**** BEGIN LOGGING AT Tue Feb 05 02:59:59 2013 **** ENDING LOGGING AT Tue Feb 05 05:35:20 2013 **** BEGIN LOGGING AT Tue Feb 05 06:02:19 2013 Feb 05 08:17:39 good morning Feb 05 08:36:51 good morning Feb 05 08:59:33 morning all Feb 05 09:22:22 morning all Feb 05 09:34:57 hi bluelightning, all Feb 05 09:35:32 morning eFfeM_work Feb 05 10:23:04 hi, while trying to build core-image-minimal for hawkboard from poky and meta-ti layer (master branch), I am facing some LICENSE related errors Feb 05 10:24:18 like : ERROR: This recipe does not have the LICENSE field set (cloud9-gnome-image) Feb 05 10:25:56 is it safe to assume the LICENSE as "gpl" ? Feb 05 10:26:12 for recipes yocto normally uses MIT Feb 05 10:26:30 rburton: thanks.. Feb 05 10:26:32 but that's in meta-ti, so you should tell them Feb 05 10:27:14 rburton: this is the full error long . http://pastebin.com/9FdthKwb Feb 05 10:29:26 rburton: thanks, I will mail. Can anybody please share the contact details of people responsible for meta-ti ? Feb 05 10:31:22 maxin: see meta-ti/README Feb 05 10:32:12 rburton: got it .. Thanks :) Feb 05 12:48:36 Corneliu: what remains for running ptests in/against/with Testopia? Feb 05 13:09:11 help Feb 05 13:09:23 sorry /help Feb 05 13:12:17 Zagor: I got some network problems atm Feb 05 13:12:54 Zagor: we would like to introduce ptest test cases but we are very busy atm Feb 05 13:13:49 Zagor: so if someone could help out with the steps / expected results phase it would be great Feb 05 13:14:46 Corneliu: what is it you need? Feb 05 13:15:06 I must admit the Testopia concept is still a bit unclear to me Feb 05 13:16:17 well basically something like this: https://bugzilla.yoctoproject.org/tr_show_case.cgi?case_id=418 Feb 05 13:17:17 Zagor: if you have any questions I will gladly try to help Feb 05 13:17:19 Corneliu: but those are manual tests, right? I though we would handle automatic tests differently. Feb 05 13:17:52 some ptests have 15000 test cases. we don't want to write up manual test protocols for those... Feb 05 13:18:03 Zagor: indeed but that wil come at a later stage Feb 05 13:18:29 Zagor: we are working on an automation framework based on the autobuilder Feb 05 13:18:47 Zagor: in the meantime many runtime tests are done manually Feb 05 13:19:47 Zagor: besides that, having defined test cases wirh how features should be tested (ex: manual test example) will help us create an optimal automated model later Feb 05 13:25:04 Zagor: basically Testopia is a test case management platform Feb 05 13:25:29 Zagor: how we choose to execute the test cases is up to us, though it offers support for automated scripts Feb 05 13:26:43 we are working on other nice features like automatd reports. Here is an example: https://wiki.yoctoproject.org/wiki/Test Feb 05 13:27:02 this should be up and running this week Feb 05 14:46:36 Good day folks. I'm looking to create a recipe for a few python packages that something I'm working on depend upon. Unfortunately, I'm neither a bitbake nor a python expert and my ignorance of both is making it somewhat difficult. I've created recipes for the packages that inherit 'distutils', and apparently part of that class translates the shebangs at the top of scripts to #!