**** BEGIN LOGGING AT Sun Feb 21 02:59:57 2010 Feb 21 11:16:20 03Martin Jansa  07org.openembedded.dev * r9a0b80fa46 10openembedded.git/recipes/xorg-lib/pixman_0.17.6.bb: Feb 21 11:16:20 pixman: disable gtk, to break circular dependency gtk+->cairo->pixman->gtk+ and also libtool issue Feb 21 11:16:20 Signed-off-by: Martin Jansa Feb 21 12:09:24 Ainulindale: vous iciĀ ?! Feb 21 12:09:52 Corsac: depuis longtemps Feb 21 12:44:59 ~seen florian Feb 21 12:45:02 florian <~fuchs@Maemo/community/contributor/florian> was last seen on IRC in channel #oe, 1d 17h 24m 27s ago, saying: 're'. Feb 21 13:36:12 what about using DEPENDS if DEPENDS_virtclass-native is empty for a BBCLASSEXTEND="native" package? Wouldn't that be more natural and avoid adding alot of DEPENDS_virtclass-native ? Feb 21 13:36:59 blindvt: thats how it already works Feb 21 13:37:11 blindvt: it just add -native to everything in DEPENDS line Feb 21 13:37:43 XorA, my native.bbclass doesn't seem to do that? Feb 21 13:39:52 XorA, consider this git-1.7.0 package: http://paste.debian.net/60774/ Feb 21 13:40:17 DEPENDS_virtclass-native ?= "${ORIG_DEPENDS}" Feb 21 13:40:22 in native.bbclass Feb 21 13:40:57 newdeps.append(dep + "-native") Feb 21 13:41:01 in native.bbclass Feb 21 13:41:08 XorA, hm, you're right. Feb 21 13:41:14 hehe Feb 21 13:42:30 the only thing I think it should do but doesnt is when the non native depends on ${PN}-native, that should dissapear when it matches in the -native version. At the moment it causes bb parse to quit because of infinite loop Feb 21 13:50:35 if i ask it to bake git-native it picks git-native_1.6.0.4.bb and not the -1.7.0. Huh? provider-picking never mentions the git_1.7.0.bb as candidate provider for git-native ? Feb 21 13:53:29 only seems to be set to 1.6.0.4 for openmoko Feb 21 13:54:39 XorA, yea, and i'm playing with micro Feb 21 13:56:48 not sure why that is happening, Ive seen it work right in other recipes I did Feb 21 13:57:56 it's quite likely that i goofed that package, but my errors isn't immediately obvious to me.. It looks pretty standard Feb 21 13:58:29 s/isn/aren/ Feb 21 14:13:47 if i wipe the git-native_1.6.0.4.bb then it works as expected. Sounds like it has a problem handling virtual:native (at least if a non-virtual -- but older, in this case -- provider exists, or something like that) Feb 21 14:15:40 for reference: Feb 21 14:15:41 has-native.log:DEBUG: adding /recipes/git/git-native_1.6.0.4.bb to satisfy git-native Feb 21 14:15:42 just-1.7.0.log:DEBUG: adding virtual:native:/recipes/git/git_1.7.0.bb to satisfy git-native Feb 21 14:22:07 hi all Feb 21 14:59:13 pb___, using SRCREV = "${AUTOREV}" doesn't work either; It suggests that 'FetchData' object has no attribute 'basecmd' while parsing Feb 21 15:16:29 hi mickey|sofa Feb 21 15:16:35 hi florian Feb 21 15:16:46 is a spam checker running on the new ltg? Feb 21 15:16:54 i have noticed an increasing spam coming through my ltg acconut Feb 21 15:17:15 +amount of Feb 21 15:21:15 mickey|sofa: yes we have sa and use greylisting. Feb 21 15:21:26 hmm, ok Feb 21 15:22:28 I currely see quite some pams passing all filters but most are very unsepecific and hard to filter. Feb 21 15:25:09 btw., what happened to the OEDEM reimbursements? Feb 21 15:29:16 I think we finished processing these end of December. Feb 21 15:29:56 The next step for the financial department is declaring taxes. But not before Embedded World. Feb 21 15:31:05 hmm Feb 21 15:31:10 i'm talking about the member reimbursements Feb 21 15:31:36 i was somewhat expecting to find the money on my account during the last couple of weeks Feb 21 15:31:42 that's why i'm asking Feb 21 15:33:37 good point... I thought I processed