**** BEGIN LOGGING AT Wed Jun 12 02:59:58 2013 Jun 12 05:58:44 is it possible to do shallow git fetches with oe? Jun 12 05:58:46 JaMa: thanks. It works :) Jun 12 07:23:55 I'm jumping from denzil to dylan. Did something happen to the kernel-modules package? Jun 12 07:29:22 how should spaces in FILES_${PN} be escaped? Backslash? Jun 12 07:46:21 good morning Jun 12 07:51:02 morning Jun 12 07:53:34 looking at populate_packages() in package.bbclass it seems I'm out of luck with escaping spaces, I might be able to work around it by replace " " with "*" Jun 12 08:11:25 morning all Jun 12 08:24:49 dylan isnt using gcc 4.8, is it? Jun 12 08:26:21 no, 4.7.2 Jun 12 08:26:56 will the dylan successor use 4.8 only? or will 4.7 and 4.8 somehow coexist? this is relevant, because if it will use 4.8 only, then I can add new flags that were introduced in 4.8 to some tune variables for a BSP layer Jun 12 08:28:11 that I am not sure of Jun 12 08:28:47 one key thing will be if the current issues with 4.8 are resolved with 4.8.1 Jun 12 09:00:32 hi JaMa, as discussed yesterday, i am starting to use meta-qt5, i am trying to build the examples ( i remove the -nomake examples), and i get a lof of QA errors like this: http://pastebin.com/3B9iPUZZ. Jun 12 09:03:20 well, apparently, the -examples packages aren't managed yet... Jun 12 09:03:26 that probably explains why. Jun 12 09:26:59 ndec: they are in master branch, feel free to backport it and send it after testing Jun 12 09:28:07 JaMa: argh... thanks.. i was actually redoing it.. as i missed them in master... Jun 12 09:33:57 JaMa: right now examples aren't built by defaut, will you change that at some point? or are we expected to override this policy in a .bbappend? Jun 12 09:35:28 similarly opengl is 'hardcoded' as opposed to gles/eglfs. so how is it recommended that we pick gles? is there going to be a 'FEATURE' to pick, or do we have to customize in .bbappend? Jun 12 09:36:11 stupid question : how can i append day month and year as cflags into a recipe ? Jun 12 09:39:13 rob_w: I'd guess you could add to CFLAGS an expression that included a python call i.e. ${@...} Jun 12 09:39:40 the python call can do almost anything; you could define your own function to return the appropriate values Jun 12 09:40:01 hmm right Jun 12 09:40:07 python is not my strong side .. but thanks Jun 12 09:40:53 the other alternative would be to do a do_configure_prepend or do_compile_prepend (depending on how CFLAGS is picked up for the piece of software) and do it in shell script Jun 12 09:42:49 ndec: changing gles/eglfs is described in recipe Jun 12 09:43:00 ndec: no strong opinion about enabling examples by default Jun 12 09:43:45 i saw that. so you expect anyone who need eglfs to have their own .bbappend, instead of making it a 'feature', right? Jun 12 09:46:04 rob_w: it doesn't have to be really complicated python. Something like this should work, http://pastebin.com/m92WD2Ta Jun 12 09:46:14 right Jun 12 09:46:35 thx ... trying it Jun 12 09:46:58 indeed, that was the sort of thing I was thinking about Jun 12 09:47:16 rob_w: you'll have to adapt it so myextracflags has the format you want of course Jun 12 09:48:17 sure Jun 12 09:49:15 looking good Jun 12 09:49:16 thx Jun 12 09:49:28 rob_w: btw, if you want the recipe to rebuild every time the date changes you may also need to do CFLAGS[vardeps] += "DATE" Jun 12 09:50:01 depending on the situation of course Jun 12 09:50:26 no actually not Jun 12 09:50:51 ok, no worries then :) Jun 12 10:00:13 hi all Jun 12 10:01:24 hi pb_ Jun 12 10:13:41 hi pb_ Jun 12 10:15:28 pb_: I have an increscious problem with a Makefile Jun 12 10:18:02 pb_: does --no-undefined really matter in case of -static? Jun 12 10:21:26 JaMa: the qt5 example patch from master applies cleanly on dylan. however the -examples package includes lots of stuff, including all the source code. is that on purpose? Jun 12 10:26:24 ndec: at least in qt4, qt-demo shows corresponding source code when running an example so I guess it is required Jun 12 10:26:49 ok. Jun 12 12:34:18 help dehilight Jun 12 12:34:24 sorry. Jun 12 12:49:28 Hi Jun 12 13:01:33 ndec: yes the question is more about whether example binaries should be there or packaged separately Jun 12 13:02:05 JaMa: yes, indeed. Jun 12 13:05:11 JaMa: btw, i just finished building/testing a rebase of master on top of dylan in meta-qt5, and it all looks good. Jun 12 13:05:31 i am testing with qemuarm, QPA=minimal,and testing qtbase only. Jun 12 13:05:52 but looking at the branch it makes sense to merge everything in