**** BEGIN LOGGING AT Wed Apr 24 02:59:58 2013 Apr 24 07:24:02 Good morning Apr 24 07:48:24 morning apelete Apr 24 07:52:06 morning all Apr 24 07:57:39 morning silvio_ Apr 24 07:59:24 hi silviof Apr 24 08:11:17 updating to dany 8.0.1 simply means "git pull", right? Apr 24 08:11:49 * eren is a bit out-of-date :/ Apr 24 08:17:30 morning eren, all Apr 24 08:18:43 morning bluelightning Apr 24 08:21:34 bluelightning: I failed to find the wiki page of "release name -> version" Apr 24 08:21:59 one sec Apr 24 08:22:10 eren: https://wiki.yoctoproject.org/wiki/Releases Apr 24 08:22:48 bluelightning: oh thanks, I searched the wiki with "release names", "release numbers" keywords but no luck :) Apr 24 08:26:45 mediawiki's search function is infuriating at times :/ Apr 24 08:27:04 yes Apr 24 08:27:08 total useless Apr 24 08:28:27 hi woglinde Apr 24 08:28:38 I just added some keywords to that page, it does now show up but at the end of the search Apr 24 08:29:35 bluelightning: which keywords? Apr 24 08:30:52 it doesn't show up with "release names", or "release numbers" Apr 24 08:32:55 eren`: are you perhaps searching the OE wiki? Apr 24 08:33:19 maybe I should add a "Yocto Project release names" page on the OE wiki with a pointer to the above Yocto Project wiki page Apr 24 08:33:33 hi bluelightning , woglinde Apr 24 08:33:54 bluelightning: I don't think so -> https://wiki.yoctoproject.org/wiki/index.php?title=Special%3ASearch&redirs=0&search=release+numbers&fulltext=Search&ns0=1 Apr 24 08:35:02 eren`: ah, I neglected to add "numbers" - refresh and you should see it at the end now Apr 24 08:35:23 bluelightning: oh, thank you Apr 24 08:45:53 Hi silviof, silvio_, bluelightning Apr 24 08:47:07 hi apelete Apr 24 08:48:18 what's the purpose of STAGING_DIR_NATIVE? I'm reading oe-commits and PHP package has "mv ${D}/${STAGING_DIR_NATIVE}/${sysconfdir}/* ${D}/${sysconfdir}/" Apr 24 08:48:56 why are php configuration files are installed in ${D}/${STAGING_DIR_NATIVE} in the first place and moved to ${D}/${sysconfdir}? Apr 24 08:50:14 hi apelete Apr 24 08:50:17 hi silvio_ Apr 24 08:50:38 eren`: the php recipe is full of badness :( Apr 24 08:50:47 hi apelete Apr 24 08:50:53 probably one of the messier recipes we have Apr 24 08:51:23 bluelightning: oh ok, I'm skipping it then Apr 24 08:53:19 bluelightning: however, I couldn't understand the purpose of STAGING_DIR Apr 24 08:53:53 normally, packages are installed in its directories, e.g ${D}, which is a temporary directory Apr 24 08:54:17 afterwards, they should be moved to somewhere to be processed by {deb, ipk, rpm} ? Apr 24 08:54:23 eren`: it doesn't make a huge amount of sense to me either; I suspect that directory is leaking into the php build script Apr 24 08:55:08 I wonder how the modphp recipe I updated for meta-webserver compares in that regard Apr 24 08:55:19 at some point I hope to unify the two recipes Apr 24 08:55:25 or at least have a common inc file Apr 24 08:55:54 as far as I remember, we compiled php 2 times to produce php binary, php-cli and mod_php Apr 24 08:56:14 I mean, back in I was busy with our distribution here Apr 24 08:56:43 we created a dir, "say php", and did "../configure ......" , "../make" Apr 24 08:56:50 for php-cli, same process Apr 24 09:06:08 need to go, see you all! Apr 24 09:31:55 bluelightning: ping Apr 24 09:33:05 bluelightning: is it possible that those parsing errors were produced when bblayers.conf had both meta-webos and meta-webos-ports included? Apr 24 09:33:31 JaMa: could be; it'll enable whatever is specified in the list of layer dependencies Apr 24 09:35:30 bluelightning: ok, I've updated layer description, as it doesn't depend on meta-webos anymore Apr 24 09:35:39 let me know if those errors are still there Apr 24 09:38:00 JaMa: ok, will do, thanks Apr 24 09:38:10 