all but I have to check the bank account anyway. Feb 21 15:34:18 ok Feb 21 15:38:09 * florian feels a little bit stressed currently Feb 21 15:38:33 don't worry, nothing too important, i just want to make sure it doesn't get forgotten Feb 21 15:38:48 good morning Laibsch Feb 21 15:38:58 hi Feb 21 15:39:02 good morning Feb 21 15:39:26 Enjoying the sofa? Feb 21 15:39:39 definitely Feb 21 15:39:47 have to catch up with mail backlog Feb 21 15:47:24 03Andrea Adami  07org.openembedded.dev * r072d533f58 10openembedded.git/recipes/klibc/ (27 files in 3 dirs): Feb 21 15:47:24 klibc: major patch split and reordering, recipe cleaning. Feb 21 15:47:24 * update order of patches to include upstream changes Feb 21 15:47:24 * bump PR for all three recipes Feb 21 15:47:24 * fix typo in klibc_1.5 about 'sh' Feb 21 15:47:25 * clean tabs Feb 21 15:47:25 Signed-off-by: Andrea Adami Feb 21 16:18:58 re Feb 21 16:33:16 i discovered that xinit uses mcookie if available, that means that if someone builds util-linux-ng before building xinit, there is a dependency but if util-linux-ng is not build before there is not (because then /dev/random or so is used); what would be the way to fix this? let xinit depend on util-linux-ng ? Feb 21 16:38:13 eFfeM: is there a --disable-mcookie? Feb 21 16:42:07 SRCREV is driving me nuts. How can i just use HEAD and not some arbitrary version? Feb 21 16:42:53 blindvt: set it to ${AUTOREV} Feb 21 16:43:45 XorA, no, that doesn't work (for me): 2 hours ago i said: using SRCREV = "${AUTOREV}" doesn't work either; It suggests that 'FetchData' object has no attribute 'basecmd' while parsing Feb 21 16:44:16 blindvt: what version of bitbake? Feb 21 16:44:29 XorA, master Feb 21 16:44:42 blindvt: well thats your problem then, use the correct version Feb 21 16:45:01 XorA, and the correct version would be which one? Feb 21 16:45:17 XorA: will check Feb 21 16:45:17 blindvt: the released version, 1.8.18 at the moment I think Feb 21 16:46:01 blindvt: that might explain your git oddness as well Feb 21 16:46:49 XorA, that's very well possible, yes. ugh. *curses* Feb 21 16:52:05 XorA: there does not seem to be a configure option; either we need to add a dependency or override MCOOKIE=$ac_cv_path_MCOOKIE Feb 21 16:52:51 eFfeM: add the dependency then Feb 21 16:53:03 eFfeM: it sounds like it was meant to be used in the general case Feb 21 16:56:24 XorA: yes, think so, actually thinking of it, this is somewhat a general issue as lots of packages run configure to determine what to use and if there is no explicit dependency then it depends on what you did before, whether or not a specific option is selected Feb 21 16:57:00 eFfeM: it is an issue we are well aware of Feb 21 16:57:21 eFfeM: you notice the mplayer recipes EXTRA_OECONF is huge Feb 21 16:58:10 XorA: some of us are aware, others are not, I just moved from group2 to group 1 Feb 21 16:58:50 eFfeM: then you are ideal to write a wiki entry :D Feb 21 16:58:54 btw has PE a meaning in name generation (like PR): xinit recipe has a PE="1" in the recipe Feb 21 16:59:02 XorA: :-) Feb 21 16:59:25 not too sure about location (actually was considering mailing to the list instead) Feb 21 16:59:27 its the epoch so package is xinit-1:${PV} Feb 21 16:59:53 versioning is debian style Feb 21 16:59:53 ah ok Feb 21 17:00:22 epoch is used where there is no sane upgrade path for PV Feb 21 17:00:36 which happened when xf86 became xorg Feb 21 17:03:33 XorA, using bitbake from the distro made AUTOREV work, but it still picks DEBUG: sorted providers for git-native are: ['.../recipes/git/git-native_1.6.0.4.bb'] and nothing more, so bitbake-1.8.18-2 is also not an option. But thanks for the hint. I'll keep the 