**** BEGIN LOGGING AT Mon Nov 25 02:59:58 2013 Nov 25 07:57:02 good morning Nov 25 09:46:40 morning all Nov 25 09:48:07 bluelightning: Good Morning Nov 25 09:48:14 hi Zaif Nov 25 10:12:42 hi yocto....while i am compiling ,compilation was interrupted by power failure,now its throwing error , i have done " -c cleanall" , but i cant clear the stamps....plz help me Nov 25 10:13:50 elango: it should be safe to delete tmp folder. and run again. Nov 25 10:14:05 it would use the sstate anyways on next run. Nov 25 10:19:22 ndec: thank u :) Nov 25 10:29:20 ndec: i have faced this problem, while compiling mplayer package seperatly....if i remove tmp i would loss all my images...so wat should i do ..?? Nov 25 10:29:58 elango: is there anything in your image that wouldn't be re-generated by a new build? Nov 25 10:32:19 ndec: removing tmp will cause recompilation of all packages again, rite? Nov 25 10:32:31 no ;-) Nov 25 10:32:42 there is what's called 'sstate. Nov 25 10:33:24 it's a local 'mirror' of everything that was built by OE. it's by default /sstate-cache Nov 25 10:34:02 so you can delete (not ) and the rebuild will only be about pulling prebuilts from sstate. so generally a couple of mins Nov 25 10:34:16 k. Nov 25 10:34:17 fyi: http://www.yoctoproject.org/docs/1.5/ref-manual/ref-manual.html#shared-state-cache Nov 25 10:34:37 ndec: Thanks for your info. Let me try Nov 25 12:10:28 Hi, I've just ran " source ./setup-environment build" which resulted in "configuring for imx28evk ,The project root content has been updated. Please run bash again. How do I run bash again????? Nov 25 12:11:12 D4v33: just run " source ./setup-environment build" again Nov 25 12:11:42 I thought thant but I get the same message repeated Nov 25 12:13:11 D4v33: tried from a "fresh" shell? Nov 25 12:13:43 same from a fresh shell Nov 25 12:17:50 D4v33: are you somehow modifying setup-environment before source'ing it? Nov 25 12:18:26 not that i know of, i'm looking at the script and that seems Nov 25 12:18:40 to be what it's checking Nov 25 12:21:25 updated alway seems to go true Nov 25 12:21:39 D4v33: can you manually run those checks? Nov 25 12:24:13 You can also try "set -x; source ./setup-environment build; set +x" to get more debug information Nov 25 12:25:46 what is was, was that I'd changed the setup-environment script in the root and that was then out of step with the one under sources Nov 25 12:26:03 Hi, any toolchain maintainers online ? Nov 25 12:26:47 I am trying to build a toolchain with $BUILD=x86 $HOST=arm and $TARGET=epiphany-elf ... Nov 25 12:27:00 is there a support for such a build ? Nov 25 12:34:29 any help guys.. Nov 25 12:38:30 had anyone tried building a toolchain with diff values for $HOST and $TARGET ? Nov 25 12:53:24 akini: look at the nativesdk and cross-canadian variants, those are as close as you'll find Nov 25 12:55:09 @RP, thanks. Nov 25 14:38:25 hi, is there a nicer way than this to set root password? https://lists.yoctoproject.org/pipermail/poky/2012-June/007987.html Nov 25 14:38:32 the default seems to be empty for root! Nov 25 14:39:56 lpapp: Add this to your local.conf (or conf/distro/.conf): Nov 25 14:40:05 lpapp: INHERIT += "extrausers" Nov 25 14:40:22 lpapp: EXTRA_USERS_PARAMS += "usermod -p '' root; " Nov 25 14:40:39 lpapp: And replace with something appropriate. Nov 25 14:45:28 Saur: ah, yes, found the commit in the meantime for the sample, thanks. Nov 25 14:47:38 to clarify, the default root password is blank if you have "debug-tweaks" in IMAGE_FEATURES, or no password (i.e. you can't log in at all) without that Nov 25 14:47:59 I can log in. Nov 25 14:48:07 by simply pressing enter. Nov 25 14:49:22 right, so you must have debug-tweaks in your IMAGE_FEATURES value (which is the default) Nov 25 14:49:33 right. Nov 25 14:51:12 I would prefer a variable for it as it is quite common for a system to set up a root password.... Nov 25 14:52:15 sounds reasonable? Nov 25 14:54:58 there is already https://bugzilla.yoctoproject.org/show_bug.cgi?id=5365 open to cover this Nov 25 14:54:59 Bug 5365: enhancement, Medium, 1.6 M1, Qi.Chen, IN PROGRESS IMPLEMENTATION , ability to set root password in cleartext Nov 25 14:56:48 yeah, well, it covers more. Nov 25 14:57:04 it covers my case + it also requests clear text. Nov 25 14:57:10 although the title is wrong for the coverage of the report Nov 25 14:57:46 well, not doing it in