/usr/bin/python-native/python, but naturally python isn't located there Feb 05 14:50:51 Garibald1: have you checked if those recipes aren't already written somewhere somehow? Feb 05 14:51:11 Garibald1: under the OE layer index page Feb 05 14:52:09 mihai: I looked over it, nothing seemed appropriate Feb 05 14:52:35 I'm doing logilab-common, logilab-astng and logilab-pylint Feb 05 14:53:55 Garibald1: the problem is probably that the logilab build process is using $PYTHON, and assuming that it's good on the target Feb 05 14:53:59 which it isn't Feb 05 14:55:18 in fact, distutils.bbclass should be fixing that for you Feb 05 14:55:22 sed -i -e s:${STAGING_BINDIR_NATIVE}/python-native/python:${bindir}/env\ python:g $i Feb 05 14:55:50 that only runs on scripts in /usr/bin though Feb 05 14:55:59 so maybe you're scripts are not there? Feb 05 14:56:18 or, you've overwritten the distutils do_install Feb 05 14:56:23 let me take a look Feb 05 14:56:30 helps if you can share your recipe Feb 05 14:56:43 will do, 1 sec Feb 05 15:01:04 here's one of them Feb 05 15:01:06 http://pastebin.com/zD4wmxk6 Feb 05 15:02:43 that one has a script named 'pytest' Feb 05 15:04:17 and looking at tmp/work/i586-poky-linux/python-logilab-common-0.59.0-r0/image/usr/bin/pytest I see #!/usr/bin/pythbon-native/python -u Feb 05 15:04:31 s/pythbon/python/ Feb 05 15:04:55 rburton: is that the right place to be looking? Feb 05 15:05:26 :q Feb 05 15:07:55 hm Feb 05 15:08:06 you could say that distutils.bbclass should be handling that case too Feb 05 15:09:27 rburton: i'd say for now, add a do_install_append() block and sed the files again to replace that path with ${bindir}/env python Feb 05 15:10:53 for i in ${bindir}? Feb 05 15:14:56 for i in ${bindir}/*; do sed -i -e 's:${bindir}/python-native/python:${bindir}/env python:'; done Feb 05 15:15:12 err, w/o single colons Feb 05 15:16:29 yes that Feb 05 15:17:32 hey guys Feb 05 15:21:12 rburton: hum, should ${bindir} be prefixed with somethign? Feb 05 15:21:20 ${D}/${bindir}? Feb 05 15:21:26 in the loop Feb 05 15:22:39 yeah that was the right variable Feb 05 15:23:03 oh, yes, sorry Feb 05 15:23:21 rburton: no problem, I should have caught that myself. That seems to have done the trick -- thanks. Feb 05 15:27:15 nice Feb 05 15:34:24 kergoth: bb needs bb log [package] [task] Feb 05 15:34:35 to show that log in less Feb 05 15:37:10 kergoth: if you can give a hint as to how i'd get the build directory i'll give it a go Feb 05 15:40:07 Hmm, not a bad idea. It'd do the same thing bb-show does to do a full recipe parse for that recipe, then it could use d.getVar() as usual to get the 'T' variable (which is WORKDIR/temp). ideally it'd support the new bb log file name format variable, though Feb 05 15:40:42 BB_LOGFMT, thats it Feb 05 15:41:57 looks like you could do with some helper modules to remove duplicated functions Feb 05 15:43:44 yeah, I've been slowing adding some of that to the core python module. haven't needed to move the get_data() bits from bb-show yet, since it was the only command that needed a full parse Feb 05 15:44:37 I'll accept pull requests if you get bored :) My current priority is finishing off bb-search and figuring out the best output format and default behavior for it Feb 05 15:44:58 e.g. should it also search SUMMARY/DESCRIPTION (hopefully not, since I don't think that's in the cache :) Feb 05 16:00:54 YPTM: Scott Rifenbark Joined call Feb 05 16:00:56 YPTM: jzhang's on the call Feb 05 16:01:01 YPTM: Bruce Ashfield on the call. Feb 05 16:01:04 YPTM: Tom Z on the call Feb 05 16:01:20 YPTM: Paul Eggleton is on the call Feb 05 16:01:22 YPTM: Björn Stenberg on the call Feb 05 16:01:28 I have no bridge info in my recurring Outlook invitation for this meeting anymore Feb 05 16:01:38 YPTM: Weird. the bridge info isn't in my calendar. Feb 05 16:01:48 scottrif: looks like you have the same issue. Feb 05 16:01:53 song accidentally sent out a cancellation Feb 05 16:02:04 a flurry of invites went