dylan. Jun 12 13:13:54 ndec: I agree, I'm also using master with dylan build, I just wanted to keep dylan branch pointing to more tested state for a while Jun 12 13:38:57 Regarding bitbake subversion fetch. Apparently the --no-interactive flag makes svn not use the cached credentials that I actually have. Do you have the same challenge, and what do you do about it. I'm not very keen on having my username/password in recipe or env-var. Jun 12 14:21:10 SorenHolm: My svn times are way back - but you can save your svn credentials with a "test co" on console, than you have a .svn directory in you home with your credentials ... **** ENDING LOGGING AT Wed Jun 12 16:48:41 2013 **** BEGIN LOGGING AT Wed Jun 12 16:50:19 2013 **** ENDING LOGGING AT Wed Jun 12 16:52:05 2013 **** BEGIN LOGGING AT Wed Jun 12 16:54:22 2013 Jun 12 17:16:53 anybody got a feel for percentage of whole-disk formatted USB storage devices are out in the wild? Jun 12 19:02:57 mr_science: That pseudo error went away when I changed the branch to dylan Jun 12 19:10:49 okay, at least it's not on that branch... Jun 12 19:23:15 dbfi: were you building off master or danny yesterday? Jun 12 20:19:33 silviof: I know. But the current svn version does not use the store credentials when run i non-interactive mode. Oh well - I'll just move the stuff to a git repo.. :) Jun 12 20:20:40 How can I remove a part of a variable in a bbappend-file. Stuff like this does not seem to work. Jun 12 20:20:42 dirs755 := "${@'${dirs755}'.replace('${localstatedir}/run', '')}" Jun 12 20:21:07 SorenHolm: might b eable to store it in your .subversion/config, though i expect that would be a security problem. alternatively, if it's an http/https uri, you might be able to use a .netrc, i use it for http git repos Jun 12 20:21:24 I dno't see anything obviously wrong with that line. what do you mean by 'does not seem to work'? Jun 12 20:21:52 of course, that could remove bits you don't intend, the less error prone method would be to split it, remove the entry from the list, and re-concatenate Jun 12 20:21:59 or just use oe_filter_out(), which does all that for you Jun 12 20:22:05 grep around, there are exampels of its use Jun 12 20:22:12 kergoth: "bitbake -e base-files | grep dirs755" does not reflect the change. Jun 12 20:22:34 well, first make sure your bbappend is being applied at all, go from there :) Jun 12 20:22:46 best to just copy and paste an example of oe_filter_out use, though Jun 12 20:23:54 kergoth: I'll try oe_filter_out... thanks. Jun 12 20:24:41 that's also what i was looking for... Jun 12 20:25:30 mr_science: It was master branch yesterday Jun 12 20:25:49 but it turns out some things like udev and dbus don't like /run being volatile, while other things like straight non-managed dhcp don't like cruft being left in /run either... Jun 12 20:26:52 so my current work-around is to test/delete the dhcp..pid in the network startup script Jun 12 20:27:18 unless someone else has a better idea... Jun 12 20:27:27 kergoth ^^ Jun 12 20:28:11 kergoth: hmmm oe_filter_out works, but only if I store the result in another variable. Putting the modified content back into dirs755 does not work. Jun 12 20:28:55 dbfi: that's a little odd since i haven't seen it Jun 12 20:29:09 and i've been building off master for the last few weeks Jun 12 20:29:25 with several pulls along the way... Jun 12 20:30:29 my guess is it was broken/fixed in between my pulls and you just happened to hit it at the wrong time Jun 12 20:31:26 SorenHolm: well, people use it for exactly that case all the time, without a temporary variable Jun 12 20:31:30 don't know what to tell you Jun 12 20:31:48 mr_science: Maybe that's the case Jun 12 20:31:57 i use it in 13 bbappends in meta-mentor alone Jun 12 20:32:50 kergoth: hmmm maybe some layer precedence then... Anyway thanks for the oe_filter_out hint. :) Jun 12 20:34:19 it's possible, definitely examine bitbake -e to see the order of operations Jun 12 20:34:37 <3 seebs' variable tracking patches Jun 12 20:48:25 kergoth: doh ... my own very minimal, but almighty layer had a priority of 1. Thanks for the hints. I did not know that operations are visible in bitbake -e.. Jun 12 20:48:43 ah :) that'd explain it, glad you got it figured out Jun 12 20:49:00 yeah, bitbake -e is invaluable, both for examining global configuration metadata and per-recipe Jun 12 20:49:13 really good for checking assumptions, making things are what we think they are Jun 12 20:49:30 indeed, thanks to Peter & Richard for that **** ENDING LOGGING AT Thu Jun 13 02:59:59 2013