cleartext results in the usability problems outlined in the report Nov 25 14:58:35 I left a comment. Nov 25 15:00:30 bluelightning: what usability problem? Nov 25 15:00:35 FOO = hash would work fine Nov 25 15:00:37 read the bug report Nov 25 15:00:54 I did, and I do not understand what you are referring to. Nov 25 15:01:23 anyway, I do not wanna have cleartext. Nov 25 15:01:43 so in that case, I need a different variable for this. Nov 25 15:02:18 although the hash might be different on different systems. Nov 25 15:02:26 I mean the algorithm. Nov 25 15:02:34 so we probably need to compromise security. :/ Nov 25 15:03:36 of course it depends on what you're trying to do, but I might suggest if you have the same default root password on more than one device your security is probably already somewhat compromised... Nov 25 15:06:29 Is there any way to use an if construct within ${@...} ? RP? bluelightning? Nov 25 15:08:15 Saur: assuming base_conditional() / base_contains() won't work, yes you can also express an if statement on one line in python Nov 25 15:08:57 bluelightning: Well, I tried to do this: SRC_URI += "${@if int(d.getVar('PV', True)) >= 208: 'file://mypatch.patch'}" Nov 25 15:09:04 But it didn't work... Nov 25 15:10:24 that doesn't look like valid python syntax to me Nov 25 15:11:35 bluelightning: Well, it works when I try it in the interactive python, but I have to have an empty line to terminate the if statement... Nov 25 15:11:47 Saur: I think you may have to do something ugly like ${@['', 'file://mypatch.patch'][int(d.getVar('PV', True)]} Nov 25 15:12:03 i.e. treating the boolean condition as an index into the list Nov 25 15:12:10 Ah, I see :) Nov 25 15:12:21 oops, I missed a closing ) at the end there Nov 25 15:12:21 hi guys, trying to figure out what to do to enable gettext for uclibc (libintl) Nov 25 15:12:22 Python magic... Nov 25 15:14:29 grepping all possible DISTRO_FEATURES in poky for gettext/libintl gives no results Nov 25 15:16:34 bluelightning: Thanks a lot. It worked like a charm. :) Nov 25 15:17:21 bluelightning: Needed this to make use of the new wildcard support for bbappends, but where different patches are needed for different versions. :) Nov 25 15:17:47 Now all I need is the wildcard support backported to dora... Nov 25 15:18:38 or if you wanted to get fancy, OVERRIDES =. "version/${PV}:"; FOO_version/1.0 = "bar" ;) Nov 25 15:18:42 morning all Nov 25 15:19:54 kergoth: Hmm, interesting alternative... Nov 25 15:21:03 also, you can use python's inline if statement support Nov 25 15:21:14 ${@'foo' if bar else 'baz'} Nov 25 15:21:26 slightly cleaner than the indexing the list approach Nov 25 15:21:30 kergoth: Though I think I prefer the possibility to use the >= operator so that we don't accidentally miss the patch when version 209 comes around... Nov 25 15:21:32 kergoth: ah yes, that is nicer... Nov 25 15:21:41 * kergoth nods at Saur Nov 25 15:22:03 I remember playing with stuff like this when poking at BBVERSIONS, which never ended up being used Nov 25 15:26:32 kergoth: Nice. This worked fine: SRC_URI += "${@'file://mypatch.patch' if int(d.getVar('PV', True)) >= 208 else ''}" Nov 25 15:26:44 cool Nov 25 15:35:49 bluelightning: XTRA_IMAGE_FEATURES = "debug-tweaks" Nov 25 15:35:56 that is explicitly there for us, apparently uselessly? Nov 25 15:41:10 Hi all. I have one local package - not present on the network. How to configure SRC_URI? Nov 25 15:42:44 lpapp: sorry, I don't understand what you mean Nov 25 15:43:13 bluelightning: you said it is the default.... Nov 25 15:43:32 lpapp: it's the default, by way of that line in the default local.conf template... Nov 25 15:48:28 is there any way to tell bitbake not to search ackage on the web, but that it is in local directory? Nov 25 15:49:15 SRC_URI = "file:///some/where/file.tar.gz" ? Nov 25 16:03:44 lexszero, this is relative path? Nov 25 16:06:28 drasko: https://www.yoctoproject.org/docs/latest/mega-manual/mega-manual.html#var-SRC_URI afair absolute paths worked too, note extra slash Nov 25 16:09:04 better way is to put your source package to the some dir in FILESPATH or even run a local http server if build supposed to be run on different machines (e.g. developers in the company) Nov 25 16:30:28 lexszero, my package is in the FILESPATH, but I thought that you have to put something in SRC_URI anyway Nov 25 16:31:26 yes, just put basename to SRC_URI