**** BEGIN LOGGING AT Tue Jun 25 02:59:58 2013 Jun 25 06:21:56 hi guys, is there any good packages to develop an embedded web server on ARM platform ? Jun 25 06:59:04 ll Jun 25 09:37:11 morning all Jun 25 09:50:13 mks: I use hiawatha, which is a Web Server; and I use it on ARM, therefore an Embedded Web Server? If you want to develop a web server then you don't really need any packages just get going in C and eglibc Jun 25 10:10:31 JaMa: I'm trying to put together a recipe for nodejs 0.10.2, and have an openssl issue. https://groups.google.com/forum/#!msg/nodejs/D6Xzd4OHV-o/ZJMfJcLLLvMJ mentions it, and also mentions you :) Jun 25 10:23:18 tasslehoff: why not start with 0.10.4 recipe I've sent earlier? Jun 25 10:23:49 tasslehoff: only thing missing is to preserve 0.8 version in recipe with different name for gateone to work Jun 25 10:24:05 JaMa: ah, didn't see that one Jun 25 10:26:18 JaMa, where do we stand on the gnuradio patch? Jun 25 10:26:38 can you override PARALLEL_MAKE on a per recipe basis from local.conf? Jun 25 10:28:30 Crofton|work: not yet Jun 25 10:29:07 waiting for more changes before I start new builds Jun 25 10:29:29 k Jun 25 10:29:33 sorry for the pain Jun 25 10:29:54 JaMa: same issue there. Jun 25 10:31:42 JaMa: could this be related to the path length? Jun 25 10:32:44 I have my tree in /home/tas/angstrom-setup-scripts, and from 1.3 to 1.4 I think maybe some other path components have become longer as well. Jun 25 10:32:50 yes looks like it Jun 25 10:33:50 locally I have just /OE but on jenkins I have even longer TOPDIR and it was building fine Jun 25 10:34:00 locally gentoo on jenkins ubuntu Jun 25 10:35:44 If I rename my folder to oe, there is no way to get out of a complete rebuild? Jun 25 10:39:30 you can reuse sstate-cache Jun 25 10:40:20 but it would be better to find out which command is failing with execve Jun 25 10:40:20 JaMa: ok. will test that then. Jun 25 10:40:28 ok. will test that then :) Jun 25 10:41:50 Hello. I am having a problem creating a new bitbake recipe. I am trying to run a Makefile.PL file after extracting the package, but it seems like bb is unable to locate said Makefile.PL. This is my recipe: http://pastebin.com/uLUzmcLd and this is the error: http://pastebin.com/n1fU8dhr Jun 25 10:42:02 I have been working on making recipes for a number of CPAN modules I need, I will publish them once I get them actually working. Jun 25 10:42:38 JaMa: not sure how, though. add some verbosity when building? Jun 25 10:43:48 tasslehoff: run with -j1 and when it fails heck first command on make -n Jun 25 10:43:57 Stygia: you have MakeFile.PL in recipe, but the file is called Makefile.PL Jun 25 10:44:23 erbo, Dear gods do I feel foolish now. You are right.... Jun 25 10:44:45 Stygia: the brain usually reads what it expects :) Jun 25 10:45:03 Stygia: btw, you might want to try inherit cpan Jun 25 10:45:05 erbo, Yea, it does that quite well. :) Fair point Now I feel slightly less foolish. Jun 25 10:45:52 bencoh, Oh? We are using Yocto, AFAIK there may be a CPAN in there, but as CPAN installs are somewhat... error-prone at times, we prefer individually installing the packages we need (as opposed to running cpan install). Jun 25 10:46:12 Stygia: cpan is a bb class Jun 25 10:46:51 bencoh, So, what I would set as "SECTION"? I am not super familiar with BB, I am trying to get some of my perl scripts to run, which led to this. Jun 25 10:46:51 (available both in oe-core/meta and the old meta-openembedded) Jun 25 10:47:28 Stygia: http://pastebin.notk.org/pastebin.php?show=d2efac295 as an example Jun 25 10:48:01 bencoh, What would that accomplish for me? Jun 25 10:48:20 Not having to actually build the package? Jun 25 10:48:39 Stygia: it's a recipe Jun 25 10:48:57 classes are here to help you write recipes : Jun 25 10:48:58 :) Jun 25 10:49:19 Stygia: the bbclass adds the do_configure and do_install stuff so you don't need to do it manually Jun 25 10:49:34 ergo, Ah, just as I expected. That's certainly useful. Jun 25 10:49:36 Stygia: look in poky/meta/classes/cpan.bbclass to see what it does