out about two weeks ago and they disappeared Feb 05 16:02:10 YPTM: polk is here Feb 05 16:02:14 YPTM: Darren has joined Feb 05 16:02:18 YPTM: Welcome to the technical team meeting, please let me know who's on the bridge Feb 05 16:02:22 Song_Liu: YTPM: RP is here Feb 05 16:02:24 YPTM: Kevin Strasser is here Feb 05 16:02:26 belen is on the call Feb 05 16:02:38 YPTM: Chris Larson here Feb 05 16:02:40 YPTM: would someone PM me the dialin info? Feb 05 16:02:48 YPTM: Laurentiu Palcu joined Feb 05 16:02:59 YPTM: nitin is on the call Feb 05 16:03:02 YPTM: davest in the house Feb 05 16:04:14 kergoth: Thoughts on http://git.yoctoproject.org/cgit.cgi/poky-contrib/commit/?h=rpurdie/t1&id=7cccf54b342fc205e915422c6eeabfc7fd63d7e8 and then http://git.yoctoproject.org/cgit.cgi/poky-contrib/commit/?h=rpurdie/t1&id=99788945bce2167435f5d845f49939f5e24a0d5f ? Feb 05 16:04:35 YPTM: Cristian Iorga is present Feb 05 16:04:37 * RP isn't entirely happy with BB_ORIGENV but I'm not sure how else to do it. I dislike the hardcoded lists Feb 05 16:04:40 YPTM: Corneliu joined Feb 05 16:04:47 YPTM: ross joining Feb 05 16:05:03 YPTM: you have me for now but I night drop for a conflict Feb 05 16:05:05 YPTM: Song_Liu: Sean Hudson here. Feb 05 16:05:14 YPTM: AlexG here Feb 05 16:05:15 RP, I was considering something similar for the fetcher env list Feb 05 16:05:19 YPTM: any opens? Feb 05 16:05:21 YPTM: Saul is here Feb 05 16:05:25 RP, something like BB_PROXY_VARS Feb 05 16:05:36 YPTM: Song_Liu: Chris Kaminski is here Feb 05 16:05:36 Song_Liu: Calendar invite no longer has the dialin info. Feb 05 16:05:48 dvhart: its that work which prompted me to look at this. The fetcher is next on my list, I wanted to kill those ones first Feb 05 16:06:04 YPTM: Bogdan joined Feb 05 16:06:13 https://wiki.yoctoproject.org/wiki/Sample_-_1.4_M3.rc2_status Feb 05 16:06:24 RP, sure, I can do it to as a follow-one of the proxy work (as I'm sure another version will be needed) Feb 05 16:06:34 *smirk* Feb 05 16:06:35 breaker breaker buddy Feb 05 16:06:49 dvhart: I didn't really see an issue with that patch series tbh Feb 05 16:07:31 RP, huh... I figured at least the last one would get people squirming :) Feb 05 16:07:42 * dvhart notes he needs to try harder Feb 05 16:07:49 dvhart: which one? Feb 05 16:08:05 the one where I delete all the existing GIT_PROXY_COMMAND implementations Feb 05 16:08:40 alex in ? Feb 05 16:08:45 RP: Hmm, yes, looks good, though I also dislike the name.. I suck at names. bb_environment, bb_env, bb_shell_env, etc, none of those sound all that much better Feb 05 16:08:58 hmm Feb 05 16:09:15 alex in ! Feb 05 16:09:26 RP, are these all something that are needed before the BB_ENV_EXTRAWHITE kicks in? Feb 05 16:09:39 RP, if so, can they be shared with the fetcher environment list? Feb 05 16:11:51 dvhart: The key issue is we probably don't want to export the variables to all tasks, only the fetcher Feb 05 16:11:56 kergoth: $ bb log avahi configure <-- works now :) Feb 05 16:12:07 nice Feb 05 16:12:26 anything to avoid having to drill down deep into tmpdir should make people happy, heh Feb 05 16:13:44 rburton, ooooh, that makes *me* happy Feb 05 16:13:53 rburton++ Feb 05 16:15:50 remove cdt?! Feb 05 16:19:04 rburton: its not being supported past the 3.0 kernel Feb 05 16:19:16 that group are idiots Feb 05 16:19:19 * rburton sighs Feb 05 16:20:42 rburton: it will continue to work with danny but not in master Feb 05 16:21:43 RP: sure Feb 05 16:22:34 example: https://wiki.yoctoproject.org/wiki/Test Feb 05 16:23:36 sample: https://wiki.yoctoproject.org/wiki/Sample_-_1.4_M3.rc2_status