Nov 25 16:35:11 In order to determine which MACHINE I can set I look in /source/meta-fsl-arm/conf/machine this lists the MACHINE I want "imx28evk.conf" I then set it in local.conf or set and export on the cmd line but bitbake still comes up with please set a valid MACHINE in your local.conf or env message any idea's Nov 25 16:35:52 lexszero, what do you mean exactly? Nov 25 16:36:08 I have got Failure expanding expression ${@bb.fetch.get_checksum_file_list(d)} which triggered exception MalformedUrl Nov 25 16:36:31 : The URL: 'hostapd-0.8.tar.gz' is invalid and cannot be interpreted Nov 25 16:36:35 drasko: SRC_URI = "file://yourpackage.tar.gz" Nov 25 16:36:43 ah Nov 25 16:36:55 and you need to put checksums into recipe Nov 25 16:37:09 ERROR: Function failed: Fetcher failure for URL: 'file://hostapd-0.8.tar.gz'. Unable to fetch URL from any source. Nov 25 16:41:39 bluelightning: yes... Nov 25 16:41:48 so it means our line is superflous Nov 25 16:42:58 lpapp: if you've added it again later, yes Nov 25 16:43:26 lexszero, any idea what's missing? Nov 25 16:43:49 SRC_URI = "file://yourpackage.tar.gz" does not seem to work for me Nov 25 16:44:37 what bitbake says now? Nov 25 16:45:17 D4v33: MACHINE should be imx28evk, not imx28evk.conf. and make sure meta-fsl-arm is in BBLAYERS in bblayers.conf Nov 25 16:46:18 bluelightning: yeah, right. Nov 25 16:48:55 kergoth: your right I way just referring to the file of that name. I had this working last week but decied to start a fresh so that I knew I could repeat it with confidence but I'm seeing the same problem with MACHINE, last week all that was required was to export it but this doesn't work this time Nov 25 16:51:13 I think you told me this last week but I had foggetten about the paths being in bblayers.... I do that now thanks Nov 25 16:51:15 lexano, it says: ERROR: Function failed: Fetcher failure for URL: 'file://hostapd-0.8.tar.gz'. Unable to fetch URL from any source. Nov 25 16:54:11 kergoth: You were right again and I've made a metal and perm note to the effect... many thanks Nov 25 16:54:29 s/metal/mental Nov 25 16:56:03 np Nov 25 16:58:10 Next problem I'm getting red ERROR messages following failed to fetch URL warnings, does this mean that no download was possible or something else? Nov 25 17:02:03 D4v33: depends, what does the error say? Nov 25 17:02:37 bluelightning: Error executing a python function in /home/dave/yocto/fsl-community-bsp/sources/poky/meta/recipes-devtools/gnu-config/gnu-config_20120814.bb: then a stack trace Nov 25 17:02:58 so, stack trace = bug Nov 25 17:03:09 D4v33: can you pastebin the full error message? Nov 25 17:04:47 bluelightning: http://pastebin.com/mKS91zQM Nov 25 17:10:38 bluelightning: looks to me that its the lack of a downloads directory "Exception: OSError: [Errno 2] No such file or directory: '${BSPDIR}/downloads/' " but what path should {BSPDIR be? Nov 25 17:12:47 bluelightning: jsut looked in my build dir to find a recursive folder named ${BSPDIR} ! Nov 25 17:13:22 3 levels and not infinite! Nov 25 17:13:55 sounds like you need to talk to fsl Nov 25 17:14:34 kergoth: I'll post on there community Nov 25 17:14:50 s/there/their Nov 25 17:16:45 yep sounds like some setup script brokenness to me Nov 25 19:44:15 bluelightning: is it just my feeling or the fsl-community-bsp thing is really obscure piece of c..p compared to yocto ? Nov 25 19:44:47 Marex: I haven't looked at it at all Nov 25 19:45:01 recent support requests seem to suggest it might be slightly broken though Nov 25 19:45:34 * mranostay spots Marex Nov 25 19:45:59 mranostay: hello kitty :) Nov 25 19:46:14 mranostay: I'm like a ghost, I'm everywhere ;-) Nov 25 19:47:46 ghosts are omnipresent? Nov 25 20:09:41 Hmm, a way to backfill PACKAGECONFIG would be useful Nov 25 20:32:49 kergoth: something like DISTRO_FEATURES + backfill + considered? What's the use case for that? When I was adding PACKAGECONFIGs I was trying to guess default value based on what was most likely built before (in cases where it was for autodetected dependency) Nov 25 20:33:27 Kangkai: but of course it was only guess, so I don't know what else I would put in backfill if there is special variable for that Nov 25 20:35:32 JaMa: Any case where a recipe defaulted to enabling something transparently, and you switch to user control. If the user is currently setting packageconfig, then they'd not have the new