1.8.x installed, just in case Feb 21 17:14:08 XorA: guess the ultimate solution to the configure issue is to use only those packages that are mentioned in DEPENDS (recursively ofc). guess this could be done with packaged staging Feb 21 17:14:53 eFfeM: we did have a discussion on that Feb 21 17:15:05 eFfeM: Im sure RP would love to see an implementation Feb 21 17:15:12 :-) Feb 21 17:15:36 i recall vaguely that there was a discussion but not really the outcome Feb 21 17:15:39 * XorA looks into whether it could be done using some form of unionfs, but that requires root access or delicate glibc hacking :-( Feb 21 17:16:03 don't think my knowledge of the build system is good enough to actually implement this. Feb 21 17:17:09 another (also not very nice) solution would be to give each package its own dir in staging and use explicit -I and -L flags, but that could lead to very long command lines Feb 21 17:17:26 and fixing every busted autoconf out there Feb 21 17:17:34 which is hundreds of thousands Feb 21 17:17:58 03Frans Meulenbroeks  07org.openembedded.dev * r7f6cec6f0a 10openembedded.git/recipes/xorg-app/xinit_1.2.0.bb: (log message trimmed) Feb 21 17:17:58 xinit: add dependency on util-linux-ng Feb 21 17:17:58 if util-linux-ng is build before xinit, xinit will use mcookie (exported by util-linux-ng) Feb 21 17:17:58 Howwever if util-linux-ng is not build before xinit it will not use mcookie. Feb 21 17:17:58 This leads to an undesired scenario. E.g. the version of xinit on the angstrom feed needed Feb 21 17:17:58 actually that packaged staging bitty would make a good GSoC project Feb 21 17:17:59 mcookie, but as my system did not have util-linux-ng installed, startx failed. Feb 21 17:18:00 By making the dependency explicit at least we get consisency and apparently using mcookie Feb 21 17:18:04 hm, i always call it autohell :-( Feb 21 17:19:51 good call :-) Feb 21 17:20:23 * XorA especially hates people who edit configure script in release tarballs Feb 21 17:20:41 yup Feb 21 19:17:03 echo mem > /sys/power/state Feb 21 19:17:34 #echo mem > /sys/power/state Feb 21 19:18:12 sorry, wrong screen Feb 21 19:18:49 morning Feb 21 19:26:19 . Feb 21 20:26:38 re Feb 21 20:34:48 hi florian you've got an imx_somenumber and want to make vpu work right? Feb 21 20:35:51 GNUtoo: yes right, that's an i.MX51x Feb 21 20:36:09 ok what's missing exactly? some addresses? Feb 21 20:37:02 because I'd like to make my i.MX31 work too Feb 21 20:37:59 would be cool to stream h264 Feb 21 20:38:20 because mjpeg is not very good,theora uses too much cpu Feb 21 20:38:28 s/good/efficent/ Feb 21 20:38:40 GNUtoo: no the driver in the kernel I got from the manufacturer did not work as well as the one from freescale. Feb 21 20:38:56 ah ok,mine doesn't even compile Feb 21 20:39:10 missing addresses that I found in some patches over the net Feb 21 20:39:25 GNUtoo: With the freescale driver I managed to play a few frames. Feb 21 20:39:32 ah ok nice Feb 21 20:40:24 But the VPU in the i.MX3x should be quite similar to the i.MX51x ones but supporting less codecs. Feb 21 20:40:29 ok Feb 21 20:40:45 are the userland tools and kernel driver floss? Feb 21 20:42:21 kernel should be Feb 21 20:42:36 as I looked into the sources.... Feb 21 20:42:58 but for instance for TI there are a lot of different things Feb 21 20:43:10 dsp-link,dsp-bridge and a third one Feb 21 21:16:51 * mwester mutters swear words about mfpr -- again. Feb 21 21:28:53 khem: ping? Feb 21 21:45:30 evening all. Hit a problem with building meta-toolchain. Specifically ncurses-sdk. I'm getting lots of dangling symlink warnings then an inability to find libc.so.6. Anyone have any ideas? Feb 21 22:02:26 anyone using