bluelightning: hi. I sent a patch for the uicmoc issue we talked about a few days ago (don't know if you remember, I've been busy with work lately) Apr 24 09:38:26 apelete: yep, saw that; will apply that this evening - thanks Apr 24 09:40:26 bluelightning: I hope it won't break something on your side, I don't know if the issue is specific to my setup or not Apr 24 09:40:43 apelete: the patch looks OK to me, FWIW Apr 24 09:40:54 ok, cool Apr 24 09:44:27 I try to do 2 packages of the same software, one develop and one release, I manage them with the option ' DEFAULT_PREFERENCE = "-2" ' inside the develop pakages which I called foo_git.bb while the release package is called foo_0.1.0.bb Apr 24 09:45:36 I think that is not the right way to do things, because I have an error "ERROR: Package version for package qscpecg-dbg went backwards which would break package feeds from (0:git-r0 to 0:0.1.0-r0)" Apr 24 09:46:12 How I must do to manage situation in right way to OE? Apr 24 09:47:01 woglinde: does meta-java need qemu for builds? Apr 24 09:47:23 woglinde: I see one patch adds qemu calls and another one that removes some Apr 24 09:49:28 silvio_: if you don't care about maintaining feeds that will upgrade properly from the target you can safely ignore that error Apr 24 09:50:17 silvio_: it's just telling you that the new package placed in the feed is older than the previous one and therefore will not be installed by the package manager on upgrade Apr 24 09:52:04 bluelightning, I need feeds that maintaing upgrade properly, which is the best practise? Apr 24 09:52:57 silvio_: in that case, assuming the output of the build is provided directly as feeds pointed to by the package manager on the target, don't do what you just did Apr 24 09:53:14 silvio_: package versions can't go backwards in that case, the package manager won't handle it automatically Apr 24 09:53:46 silvio_: you can bump PE in the case where upstream changes the version in such a way that the version effectively goes backwards, but that's a different situation Apr 24 09:59:03 bluelightning, sorry, what means "bump PE"? Apr 24 10:00:25 silvio_: http://www.yoctoproject.org/docs/1.4/ref-manual/ref-manual.html#var-PE Apr 24 10:00:37 bluelightning, thanks Apr 24 10:00:45 silvio_: by "bump" I mean "increase" e.g. PE = "1" since the default is "0" Apr 24 10:05:21 bluelightning, so pls correct me if i lost something, I want 2 branch o develop: "Devel" on git with foo_git.bb and "Release" on a server with a "foo.tar.gz" of the tagged recipe, so if I want only things from the "Release" branch I have to put inside bb the PE set to 1? Apr 24 10:06:22 bluelightning, I mean if I want that opkg manage in right way the "Release" , while "devel" package could be forced by hand Apr 24 10:07:19 silvio_: I would suggest having separate feeds to handle that if you need to; you'd need to manage the feeds by hand Apr 24 10:07:26 silvio_: PE is not going to help for this Apr 24 10:09:51 bluelightning, u mean change feed inside bb each time? Apr 24 10:14:28 silvio_: I mean, instead of just copying the entire contents of tmp/deploy/ipk|rpm|deb to a server, you have some scripts to do that and split out the development versions to a separate feed Apr 24 10:14:55 silvio_: you could make it easier by keeping separate distro configs for dev/release which select the preferred versions in each case Apr 24 10:20:35 yay, my university allocated a server for OE build Apr 24 10:20:47 I will see their capabilities today Apr 24 10:20:59 I hope they allocated at least 8 core and 8GB ram Apr 24 10:25:07 Hello. I would like to patch an existing package using a .bbappend file found in another bblayer, but I noticed that the my patch cannot find the desired file to patch unless I add ../ to my patch file which is not an elegant