Jun 25 10:49:47 meta/recipes-devtools/perl/libxml-simple-perl_2.20.bb would be an example recipe Jun 25 10:50:51 bluelightning, Noted. Jun 25 10:51:02 This looks interesting. Would make it pretty easy to add CPAN modules. Jun 25 10:51:24 Does anyone have recommendations for where I should make this available when I finish? Plan to just put it on my public github. Jun 25 10:51:45 JaMa: http://pastebin.com/ndM9Vk9i Jun 25 10:55:21 Stygia: hmm... normally I'd suggest meta-oe, but I wonder if this would help us better test perl in OE-Core Jun 25 10:55:35 bluelightning, Interestingly enough, after modeling my recipe on this example (http://pastebin.com/SG3JGsPw), it still complains about Makefile.PL being unavailable - it is no longer mentioned in my recipe at all. Jun 25 10:56:09 Stygia: you'll need ${PV} not {PV} in your value for S Jun 25 10:56:33 bluelightning, Yea, I just noticed. Damn, this keeps throwing me off with these scripts. Jun 25 10:56:36 bluelightning, Sorry. Jun 25 10:58:31 Stygia: you can drop the DEPENDS now you're inheriting cpan btw, and PR doesn't need to be set to "r0" since that's the default (we're slowly tidying these up in existing recipes) Jun 25 10:58:36 Stygia: also if your recipe has _3.28 in the name you won't need to set PV Jun 25 10:59:01 bluelightning, Hmm, noted. I will slim them down, then. Jun 25 11:00:53 are rprovides and rdepends of perl packages calculated by the build systems or must they specified manually? Jun 25 11:00:54 bluelightning, Which section do you suggest? Jun 25 11:01:13 ensc|w: I'm pretty sure it's manual Jun 25 11:01:31 tasslehoff: is it from rm or ar? Jun 25 11:01:40 ensc|w: I know kergoth is working on automatic runtime dependencies for python, but AFAIK nothing exists for perl Jun 25 11:01:51 does it fail when you run it in your shell manually? works fine here Jun 25 11:02:30 Stygia: you mean the value of SECTION ? if so, it's not critical Jun 25 11:02:46 tasslehoff: well rm is short so probably ar Jun 25 11:03:56 ensc|w: manual Jun 25 11:04:00 bluelightning: ok; last time when I packaged a perl program, it was an horror to get the deps correctly (especially due to different spellings). fwiw, rpm contains tools calculating perl deps/provides Jun 25 11:04:07 JaMa: not following. are you looking at the rm in the clean target? Jun 25 11:04:12 (I ran into deps hell, so ... definitely manual :p) Jun 25 11:04:27 ensc|w: indeed, I had the same pain constructing a recipe for webmin Jun 25 11:04:48 ensc|w: yes, we would probably be best advised to interface with that Jun 25 11:05:35 ensc|w: the only problem is we can only add runtime dependencies during packaging, which means if they haven't also been built (and they won't be automatically) then you'll get failures when you attempt to include those packages in an image Jun 25 11:06:19 (well, I mean we can only add these kinds of runtime dependencies that are calculated based on the source/compiled code at packaging time) Jun 25 11:06:50 tasslehoff: I'm looking at rm in your paste Jun 25 11:07:04 yes; we need some rules to express PROVIDES of perl packages in a common way Jun 25 11:07:24 (e.g. libxml-simple-perl vs. perl-xml-simple) Jun 25 11:08:07 RPROVIDES, yes Jun 25 11:08:16 that still won't ensure the dependencies are built though Jun 25 11:08:33 JaMa: ahh, didn't notice the ar in there until now Jun 25 11:08:48 ensc|w: the only way for that to work ultimately is for those dependencies to eventually get added back into the recipe Jun 25 11:09:33 JaMa: works fine in my devshell Jun 25 11:09:55 tasslehoff: check https://bugs.launchpad.net/ubuntu/+source/binutils/+bug/1043507 Jun 25 11:10:32 bluelightning: no; I think a common naming in PROVIDES would be enough for build time dependencies. The RPROVIDES and RDEPENDS can be calculated automatically. Jun 25 11:11:22 ensc|w: the problem is when can you determine these? if it's not at parse time bitbake cannot know that they need to be built Jun 25 11:11:42 tasslehoff: maybe you'll need to fix your distro make https://github.com/adobe/webkit/blob/master/Tools/gtk/patches/make-3.82-arg-list-length.patch