recipe is preferred, etc. Aug 02 11:18:28 hrw: unfortunately I think you may just have to bear the extra time, the system isn't designed to handle that corner case... Aug 02 11:18:38 ok, just wanted to know Aug 02 11:20:02 khem: Hi. You added boot 1.50 to OE-core this july. Do you see any reason why the recipe uses the build-hosts compiler? Aug 02 11:20:33 rob_w: FYI you can set SANITY_TESTED_DISTROS = "" in your local.conf to avoid the warning if it's annoying Aug 02 11:21:06 no worrys Aug 02 11:21:49 i am pretty sure it will work ;-) Aug 02 12:41:39 hi all Aug 02 12:42:25 It is possible change/force building an image with different package? Aug 02 12:43:01 for example change qt4-x11-free with qt4-x11-free-gless Aug 02 12:43:28 so that the "gless" one is the default Aug 02 12:46:23 silvio_: PREFERRED_PROVIDER_qt4-x11-free = "qt4-x11-free-gles" I would think Aug 02 12:46:28 presumably this is OE-Classic Aug 02 12:48:12 : i try but doesnt work in this way Aug 02 12:48:52 silvio_: after doing that what do you get if you do: bitbake -e | grep "PREFERRED_PROVIDER_qt4-x11-free" Aug 02 12:48:59 : instead of = was ?= i canghed in =, but still try to compile the xfree without gless Aug 02 12:49:43 PREFERRED_PROVIDER_qt4-x11-free="qt4-x11-free-gles" Aug 02 12:50:07 ok, so that setting is being used at least Aug 02 12:50:16 i rebuild with -c clean need maybe "cleanall"? Aug 02 12:50:26 I very much doubt that will help Aug 02 12:51:24 I thought to install package after, but i think that some dependencies will be breack Aug 02 12:51:32 sorry for my english Aug 02 12:52:20 silvio_: your english is just fine :) Aug 02 12:52:42 silvio_: hmm... do you have a valid provider of virtual/egl ? Aug 02 12:53:19 silvio_: oh, wait a second Aug 02 12:53:49 silvio_: I'm guessing your MACHINE is not omap3 or ti816x, is that correct? Aug 02 12:54:06 I put and it compile the qt4-x11-free-gless.bb recipe Aug 02 12:54:38 yes it is a nvidia, i use the nvsamples egl Aug 02 12:54:51 is tegra2 Aug 02 12:55:00 so MACHINE is set to tegra2? Aug 02 12:55:27 i add it yes Aug 02 12:55:46 real is a colibri-T20 from toradex, and use tegra2 chip Aug 02 12:56:25 ok, FYI the qt4-x11-free-gles recipe does "require recipes/egl/egl.inc" Aug 02 12:56:42 and recipes/egl/egl.inc sets COMPATIBLE_MACHINE = "omap3|ti816x" Aug 02 12:56:57 which means if your machine is not either of those, it will be skipped Aug 02 12:57:29 i added colibri-t20 to egl.inc Aug 02 12:57:39 ok, good Aug 02 13:01:28 bluelightning, we have BBFILES := "${OEHOME}/stuff/btw/recipes/*/*.bb ${OEHOME}/stuff/trdx/recipes/*/*.bb ${OEHOME}/stuff/openembedded/recipes/*/*.bb" Aug 02 13:01:28 silvio_: bitbake -D may give some information about PROVIDES and DEPENDS and how they are being processed which may help Aug 02 13:02:00 and I created a btw/recipes/egl/egl.inc with COMPATIBLE_MACHINE = "omap3|ti816x|colibri-t20" Aug 02 13:05:41 silvio_: MACHINE="tegra2" or "colibri-t20" ? Aug 02 13:06:21 in build/conf/local.conf MACHINE="colibri-t20" Aug 02 13:33:29 morning. I'm having some problems building images with poky/oe-core after I add meta-oe to bblayers. Target arch is x86 and it looks like the udev in meta-oe doesn't put udevd/udevadm in the right places (/lib/udev instead of /sbin). Any advice? Aug 02 13:38:03 coldsoup: depending on what you're trying to do, I would say set PREFERRED_VERSION_udev to 164 so that the OE-Core/Poky version is used Aug 02 13:39:17 Yea, I noticed similar lines used for the kernel in the distro recipe/confs. Does that go in local.conf? Aug 02 13:39:37 Do somone of you have time to check my boost compilation error? Aug 02 13:39:45 coldsoup: it can, yes Aug 02 13:40:25 SorenHolm: I can look but I'm by no means any kind of expert on boost... Aug 02 13:40:33 bluelightning: thanks, I'll give it a whirl. Aug 02 13:41:08 SorenHolm: same (definitely no expert), but I'm doing things involved with boost and I think it compiles for me Aug 02 13:42:30 coldsoup: bluelightning: great.... It actually compiles, It just uses the wrong build systems compiler. Which means that on a 32bit system you'll get no error problably. Aug 02 13:42:56 bluelightning, any idea on why it says me ERROR: Multiple .bb files are due to be