way... How can I fix that ? Apr 24 10:26:21 vadmeste: you probably need to set an appropriate striplevel= Apr 24 10:26:55 vadmeste: http://www.yoctoproject.org/docs/latest/ref-manual/ref-manual.html#var-SRC_URI Apr 24 10:29:25 bluelightning: cool thanks ^^ Apr 24 10:36:06 suihkulokki hm normaly we need qemu because of the sizer stuff Apr 24 10:36:21 and yes the patches could be reworked Apr 24 10:41:48 bluelightning, with "keeping separate distro configs for dev/release " u mean have two separate recipes to build the image? Apr 24 10:42:15 silvio_: no, I mean separate conf/distro/distroname.conf Apr 24 10:43:52 bluelighting is it better to use in PACKAGES = "${PN}-something" too and FILES_${PN}-something ? Apr 24 10:54:58 woglinde: yes, if you mean ${PN}-something instead of foo-something when PN = "foo" Apr 24 10:55:50 bluelightning, I m sorry it is not clear to me Apr 24 10:56:28 I now have qscpecg_0.1.0.bb and qscpecg_git.bb Apr 24 10:57:07 woglinde, U mean inside the distro recipe? Apr 24 10:58:36 morning all Apr 24 10:59:05 hi pb_ Apr 24 11:01:15 silvio_: the dev distro config file would have PREFERRED_VERSION_qspecg = "git" (or whatever PV is set to) and the release config would have PREFERRED_VERSION_qspecg = "0.1.0" Apr 24 11:01:32 hi pb_ Apr 24 11:04:58 hi florian Apr 24 11:05:06 bluelightning, thanks Apr 24 11:05:23 hi silvio_, florian, bluelightning Apr 24 11:29:06 gm pb Apr 24 11:37:39 ~curse libxml-parser-perl Apr 24 11:37:41 May the fleas of a thousand camels infest your most sensitive regions, libxml-parser-perl ! Apr 24 11:40:00 hrw: lol Apr 24 11:40:03 ups we have ibot Apr 24 11:40:24 ~curse ibot Apr 24 11:40:26 May you be reincarnated as a Windows XP administrator, ibot ! Apr 24 11:40:42 curses list is short Apr 24 11:45:10 Hi guys I am doing a project similar to implementing http://arduino.cc/en/Reference/SD .. But my prof said this is not quite enough work to get good grades. Please tell me how can I add something extra in it. I am not too much familiar with embedded devices. Apr 24 11:45:51 ? Apr 24 11:46:29 wooo_ type /topic and understand what you read than Apr 24 11:46:52 woglinde: how is your jenkins setup going? Apr 24 11:47:06 hi woglinde Apr 24 11:47:09 hi zecke Apr 24 11:47:17 zecke hm still no images test Apr 24 11:47:34 who broke devshell? Apr 24 11:47:34 but building is okay now Apr 24 11:47:43 hrw git blame Apr 24 11:49:03 woglinde: how do you plan to test images? Apr 24 11:53:26 hrw qemu Apr 24 11:53:51 and than ssh into it and run some tests Apr 24 11:54:35 woglinde: or run with stdio serial and do tests on serial Apr 24 12:00:18 JaMa: do you know why http://patches.openembedded.org/patch/17039/ was rejected? Apr 24 12:12:11 hrw yes Apr 24 12:12:31 hrw but ssh might be easier Apr 24 12:13:16 hm. how does libxml-parser-perl work with OE toolchain and fail with external... Apr 24 12:13:42 hrw compiling? Apr 24 12:14:36 woglinde: it compiles simple test for expat.h header (which is present) and then tries to run binary Apr 24 12:15:11 will have to check with OE toolchain... Apr 24 12:15:15 do we need the test? Apr 24 12:15:16 be back in ~30 minutes Apr 24 12:19:42 pb_: IIRC the only concern was that it creates new path which should be tested Apr 24 12:20:25 pb_: and from my perspective, path with sstate-cache archive not existing is well tested and quite common, so only not creating them is "new" Apr 24 12:46:13 what's the difference in bitbake recipes between "PACKAGES" and "PROVIDES"? Apr 24 12:47:40 those are two different things Apr 24 12:48:06 i can't find any documentation describing them Apr 24 12:48:13 i mean, i can, but it's pretty vague Apr 24 12:48:30 PACKAGES lists binary packages produces by recipe Apr 24 12:48:49 PROVIDES is an optional list of things which are provided by recipe Apr 24 12:48:50 so