Feb 05 16:25:03 * zeddii clicks and looks Feb 05 16:25:04 rburton, we may resurrect it with gma500 Feb 05 16:25:15 but for now, nothing we can do :/ Feb 05 16:25:34 dvhart: so its entirely because of the binary drivers? Feb 05 16:25:56 rburton, yes graphics for this board not supported beyond 3.0 Feb 05 16:26:02 rburton, *sigh* Feb 05 16:26:17 rburton, rest assured I've *discussed* this with them Feb 05 16:26:33 sure, and obviously there's nothing we can do Feb 05 16:26:44 dvhart: is anyone maintaining the open driver now alan has gone? Feb 05 16:27:00 dvhart, rburton: we can forward port the kernel to 3.8, shouldn't be a big deal Feb 05 16:27:23 dvhart, rburton: we've offered to do that too, just have no takers Feb 05 16:27:49 yeah, it's not the work, it's the support matrix they object to Feb 05 16:28:01 dvhart, rburton: or rather we can't put resources into it without some kind of buy-in Feb 05 16:28:11 right Feb 05 16:28:34 dvhart: right, if we took it over we wouldn't need to worry about that Feb 05 16:29:01 dvhart: rather see the pvr driver dumped and try with gma500 Feb 05 16:29:09 YPTM: if you are interested in reviewing the Web Hob interface, contact:  Feb 05 16:29:10 chris.kaminski@tobias.tv Feb 05 16:29:10 • review Web Hob UI Feb 05 16:29:10 • via remote session Feb 05 16:29:10 • approx. 30 min Feb 05 16:29:11 • recorded via Skype or TeamViewer  (but you can opt out of recording) Feb 05 16:29:20 dvhart: at any rate, no resources for it at the moment Feb 05 16:29:31 YPTM: Web Hob wiki is at: Feb 05 16:29:31 https://wiki.yoctoproject.org/wiki/Web_UI Feb 05 16:32:27 dvhart: anyway, feb 8 it will be removed from master if there's no progress Feb 05 16:33:16 dvhart: i guess the autobuilder will still break finding no bsp, but we've already requested it be removed from the autobuilder build Feb 05 16:34:00 tomz1, right, honestly the removal from the AB is the more pressing issue :-) Feb 05 16:34:42 YTPM: Web Hob mailing list: Feb 05 16:34:43  https://lists.yoctoproject.org/listinfo/webhob Feb 05 16:38:12 when was sound removed from qemu? Feb 05 16:38:26 I demo'd sound in 2010 on qemu as part of the kernel demo... Feb 05 16:39:50 we are still attempting to enable sound in the configs. I haven't tested it recently myself. Feb 05 16:40:02 KERNEL_FEATURES_append_qemux86=" cfg/sound.scc" Feb 05 16:40:02 KERNEL_FEATURES_append_qemux86-64=" cfg/sound.scc" Feb 05 16:40:15 ech. whitespace. Feb 05 16:40:26 sorry RP Feb 05 16:44:47 yea! Feb 05 16:48:40 what's the proper way for configure to detect the target arch? Feb 05 16:48:49 right now, I have a configure.ac that's using uname Feb 05 16:48:51 which is clearly wrong Feb 05 16:49:03 I can pass in TARGET_ARCH from the yocto build env, but that does not seem quite right Feb 05 16:49:12 i'm sure there is a fine example somewhere too... Feb 05 16:49:27 if you're talking autoconf style, I believe AC_CANONICAL_HOST does it Feb 05 16:51:05 it sets the $host variable in cpu-machine-OS style Feb 05 16:51:46 B4gder: sounds like what I might be looking for Feb 05 16:53:09 B4gder: AC_CANONICAL_SYSTEM sounds right? Feb 05 16:53:55 yeah, that's a somewhat extended version Feb 05 16:54:33 ok Feb 05 16:54:34 thansk Feb 05 16:54:36 http://www.gnu.org/software/autoconf/manual/autoconf.html#Canonicalizing seems to be the docs Feb 05 16:57:45 Song_Liu: YPTM: Had to leave call early to open bridge for another call. Feb 05 17:32:50 While compiling danny php_5.3.6.bb Feb 05 17:32:50 I frequently see this kind of error: Feb 05 17:32:50 Wrote PEAR system config file at: /sboxm7/jim/dan5/build/tmp/work/x86_64-poky-linux/php-5.3.6-r2.1/image//ssd/jim/danm6/build/tmp/sysroots/x86_64-linux/etc/pear.conf Feb 05 17:32:50 You may want to add: /usr/lib/php to your php.ini include_path Feb 05 17:32:50 Installing header files: /sboxm7/jim/dan5/build/tmp/work/x86_64-poky-linux/php-5.3.6-r2.1/image/usr/include/php/ Feb 05 17:32:53 mv: cannot stat `/sboxm7/jim/dan5/build/tmp/work/x86_64-poky-linux/php-5.3.6-r2.1/image//sboxm7/jim/dan5/build/tmp/sysroots/x86_64-linux//etc': No such file or directory Feb 05 17:32:56 ERROR: Function failed: do_install (see /sboxm7/jim/dan5/build/tmp/work/x86_64-poky-linux/php-5.3.6-r2.1/temp/log.do_install.24746 for further information) Feb 05 17:32:59 I usually find I have to blow away the php sstate files and try again til it works... Feb 05 17:33:01 Has anyone see this or got a fix? Feb 05 17:33:03 thanks Feb 05 17:41:13 JimNH: sounds like broken parallel do_install... you could work around it by doing PARALLEL_MAKEINST = "" in the recipe Feb 05 17:49:15 ok I just tried a PARALLE_MAKE = " " and it failed, will try the MAKEINST flavor now Feb 05 17:54:20 hmm, PARALLEL_MAKEINST gets set from PARALLEL_MAKE by default, so maybe that's not the issue... but if it isn't I'm not sure what it could be Feb 05 17:54:29 the php recipe is a horrible nightmare :( Feb 05 17:55:39 also fails... same way; I suspect its something about it using sstate, and putting into my image dir the path to the sandbox that build the ssate items, where as the error above you see its put a ref in to the path twice.. Feb 05 18:05:29 is there a way in a .bb file to say 'do not use any sstate info' for this recipe? Feb 05 18:06:27 sstate won't be the problem Feb 05 18:10:56 What I see is that my image dir is in a path like: /sboxm7/jim/dan4/build/tmp/work/x86_64-poky-linux/php-5.3.6-r2.1/image, and in there I see a usr dir as I expect, but also a dir path of ssd/jim/danm6/build/tmp/sysroots/x86_64-linux/etc/pear.conf ; /ssd is me building on another system in another pool.. so I had assumed the only thing shared between the sandboxes in sstate... Feb 05 18:14:28 JimNH: I would say don't take the path being repeated under image to necessarily indicate a problem by itself Feb 05 20:27:01 anyone here using smart as a package updater? Feb 05 20:32:41 ftonello: oe-core is transitioning to smart for the 1.4 release Feb 05 21:48:40 How do I create an iso? Feb 05 21:59:11 sgw1: when is suppose to release 1.4? Feb 05 22:03:09 ftonello: April Feb 05 22:06:41 RP: 10x Feb 05 22:32:21 poll: is it more useful to have a subcommand to search recipes/provides, and another to search packages/rprovides/packages_dynamic, or a single command which outputs both, distinguished in its output? I want to avoid subcommand proliferation, but we also want to keep usability, which means we don't want the output to be too complex or confusing Feb 05 22:35:17 kergoth: I vote for both or with option to limit search to one group Feb 05 22:35:39 kergoth: both types in output Feb 05 22:45:34 JaMa: my current command searches both namespaces, and has arguments to specify just one or the other. is that what you're thinking, as long as the output is clear about which is which? Feb 05 23:18:02 kergoth: yes Feb 05 23:19:01 okay, thanks for the input Feb 05 23:19:02 kergoth: sometimes people will be looking just for "python" and it's good to show that there is python recipe, which does not have python package, but have python-core as RPROVIDER python Feb 05 23:19:11 yeah, thats what i was thinking as well Feb 05 23:24:22 JaMa: https://gist.github.com/kergoth/4718162 is the current implementation