**** BEGIN LOGGING AT Wed Feb 20 02:59:58 2013 Feb 20 08:06:27 good morning Feb 20 08:09:11 hi mckoan Feb 20 08:28:19 gm florian Feb 20 08:29:46 hi all Feb 20 09:16:04 gm bluelightning Feb 20 09:16:24 morning woglinde, all Feb 20 09:21:24 This software is released under either the GNU Library General Public Feb 20 09:21:24 License (see LICENSE.LGPL) or the BSD License (see LICENSE.BSD). Feb 20 09:21:28 what do i do ? Feb 20 09:21:36 i pick one ? Feb 20 09:22:33 yes Feb 20 09:22:42 ok Feb 20 09:22:52 no, don't you add all 3 licenses? Feb 20 09:22:59 there's many examples of doing so Feb 20 09:23:40 (assuming you're writing a recipe here) Feb 20 09:23:48 (of course) Feb 20 09:24:01 afournier hm or it depends file wise Feb 20 09:24:02 i did not expect it could happen :) Feb 20 09:24:41 hm? Feb 20 09:25:09 afrournier 2 files lgpl 5 files bsd Feb 20 09:25:58 afournier but if you are unsure ask a specialized laywer Feb 20 09:26:48 all files are both BSD and LGPL Feb 20 09:26:53 i'm assuming that the question pertains to writing a recipe for a software package that's multiply-license... in that case, add all the licenses to the recipe Feb 20 09:27:26 LICENSE = "GPLv2 & LGPLv2.1 & MIT" Feb 20 09:27:28 LIC_FILES_CHKSUM = "file://LICENSE.GPL2;md5=751419260aa954499f7abaabaa882bbe \ Feb 20 09:27:30 file://LICENSE.LGPL2.1;md5=4fbd65380cdd255951079008b364516c \ Feb 20 09:27:32 file://LICENSE.MIT;md5=544799d0b492f119fa04641d1b8868ed" Feb 20 09:27:34 something like that Feb 20 09:27:40 well ok thanks jonibo Feb 20 09:28:44 hang on a sec Feb 20 09:29:25 afournier: jonibo: if it's OR (i.e. you can choose the license) you need to use | instead of & Feb 20 09:29:59 no, it doesn't matter if you're only writing a recipe... only if you are adding code and need to make an explicit choice Feb 20 09:30:10 no, it does matter Feb 20 09:30:16 it's & until you combine it with something that dictates otherwise Feb 20 09:30:33 if you link it to something else then it becomes important, and that can happen after that recipe is written and built Feb 20 09:30:43 or rather, if you link something else to it Feb 20 09:30:46 good point Feb 20 09:30:59 ok Feb 20 09:31:14 guess it should be "|" then Feb 20 09:31:15 LICENSE = "LGPLv2.1 | MIT" is not supported than :) Feb 20 09:31:20 ? Feb 20 09:32:00 ? Feb 20 09:32:55 afournier: that syntax is correct yes Feb 20 09:33:12 afournier: if it matches what the license documentation for the software being built says, then it's also correct... Feb 20 09:33:14 i think it's more correct like this then Feb 20 09:33:24 because they give the choice in the license Feb 20 09:33:33 right, then that should be the right value Feb 20 11:08:15 configure.in:134: warning: AC_LANG_CONFTEST: no AC_LANG_SOURCE call detected in body Feb 20 11:08:17 any idea ? Feb 20 11:16:40 I'm no autotools expert I'm afraid, hopefully someone else can help Feb 20 11:16:54 i am fighting with unixodbc Feb 20 11:17:28 there's a reference to that here though: http://www.flameeyes.eu/autotools-mythbuster/forwardporting/autoconf.html Feb 20 11:17:51 thanks Feb 20 11:18:04 btw, it works if i do "do_configure () { autoconf oe_runconf }", that's ugly Feb 20 11:18:35 then i added "inherit binconfig" but the odbc_config is a binary not a script :/ Feb 20 11:56:26 morning all Feb 20 12:43:26 hi pb Feb 20 12:50:23 hi woglinde Feb 20 12:53:10 libtoolize: `config/compile' not found in `/oe/tmp-eglibc/sysroots/alix2d/usr/share/libtool' Feb 20 12:53:11 grrr Feb 20 12:53:26 http://patches.openembedded.org/patch/15065/ Feb 20 13:50:26 I'm having some difficulty getting a systemd timer to work, I think something is up with the recipe could anyone point me to a recipe with a systemd timer defined? May help show where I am going wrong Feb 20 14:52:23 yay ! i finally understood how to recipefy unixodbc => do_configure_prepend() { rm -rf ${S}/m4 } Feb 20 15:13:01 hey. is there a way to tell bitbake not to cache files in a certain recipe? Feb 20 15:15:05 for instance SRC_URI="blahblah;don'tcachethisone" Feb 20 15:15:21 or to at least force its retrieval from source URL each time Feb 20 15:17:14 CMoH : python do_fetch_prepend() and remove the cached version Feb 20 15:19:19 afournier, thanks Feb 20 15:19:54 is there a cache API, like "bb.cache.remove(this-file)" ? Feb 20 15:20:15 no idea Feb 20 15:21:44 thanks - on to reading bitbake's internals then Feb 20 15:27:13 so I can get my systemd timer to work by loading it manually just not on boot, have no issues with a normal service, what am I missing Feb 20 15:31:24 apparently the bb.fetch2.Fetch can be queried to obtain the