they can't be used as dependencies? Apr 24 12:49:21 thanks hrw for your answer, so i'm more relax to provide mine. on mainstream distro (debian & co) one package is binary executable and they provide functionality (like a shell) Apr 24 12:49:21 inter-recipes dependencies (aka build dependencies) are constructed from recipe names and providers names Apr 24 12:49:51 what about runtime dependencies? Apr 24 12:50:07 are those based on PACKAGES or PROVIDES? Apr 24 12:50:10 staticshock_: runtime dependenceis are in PACAKGES Apr 24 12:50:15 typo Apr 24 12:50:43 ok Apr 24 12:50:49 thanks Apr 24 12:50:50 staticshock_: note that packages which contain libraries can be renamed to follow Debian style of naming Apr 24 12:51:04 yeah, i noticed Apr 24 12:51:08 so libz1 instead of zlib1 Apr 24 12:51:22 i saw it done in external-toolchain-csl Apr 24 12:51:26 hm. it was wrong example ;d Apr 24 12:51:41 staticshock_: external toolchains can be evil ; Apr 24 12:51:41 glibc -> libc6 Apr 24 12:52:23 yeah. fair enough. Apr 24 12:54:33 can i run oe for my desktop computer ? Apr 24 12:56:07 remide: yes, sure Apr 24 12:56:36 remide: I started using OE on pc which was slower then devboards I use now Apr 24 12:57:17 ok, so i can and you do ? Apr 24 12:58:09 you know other people ? that don't look to be common Apr 24 12:58:21 you use os core ? or other branch Apr 24 12:58:55 remide: just do a build. in worst case it will take more time Apr 24 12:59:12 remide: you do not have athlon-xp, don't you? Apr 24 13:01:06 no, i hav already build oe. poky maybe ? but the doc speak a lot about dev, not just using day to day Apr 24 13:01:40 so i come here to don't be alone ^^ Apr 24 13:01:59 remide: what do you mean by 'using day to day'? Apr 24 13:02:52 using OE on a desktop is fine, and if you have a slower desktop, OE can prevent you even from RSI! Apr 24 13:03:21 just like debian Apr 24 13:03:33 I guess for update it's not so easy Apr 24 13:04:34 remide: ah, you mean 'using OE derived distribution on desktop PC'? Apr 24 13:05:01 remide: cause OE is buildsystem not distribution Apr 24 13:05:44 i look for somethings like netbsd, a base with hardware support and gcc Apr 24 13:05:47 remide: why do you want to use OE to build a desktop distribution? Would you not be better off using something like Gentoo? Apr 24 13:06:07 in oe there is sdk but their is a lot of useless stuff for me Apr 24 13:09:06 i already used gentoo. oe seems to be more interesting to understand Apr 24 13:09:41 remide: I sugest you then to grab an x86 image of angstrom and take a look Apr 24 13:09:42 i have little laptop and a big desktop computer will build system fot it Apr 24 13:10:03 ok, you suggest angstrom Apr 24 13:10:25 why ? that's difficult to choose, there is so many Apr 24 13:13:06 because its alphabetically seen the logic point to start ;)(and that it's quite well maintained helps too) Apr 24 13:13:09 remide: many users, good distro maintainers Apr 24 13:17:18 good, i will try Apr 24 13:17:24 thanks Apr 24 13:17:25 by Apr 24 13:26:46 mm, you're quite a diehard if you want to use angstrom / any OE derived distribution on your desktop... Apr 24 13:33:32 tsjsieb: who knows, maybe he will be even more hardcore and will build angstrom under angstrom Apr 24 13:36:13 ok. cpan_do_configure now... Apr 24 13:36:21 hrw: which of course is not too difficult Apr 24 13:38:58 heh... Apr 24 13:39:03 He can make it a little bit more difficult if he's cross compiling on a ARM device for a X86 Apr 24 13:39:25 But I would start my own kernel at first, if I was him! Apr 24 13:39:53 Get al the unnecessary stuf out! ;) Apr 24 13:40:08 http://pastebin.com/MeHKRKUi is a fix for my issue Apr 24 13:40:57 I prefer not to check who had idea that "sed -e s/something/something with a/path/in/it/probably/g" will work Apr 24 13:42:02 