built which each provide qt4-x11-free (/home/oe/stuff/openembedded/recipes/qt4/qt4-x11-free-gles_4.7.3.bb /home/oe/stuff/openembedded/recipes/qt4/qt4-x11-free_4.7.3.bb). ? Aug 02 13:44:56 silvio_: because you can build two flavors of QT (a version for X11 only, and a GLES-supporting version). Both provide qt4-x11-free Aug 02 13:45:12 SorenHolm: what recipe are you building? Aug 02 13:45:23 coldsoup: boost Aug 02 13:45:59 SorenHolm, that's clear to me. why they are both being built if i put PREFERRED_PROVIDER_qt4_x11-free = "qt4-x11-free-gles" in my machine.conf ? Aug 02 13:47:07 silvio_: hmmm ...... I don't know that. Are you sure they are both build. It might just be a warning+ Aug 02 13:47:13 silvio_: ? Aug 02 13:48:26 SorenHold: are you sure it's not building both a native and target version? it DEPENDS = "boost-native..." Aug 02 13:49:11 SorenHolm, it is building for sure qt4-x11-free now. Aug 02 13:49:24 silvio_: strange Aug 02 13:49:27 the gles version is already compiled Aug 02 13:49:28 oops, butchered your name ;-) Aug 02 13:49:41 coldsoup: :D Aug 02 13:50:02 coldsoup: it builds both yes. The native one does not fail whereas the target-version does Aug 02 13:50:13 oh Aug 02 13:51:17 coldsoup: user-config.jam contains the correct compiler path Aug 02 13:52:01 SorenHelm: I'm enjoying the comments left in this recipe, heh Aug 02 13:53:01 silvio_: it's usually because one of the recipes provides something the other does not Aug 02 13:53:41 bluelightning, I resolved adding RPROVIDES_${PN} = "qt4-x11-free" Aug 02 13:53:41 RPROVIDES_${PN}-dev = "qt4-x11-free" to gles recipe Aug 02 13:54:05 er, you don't want the latter Aug 02 13:54:20 the dev ? Aug 02 13:54:23 ${PN}-dev should not be providing qt4-x11-free Aug 02 13:54:42 qt4-x11-free-dev maybe if that's really required, but I wouldn't think so Aug 02 13:56:37 bluelightning, next one :) Aug 02 13:56:39 | * satisfy_dependencies_for: Cannot satisfy the following dependencies for task-qt4-x11-base: Aug 02 13:56:39 | * qt4-x11-free Aug 02 13:56:49 | * resolve_conffiles: Existing conffile /home/oe/output/rootfs/btw-x11-demo-image/etc/device_table is different from the conffile in the new package. The new conffile will be placed at /home/oe/output/rootfs/btw-x11-demo-image/etc/device_table-opkg. Aug 02 13:57:20 silvio_: I'm guessing qt4-x11-free-gles do_package did not get re-run since you added that RPROVIDES_${PN}... ? Aug 02 13:57:48 bluelightning, no :) Aug 02 13:58:03 silvio_: right, so the package does not have the runtime provides in it Aug 02 13:58:14 coool Aug 02 13:58:27 seeemed harder than real :) Aug 02 13:58:34 SorenHolm, the meaning of this recipe escapes me. Seems like it should work if you have the right path in user-config.jam Aug 02 13:59:02 this is OE-Classic so none of that happens automatically, either rebuild qt-x11-free-gles or force do_package and then do_package_write_ipk for that recipe Aug 02 13:59:20 I'm guessing those gles recipes have not had a huge amount of testing in recent times Aug 02 13:59:24 coldsoup: yeah ..... do you want some log og something? Aug 02 14:01:00 SorenHolm, eh. You probably know more about this than I do. I've only been working with oe for the past month and thought I might be able to help someone else while I was getting help. Looks like I was wrong. I can try building just boost in a minute to see if it succeeds for me Aug 02 14:01:57 SorenHold: but I think it has been building (dependency of gnuradio which is my interest), granted I'm targeting atom-pc which should be 64-bit x86. Aug 02 14:03:33 coldsoup: ..... ok ...... I'll leave now. B back later. Aug 02 14:43:38 Pseudo is not present but is required, building this first before the main build what is that ? Aug 02 14:45:32 rob_w: it's just telling you what it's doing, nothing to worry about Aug 02 14:45:45 ERROR: Nothing PROVIDES 'pseudo-native' Aug 02 14:45:53 this is how my build ends Aug 02 14:46:00 rob_w: er, that should not happen Aug 02 14:46:11 is it wrong to not build anything after the initial git clone Aug 02 14:46:12 rob_w: what is your current configuration? Aug 02 14:46:21 rob_w: no, that's not the issue Aug 02 14:46:26 and bring in meta layers before that first build ? Aug 02 