addition to the default value, and the behavior will change. if we could backfill it in, then they wouldn't be affected by it Nov 25 20:35:42 JaMa: same use case as distro features, basically Nov 25 20:35:49 * kergoth shrugs Nov 25 20:37:54 JaMa: in this case, i'd like to add a packageconfig to let you control sysvinit compatibility for systemd, without breaking anyone's existing systems who define systemd's PACKAGECONFIG Nov 25 20:37:58 ah right, sorry my head was stuck in that scenario with autodetected deps Nov 25 20:38:00 I see now Nov 25 20:38:12 obviously, could use PACKAGECONFIG_append for now, and if pressed the user could use _remove, which would take precedence Nov 25 20:38:19 but a backfill mechanism would be a bit cleaner Nov 25 20:38:46 JaMa: hey ;-) Nov 25 20:39:02 Marex: hey, zombies are comming! :) Nov 25 20:39:33 JaMa: yeah, looking forward ... glad it's not today, I might not make it through Nov 25 20:39:34 gnight (lol) Nov 25 20:43:39 you never know.. it could start today :) Nov 25 20:49:03 * mr_science tests camera streaming Nov 25 21:09:12 Opinion: should os-release be pulled in by base-files, or just by systemd and images, or via some other mechanism. Thoughts? Nov 25 21:09:24 * kergoth is working through his push-to-upstream backlog Nov 25 21:19:16 anybody have access to fix http git pulls from git.yoctoproject.org? The kernel repos appear to be broken. Nov 25 21:19:39 http://git.yoctoproject.org/cgit/cgit.cgi/yocto-kernel-tools/ shows the http: git clone url as: http://git.yoctoproject.org/git/yocto-kernel-tools Nov 25 21:20:04 But I get: Nov 25 21:20:06 $ git clone --mirror http://git.yoctoproject.org/git/linux-yocto-3.10 Nov 25 21:20:06 Initialized empty Git repository in /home/michael_e_brown/f10/sources/g/linux-yocto-3.10.git/ Nov 25 21:20:06 error: RPC failed; result=52, HTTP code = 100 Nov 25 21:20:17 Where other git repositories on that server work ok Nov 25 21:20:31 I'm going through an HTTP proxy to do my clones Nov 25 22:00:39 kergoth: Any opinions on http://git.yoctoproject.org/cgit.cgi/poky-contrib/commit/?h=rpurdie/t2&id=8372b2ff637b8c5e7f5c61121513a39ba9361d6a ? Nov 25 22:01:17 kergoth: Part of me hates special casing something like this, equally it would be a big win for usability without having to get to the full typed variable situation quite yet Nov 25 22:05:47 RP: I hate that too, but it sounds like it's an okay temporary measure to take in the meantime. I would suggest, however, changing a couple things for clarity for the reader. s/flag/word/, to avoid possible confusion with actual flags, and similarly, %s{%s} = {,Un}Set rather than [] Nov 25 22:07:58 kergoth: good point, I'll tweak that Nov 25 22:08:44 kergoth: I can't see this binding us too much in the future so it might be ok for now Nov 25 22:11:45 should be straightforeward to pull this back out in the future, we'll be neck deep in that code to checksum typed variables anyway, presumably Nov 25 22:13:15 kergoth: right Nov 25 22:13:18 are you going to add DISTRO_FEATURES to the whitelist, if that gets merged? Nov 25 22:13:36 kergoth: the beauty is I shouldn't need to Nov 25 22:13:37 also, we'll have to make sure we don't have any lurking instances of e.g. ${@'foo' if 'bar' in DISTRO_FEATURES.split() else 'baz'} Nov 25 22:13:52 kergoth: right, it will force us to do a bit of cleanup Nov 25 22:14:11 a quick grep shows about 30 or so usages in OE-COre Nov 25 22:14:15 ah, right, you don't add the contain var name to references, only to contains. cool Nov 25 22:14:18 (that need some tweaking) Nov 25 22:14:21 * kergoth nods Nov 25 22:14:47 kergoth: it turns out the code hits TUNE_FEATURES more than DISTRO_FEATURES Nov 25 22:15:31 ah, didn't think about that. makes sense, it's used all over the tune files Nov 25 22:15:47 kergoth: neither did I until I put some prints in :) Nov 25 22:16:37 Any opposition to pulling in a few small patches to deal with CLOEXEC & the like for old hosts? Those hosts aren't supported anymore, but the patches would be needed to let them build even when using buildtools-tarball. we've got af ew floating around for centos5/centos6 Nov 25 22:17:07 kergoth: For buildtools-tarball that probably makes sense Nov 25 22:18:13 all trivial changes, just wanted to check since the distros aren't officially supported anymore Nov 25 22:18:18 * kergoth adds to his queue **** ENDING LOGGING AT Tue Nov 26 02:59:58 2013