hrw: yes, it's a bit sad that / was picked as the common character to use there in all examples given the range of choice availabe Apr 24 13:42:05 available Apr 24 13:42:45 I admit that '|' is maybe not best choice either but works in this case Apr 24 13:43:54 I usually use ! Apr 24 13:44:20 of course there's an outside chance that ! could appear in a path Apr 24 13:45:49 patch sent to OECore ML Apr 24 13:46:42 looks like I am one step from eing able to build SDK images with external Linaro toolchain Apr 24 13:46:58 just need to kill eglibc-mtrace requirement Apr 24 14:01:02 what variable specifies what targetfs formats should be built, e.g. .tar.gz .tar.xz Apr 24 14:01:37 jackmitchell: IMAGE_FSTYPES Apr 24 14:02:53 bluelightning: can I override it directly in my distro conf, or does it need to be in local conf? Apr 24 14:03:20 jackmitchell: it can be in either of those places Apr 24 14:04:05 in some cases it is useful to set at the machine level if the machine supports or prefers certain means of booting Apr 24 14:05:06 bluelightning: yes it is currently set in MACHINE as an _append, which means I get tar.gz and tar.xz which makes the do_rootfs take twice as long as it strictly needs to be Apr 24 14:08:24 jackmitchell: I think the recommendation now is to use += in the machine config so that it can be overridden elsewhere Apr 24 14:08:42 (specifically for IMAGE_FSTYPES that is) Apr 24 15:04:49 JaMa: thanks. Apr 24 16:23:26 is there an openembedded/haiku project? Apr 24 16:29:01 staticshock_: sorry, can you explain what you mean? Apr 24 16:32:16 moin Apr 24 16:36:19 eh, i retract my question. i'm not really sure what i'm asking. Apr 24 16:43:15 me too... i fingered it out... Apr 24 17:19:30 hello, is there any reason why bitbake wouldn't automatically build my MLO when I try to build an image? Apr 24 17:19:36 I get u-boot, just no MLO Apr 24 17:21:22 hello all Apr 24 17:21:26 I noticed it after I upgraded frome denzil->danny Apr 24 17:25:06 jawilson: i haven't built an MLO before that i was aware of, but maybe there is a different make argument that generates the MLO Apr 24 17:25:58 if that is true, then you would probably need a do_compile_append() function that runs the make arg to generate the MLO Apr 24 17:26:29 well, "bitbake my-image-name" used to build the kernel, u-boot, x-load (my MLO), and my RFS and put it all in the images directory Apr 24 17:26:58 after upgrade, I only get u-boot, and RFS Apr 24 17:27:39 running "bitbake x-load" builds it and puts in the image directory, but that's beside the point, I'm wondering why it is no longer happening automagically Apr 24 17:30:16 from which version did you upgrade? Apr 24 17:30:40 from denzil branch to danny branch Apr 24 17:35:19 hmm, I'm thinking it may be something in meta-ti Apr 24 17:39:34 jawilson: most of the platforms in meta-ti switched from x-load to u-boot/SPL, thus MLO is provided by u-boot. what's your machine? Apr 24 17:40:32 denix yup, just found that commit, that's exactly what's wrong. It's an atlas module from iVeia (OMAP DM3730), my own custom BSP that depends upon meta-ti Apr 24 17:44:04 are you still using x-load with it? Apr 24 17:44:43 yeah, by BSP I just mean OE BSP layer, I'm using all their code Apr 24 17:49:16 jawilson: no, I meant is your board using x-load for MLO or u-boot/SPL? Apr 24 17:51:18 it's using x-load Apr 24 17:52:58 hmm, ok Apr 24 20:43:53 is it possible to provide opkg recommendations in bitbake recipes? Apr 24 20:44:12 RRECOMMENDS Apr 24 20:45:02 thanks Apr 24 20:46:07 does RRECOMMENDS stuff get used for build-time dependencies? Apr 24 20:46:17 i'm sorry, bad phrasing Apr 24 20:46:27 does RRECOMMENDS stuff get built? Apr 24 21:24:06 staticsh1ck: yes they are built before do_package is executed **** ENDING LOGGING AT Thu Apr 25 02:59:59 2013