Jun 25 11:13:30 JaMa: that so it could seem. thanks. Jun 25 11:13:54 https://savannah.gnu.org/bugs/?36451 Jun 25 11:15:23 bluelightning: it must be done by the packages; for DEPENDS, he has to read the README. The PROVIDES are more complicated indeed. For simple perl packages, they are trivial because they contain only one item. Jun 25 11:25:06 JaMa: which version of make do you have? didn't you say it worked on your jenkins running ubuntu? Jun 25 11:25:30 3.82 with patches Jun 25 11:26:05 yes ubuntu, but I have no idea what your distro is Jun 25 11:26:23 JaMa: ubuntu 13.04 Jun 25 11:26:26 64 bit Jun 25 11:27:41 tasslehoff: have you tried to move this 1 command to separate Makefile to test this theory? Jun 25 11:29:36 tasslehoff: looks like my TOPDIR on jenkins is actually a bit shorter then yours Jun 25 11:29:50 /home/jenkins/oe/shr-core Jun 25 11:31:01 JaMa: ok. I'll probably try a rebuild soon. Jun 25 11:31:11 haven't found the line in the Makefile Jun 25 11:32:58 hmm Jun 25 11:33:17 you can use the one you copy paster Jun 25 11:37:33 JaMa: it runs fine in its own makefile Jun 25 12:36:02 JaMa: worked with /home/tas/oe Jun 25 15:04:42 xchat crash Jun 25 15:04:44 grr Jun 25 15:45:37 ka6sox-away: what's the URL for the repo? Jun 25 15:45:52 Crofton|work: hey, you know the URL for the advocacy repo? Jun 25 15:47:00 denix: same question that I asked Crofton|work ^^^? Jun 25 15:47:28 no Jun 25 16:06:30 hi. i am trying to understand PREMIRRORS vs SOURCE_MIRROR_URL, and I don't quite understand why we have both. does anyone know when we should use one vs the other? Jun 25 16:07:18 ndec: own-mirrors.bbclass adds SOURCE_MIRROR_URL to PREMIRRORS. that's all it is, a convenience, to avoid having to set up your premirrors manually for a simple, common case Jun 25 16:08:07 ok, i see. but own-mirror doesn't seem to prepend, does that mean that it overrides? Jun 25 16:09:18 it does, but if you use _append or _prepend to alter PREMIRRORS, that will happen after own-mirrors is parsed, as append/prepend is postponed to the end of the parse process Jun 25 16:11:09 ok, thanks. Jun 25 16:54:23 Crofton|work: confirmed access to the repo... Jun 25 16:54:25 see email. Jun 25 16:54:46 Crofton|work: do we have a current list of action items somewhere? Jun 25 21:35:21 what's the appropriate procedure to request a change be cherry-picked onto dylan? Jun 25 21:35:48 joshc: read README Jun 25 21:36:06 JaMa: that's perhaps where I should have started, of course. thanks. Jun 25 21:48:24 JaMa, I am not sure that is clear Jun 25 21:49:34 gitweb seems to suggest koen for meta-oe? Jun 25 21:50:44 inspecting the commit log suggest ericben Jun 25 21:53:40 --subject-prefix=meta-oe][dylan][PATCH is important part adding ericben to CC is good too Jun 25 21:59:57 where woudl I find that so I can say the right thing :) Jun 25 22:00:18 maybe my gitweb interface usage is poor? Jun 25 22:00:44 Paul is doing the oe-core stuff, right? Jun 25 22:00:54 http://git.openembedded.org/meta-openembedded/tree/meta-oe/README?h=dylan#n10 Jun 25 22:01:03 right Jun 25 22:01:23 for Paul for oe-core/dylan, Ross for oe-core/danny, eric for meta-oe both Jun 25 22:02:41 gah, OK, I wasn't reading carefully enough Jun 25 22:03:00 we should include the branch maintainer info in the README's Jun 25 22:03:38 is sending the patch the best way to request a cherry-pick, or should we send the git hash? Jun 25 22:07:02 both work when the hash does apply cleanly Jun 25 22:17:15 JaMa, Crofton|work: thanks guys Jun 25 22:17:24 np Jun 25 22:17:46 I'm trying to make sure I can answer the question clearly next time Jun 25 22:18:11 especially since I nag the poky crowd about not being clear a bunch :) Jun 25 22:19:30 oe sucks everyone in eventually Jun 25 22:20:18 we must achieve world domination Jun 25 22:29:11 i'd settle for beer domination... Jun 25 22:37:51 df -h Jun 25 22:37:55 ups Jun 25 22:39:41 with load average: 33.31, 30.59, 29.11 it's easy to type faster, then screen refresh after ^A P :) **** ENDING LOGGING AT Wed Jun 26 02:59:58 2013