14:46:33 rob_w: that's fine Aug 02 14:46:47 well i cloned fresh .. took meta-ti , meta-angstrom, meta-igep Aug 02 14:47:00 configured MACHINE to igep0020 and started Aug 02 14:47:17 pseudo-native should be able to be found within OE-Core; if it can't find that something is pretty seriously wrong Aug 02 14:47:48 k Aug 02 14:47:53 checking Aug 02 14:47:54 rob_w: could you pastebin your bblayers.conf ? Aug 02 14:48:40 err Aug 02 14:48:54 just deleted the whole thing and redo it ... then pastebin all .. hang on Aug 02 14:49:11 that's a bit drastic, but ok... Aug 02 14:49:36 good for me to learn Aug 02 14:50:02 need to get confortable here , oe-classic was my thing the last years Aug 02 14:50:48 ok, sure Aug 02 14:58:45 http://pastebin.com/9FpWapPD same error ... check this out Aug 02 15:23:28 rob_w: this is the problem: ERROR: ParseError at /data/oe-core/meta-ti/recipes-misc/payload/bonescript.bb:5: Could not inherit file classes/systemd.bbclass Aug 02 15:23:40 working on it already Aug 02 15:25:07 ..building Aug 02 15:27:10 * rob_w needs more practise .. an doesnt like the fact that a quick look into recipes/* showed me all available apps ... anyway .. Aug 02 15:27:29 running home .. thx bluelightning !! Aug 02 15:27:34 np Aug 02 16:11:30 How do I use BBMASK to mask polkit_0.104.bbappend? I tried BBMASK+="polkit*bbappend", but bitbake -g still shows it's stuff (namely systemd) Aug 02 16:13:13 coldsoup: it's a regex, you probably need to change it accordingly Aug 02 16:13:55 does the type of assignment matter? (+=, =, :=) Aug 02 16:15:35 bluelightning: what's a good way to test that the regex is working? Aug 02 16:16:03 it depends on what's already in the variable Aug 02 16:16:22 it starts off blank by default Aug 02 16:21:24 bluelightning: oh dang. I think I figured out what I've been doing wrong. I was leaving out the surrounding //'s! Thanks for the help Aug 02 17:17:36 coldsoup, is the meta-oe polkit still pulling in systemd? Aug 02 17:17:45 I thought that was fixed with meta-systemd? Aug 02 17:18:32 have you rebuilt that? Aug 02 17:19:34 me, not yet :() Aug 02 17:19:36 :) Aug 02 17:21:22 Crofton|work: maybe you'll have to disable it explicitly http://git.shr-project.org/git/?p=meta-smartphone.git;a=commit;h=fdfca500137491ee3a91c457213eba9bf5fe08a6 and then reenable it in meta-systemd layer Aug 02 17:22:17 ah this one is better http://git.shr-project.org/git/?p=meta-smartphone.git;a=commit;h=17e1219e7a530e7c0801ebb12bde484ad30ffbe5 Aug 02 17:31:55 Crofton|work: ugh, it's been a mess, and it's all grc's fault. I hate that program. Aug 02 17:32:39 Crofton|work: but I finally got an image built with oe booting on x86 WITH gnuradio Aug 02 17:34:52 meta-oe is not _really_ distribution independent. I would be happier if there was a meta-scientific or something that had gr, numpy, plotting tools, etc... Maybe I'll do that on my own time ;-) Aug 02 17:36:18 talk on the list Aug 02 17:36:31 cool Aug 02 17:36:46 and we are interested in the troublesome spots Aug 02 17:37:46 the problem wound up (partially) being that udev recipes from meta-oe doesn't put binaries in to /sbin like it should. So on syslinux bootup the /init script in initramfs can't find udevd. The oe-core version works just fine though. Aug 02 17:38:46 I think that might only affect x86 since no one else probably uses syslinux Aug 02 17:39:23 at least I'm done banging my head on the keyboard for the moment :-) Aug 02 17:56:15 the problem is gnuradio tries to be all things to all people Aug 02 17:56:29 and that makes it hard to use effectievly on embedded systems Aug 02 17:56:50 we ccould try an alternate recipe more suited to your needs? Aug 02 18:34:40 Crofton|work, hi, does the grc run on the E100 now? Aug 02 18:36:49 it can Aug 02 19:13:02 is there a way to outright disable systemd for meta-oe? Aug 02 19:19:08 (for denzil release before it was separated) Aug 02 19:20:42 msm: yes, don't pull any -systemd package to image and disable it explicitly in polkit Aug 02 20:40:43 coldsoup: Did you find out anything? Aug 02 22:28:09 re Aug 02 22:30:53 hi pb_ Aug 02 22:32:16 hi ant_______ Aug 02 22:39:08 heh, I see ljp is all over the news at the moment. **** ENDING LOGGING AT Fri Aug 03 02:59:58 2012