with short-circuiting to reduce the spammy output. if searches, in order, pn, provides, packages, rprovides, packages_dynamic. without -a, it skips the following steps as soon as it matches. so if 'bash' the recipe matches, it wont show provides, packages, etc Feb 05 23:24:33 seem reasonably sane? Feb 05 23:25:03 sometimes it would be useful to get list of newly added packages from do_split_packages any reason why adding return packages in do_split_packages would be bad idea? Feb 05 23:25:43 hmm, can't think of a reason offhand.. that does seem like it would be handy Feb 05 23:27:56 kergoth: can you show that output when searching for python? (to show how rprovides look?) Feb 05 23:28:11 kergoth: not sure if you saw my attempts to use multiprocess in package.bbclass but we have a problem. It seems python 2.7 older than 2.7.3 has bugs in multiprocess :( Feb 05 23:28:14 kergoth: and how long that list will be (bash surprised me a bit) Feb 05 23:28:37 kergoth: Seems we can hang processes using it, the pool.join() never returns in some cases Feb 05 23:29:13 RP: could it also explain #3742? Feb 05 23:29:41 but that I see also with 2.7.3 Feb 05 23:29:46 JaMa: which python version is it? Feb 05 23:30:09 JaMa: hmm :/ Feb 05 23:30:15 RP: ugh, yes, i remember hitting stuff like that when trying to use a pool for parsing Feb 05 23:30:22 would just inexplicably hang at times Feb 05 23:30:54 kergoth: I just tried wrapping for exception handling but its not working for my test case :/ Feb 05 23:31:06 RP: and it's really annoying.. almost every Ctrl+C I need to open separate shell and kill with -9 Feb 05 23:31:32 RP: the same when someone tries to stop running build from jenkins UI :/ Feb 05 23:31:40 JaMa: we will get it looked at sooner or later if noone beats us to it Feb 05 23:31:54 JaMa: of course, if i have short-circuiting on, and i bb search python, it won't tell me about python-core. would have to use -a or -P.. hmm Feb 05 23:32:09 RP: might be easiest to just roll your own pool the way we do for parsing Feb 05 23:32:15 not pretty, but.. Feb 05 23:32:48 kergoth: and the search pattern is also regexp right? So I can search for ^python$ to hide all those extra python modules and recipes Feb 05 23:33:06 JaMa: yeah, there's options for substring search (default), exact match, wildcards, and regex Feb 05 23:33:14 nice Feb 05 23:33:32 bb search -w -P 'python*' Feb 05 23:33:34 or whatever Feb 05 23:33:55 * kergoth is still debating the output format for optimal effectiveness, not convinced the current behavior is ideal Feb 05 23:34:21 kergoth: ah, I now understand why that code is so ugly :( Feb 05 23:38:48 kergoth: I'm tempted to add the 2.7.3 Pool class code into compat.py Feb 05 23:39:09 assuming I can't reproduce this on 2.7.3 Feb 05 23:39:12 not a bad idea Feb 05 23:39:19 ugly, but hardly less so than rolling all our own stuff Feb 05 23:39:41 kergoth: right Feb 05 23:48:54 kergoth: using pool code from 2.7.3 on the problematic python install doesn't lock up in 500 iterations Feb 05 23:49:30 path of least resistance looks clear... Feb 05 23:54:54 I'd be curious to see if chunking the input to reduce overhead helps at all. It did appear to speed up the parsing process slightly, though it depended on the number of processes, obviously Feb 05 23:55:05 (pool supports chunking built in) Feb 05 23:57:50 * mranostay waves to kergoth Feb 05 23:58:07 hey mranostay Feb 05 23:58:25 kergoth: going to ELC? Feb 06 01:14:19 mranostay: not sure, need to ask the boss about it. would be nice to make it again this year **** ENDING LOGGING AT Wed Feb 06 02:59:58 2013