paths to the files that need to be removed Feb 20 15:39:49 are those SYSROOT_PREPROCESS_FUNCS executed before do_install ? Feb 20 15:43:28 never mind Feb 20 16:38:58 http://www.h-online.com/open/news/item/D-Bus-is-coming-to-the-Linux-Kernel-1801485.html Feb 20 16:39:01 uhuh Feb 20 16:42:10 afournier: if it has sane userspace library then I welcome it with open arms Feb 20 16:42:39 afournier: attempting to use dbus and glib the other day was horrific Feb 20 16:43:30 the time will tell Feb 20 16:44:06 afournier: I'm looking forward to it, I would love to see a simple multicast message bus Feb 20 16:44:16 if this will be it; who knows... Feb 20 20:19:13 kergoth, did you make it to elc this year? Feb 20 20:20:07 unfortunately not this year Feb 20 20:23:48 oh russ here Feb 20 20:28:55 I know OE classic is outdated, but what is the minimum I need to copy of my tmp dir between two machines (same exact path, username, OS, etc). I don't want to have to recompile anything. Feb 20 20:30:28 ? Feb 20 20:44:08 I'm trying to create a new VMware build machine identical to my current one and I'd like to copy only the minimum of files from my OE classic environment as possible. Feb 20 20:45:08 This is so I can trim down the size of the VM which grows with time. I've since found a better method of creating a second virtual disk for tmp which I can delete and re-create without having to make a new VM. Feb 20 20:49:57 o.O Feb 20 20:50:04 just copy the v Feb 20 20:50:05 m Feb 20 20:50:11 thats what vm's is for Feb 20 20:52:41 it's 100GB Feb 20 20:52:53 I'm sending it to India Feb 20 20:53:10 great! Feb 20 20:53:49 I'd like to get the size down to something that is faster than DHL Feb 20 20:54:34 I would use BZ2 Feb 20 20:55:13 I figured I'd let xz work on it overnight. Feb 20 20:55:15 I would use the shrink modus Feb 20 20:55:21 before Feb 20 20:55:27 ? Feb 20 21:00:26 woglinde: "shrink modus"? Feb 20 21:01:49 yes vmware has a disktool like virtualbox Feb 20 21:01:56 where you can shrink the vm Feb 20 21:05:36 Ah, my experience with that tool is that it only works well with Windows VMs. Feb 20 21:08:46 Are sources only cached in tmp/deploy/glibc/sources? Feb 20 21:08:59 hello all Feb 20 21:09:05 I need to make sure I don't ship some package sources. Feb 20 21:09:15 sr105 no in work too Feb 20 21:09:22 right Feb 20 21:09:26 there are sources i nthe -dbg binary packages also, for source level debugging Feb 20 21:09:28 if you do not have inherit rm_work Feb 20 21:09:29 I've cleaned that out. Thanks. Feb 20 21:10:52 Looks like a lot of duplicated data between deploy/glibc/ipk, pstage, and sys roots…. Feb 20 21:11:20 somewhat, but they're used for different purposes, and the build couldn't very well link to files inside of ipks ;) Feb 20 21:12:12 oe-lite, which is an oe based thing that split off and changed in a lot of ways implemented their binary caching using binary packages, and removed the build dependency vs runtime dependency thing in favor of just runtime, so with that, there is no separate sstate archive. has its advantages and disadvantages, though Feb 20 21:12:37 ok, just wanted to make sure it wasn't like estate-cache where things would just be automatically regenerated. Feb 20 21:13:10 estate-cache is what I meant before auto-correct took over Feb 20 21:13:16 sstate-cache Feb 20 21:13:47 I guess I've trimmed as much as I can then. Thanks. Feb 20 21:49:45 is there a class to re-path .la in sysroot ? i saw binconfig does it, but it also copy the *-config ... Feb 20 21:55:47 dont we remove the la files? Feb 20 21:55:57 i dunno Feb 20 21:56:47 i don't think so as there are so many in my sysroot Feb 20 21:57:11 but it's not a reference :) Feb 20 21:58:37 trying to rebuild without the problematic .la Feb 20 21:58:43 looks like some things moved around on the mirrors, * opkg_download: Failed to download http://feeds.angstrom-distribution.org/feeds/core/ipk/eglibc/all/task-sdk-target_1.0-r0_all.ipk, wget returned 1. Feb 20 21:59:27 I looked at the url and it stops at eglibc... what would the right one be? Feb 20 22:00:03 woglinde: also according to binconfig.class we do not erase .la we fix them Feb 20 22:07:35 What is the difference in openmembedded.git and meta-oe.git? It looks like there is a recipe for the package I want (NUT) in openembedded.git, but it isn't in meta-oe (oe-core, understandably from what i understand of the structure). I'm working with the angstrom setup scripts, which are configuring a variety on layers for me, in case it matters. Feb 20 22:08:32 My guess would be that openembedded.git is oe classic and it just means no one has bothered bringing in all of the recipes into the new oe structure, is that correct? Feb 20 22:08:58 sounds correct to me Feb 20 22:11:00 ok, so the quick and dirty would be to grab the folder for the package and add it to my own custom layer and the proper way would be to submit it as a patch for meta-oe, right? That said, the proper way might leave me waiting for a while for inclusion Feb 20 22:11:43 moving a recipe from oe-classic to oe-core needs some tweaking most of the time Feb 20 22:11:57 you'd better make your own layer and rewrite that recipe Feb 20 22:12:14 if you think it should be in oe-core then why not submitting a patch Feb 20 22:12:26 that's how i would do, but i can be wrong Feb 20 22:13:03 I've already got my layer, so step one is taken care of, but if I'm submitting a patch I should add it to my local meta-oe, then make the patch, no? Feb 20 22:13:07 alors check oe-contrib Feb 20 22:13:17 yes Feb 20 22:13:23 s/alors/also/ Feb 20 22:14:25 i don't see an oe-contrib repo Feb 20 22:15:46 meta-openembedded-contrib Feb 20 22:16:03 sorry openembedded-core-contrib Feb 20 22:17:02 i am not yet to confortable with all those repos, i just use openembedded-core Feb 20 22:19:01 I'm still not finding it. oe-core/meta/recipes-support, oe-core/meta/recipes-extended, meta-oe/meta-oe/recipes-support, and meta-oe/meta-oe/recipes-supportextended all look like potential homes for it Feb 20 22:19:28 i did not mean it was there, i meants it could be there somewhere Feb 20 22:19:32 it's really not all that difficult to pull a oe classic recipe to oe-core. make sure it uses new style staging (no do_stage task) and add LIC_FILES_CHKSUM, that's 90% of it Feb 20 22:19:44 spacecolonyone: random recipes tend to go to meta-oe, but only if there's no better place for it Feb 20 22:20:09 kergoth, looks like your the author of the one I'm looking at porting Feb 20 22:20:19 (NUT) Feb 20 22:21:03 heh, i used to sometimes just grab an interesting project on freshmeat and add a recipe for it, even if nobody needed it yet Feb 20 22:21:10 clearly i had too much time on my hands Feb 20 22:21:14 heh Feb 20 22:21:30 well hopefully your effort will make my life easier Feb 20 22:22:03 i thought oe-core was supposed to be the bare minimum not somewhere to put recipe we don't know where to put... Feb 20 22:23:26 alright, well since the angstrom setup-scripts draw from HEAD of the denzil branch, I'll look at adding it there in oe-core/meta/recipes-support on mine Feb 20 22:23:51 I'm not 100% that the setup scripts will handle my local commits (I think they try to) Feb 20 22:23:56 wakko: it is, yes Feb 20 22:24:01 meta-oe is the random recipe pile Feb 20 22:24:03 if so then I'll try making a patch and submitting it Feb 20 22:24:32 otherwise I'll stick it in my layer and mention i ported it on the mailing list Feb 20 22:24:43 oe-core is minimum. of course, one could look at it and wonder how sato is minimum, but a certain amount of stuff exists there for testability/demos Feb 20 22:24:48 heh Feb 20 22:25:04 :/ Feb 20 22:26:06 hmm, kergoth do you recall anything about the libm patch that is in there (I know, a total shot in the dark) Feb 20 22:26:06 demos can be part of the "minimum" if they are specified as essential Feb 20 22:26:31 I've built it by hand on my BBxM and I didn't patch anything Feb 20 22:26:31 not clue, sorry to say, been too long Feb 20 22:26:36 s/not/not a/ Feb 20 22:27:18 it is adding -lm to bcmxcp_usb_LDADD = $(LDADD_DRIVERS) $(LIBUSB_LDFLAGS) Feb 20 22:27:30 does that mean anything? Feb 20 22:27:57 a linker argument for the broadcom driver? Feb 20 22:28:37 I'd say drop it and if it builds, leave it out Feb 20 22:28:42 see if it fails without it Feb 20 22:29:12 it's not because -lm is on the command line that it is used for linking, i think it's not Feb 20 22:29:55 yea, Thats what I'm going to try it inherits autotools so I don't need to worry about the do_stage and I'll go google LIC_FILES_CHKSUM Feb 20 22:31:13 LIC_FILES_CHKSUM is used to record the copyright headers / license text files state as checksums (md5), so that if upstream changes their licensing, our builds fail and we know we have to update LICENSE Feb 20 22:31:27 ensures LICENSE matches reality, or at least that someone has explicitly checked Feb 20 22:31:41 is there a way to bypass ;) Feb 20 22:32:18 the only gross hack that exists is to claim it's not open source (LICENSE=CLOSED), but no one will ever accept it upstream, and the license would obviously be incorrect Feb 20 22:41:49 alright, thatnks **** ENDING LOGGING AT Thu Feb 21 02:59:59 2013