OE with a gumstix overo? Feb 21 22:05:00 bch132, seem that they use oe by default....but I'll go to bed now...sorry Feb 21 22:05:23 bch132, sort off :0 Feb 21 22:10:21 yeah, they use oe by default Feb 21 22:12:08 I have it all installed and building everything for the gumstix, but wondering how to add packages to the root fs by default Feb 21 22:13:11 right now I just do "bitbake omap3-console-image", but it doesn't include some packages that I want Feb 21 22:13:34 bch132, take the omap3-console-image.bb file, copy it, and add the packages you want Feb 21 22:16:23 grg - say I want libjpeg, is there any specific place I should put it in the bb file? Feb 21 22:17:13 or should I just add it to TOOLS_INSTALL or whatever Feb 21 22:23:00 bch132, i dont have a copy of the .bb file in question, so i'm just guessing, but.... it probably has an IMAGE_INSTALL var with a bunch of stuff listed? (or it includes a file which does this) Feb 21 22:24:13 grg - thanks, yeah, that's what it looks like, so it probably doesn't matter too much Feb 21 22:24:30 you can also add stuff from your local.conf. I forget the directive for that though; check the oe manual. Feb 21 22:30:28 grg, any updates on opkg lately? (I've been out of the country for weeks and haven't been on IRC to watch any commits) Feb 21 22:31:02 mwester, some. Feb 21 22:31:21 mwester, best bet is to look at the commit logs Feb 21 22:31:38 mwester, i plan on making another release in the next hour or so Feb 21 22:32:25 Any changes wrt to the small-memory uncompression stuff (fork/vfork etc)? Feb 21 22:32:34 nope Feb 21 22:32:36 sorry Feb 21 22:32:43 :( Oh well. Feb 21 22:33:10 how about i add a configure flag to enable that in a #defined section? Feb 21 22:33:36 (that should be an #ifdef'd section) Feb 21 22:34:07 I was hoping to look at that this weekend, but I can't get the toolchain to build anyway right now. I'll happily settle for a configure flag; I'd still need a slugos-specific recipe, but at least I could catch up to the current revisions. Feb 21 22:35:00 i suspect you would have fewer problems with low memory systems oom killing stuff now anyway. The memory use of opkg is about 35% of what it was 12 months ago Feb 21 22:37:12 It's high on my list of things to work on, if my life ever returns to something vaguely resembling normal. Feb 21 22:37:21 anyone else seen issues with ncurses-sdk do_package? Feb 21 22:38:34 jic23, I can't even get that far anymore! Feb 21 22:38:52 "ld: -lc not found" <--- Ouch. Feb 21 22:39:22 youch Feb 21 22:48:05 mwester, sudo yum install glibc-static Feb 21 22:48:28 er - seriously? Feb 21 22:48:31 on the host? Feb 21 22:48:37 I think so Feb 21 22:48:49 * mwester goes to check something... Feb 21 22:48:53 that seems to be needed for fedora systems Feb 21 22:49:43 must be a new requirement then. :( Feb 21 22:50:18 recent Fedora stopped installing it Feb 21 22:52:56 hehe! How about that? My 64-bit system doesn't list glibc-static, but there it is in the feeds for the 32-bit system (but not installed). And OE builds just fine on the 64-bit system but not on the 32-bit one. Feb 21 22:53:21 64-bit system is Fedora 10. Feb 21 22:53:29 32-bit system is Fedora 12 Feb 21 22:53:36 -static only came in in fedora 11 Feb 21 22:53:44 if I recall right Feb 21 22:54:00 I skipped Fedora 11 Feb 21 22:54:03 *carries on clutching at straws* anyone want to guess about: http://www.pastebin.com/d67e09177 Feb 21 22:54:27 * XorA has fone 10->ubuntu->11->12 :-) Feb 21 22:54:31 gone Feb 21 22:54:34 But last weekend I could build SlugOS and the packages feeds on F12 just fine, so something must have changed that now needs to link statically. Feb 21 22:54:51 I tried ubuntu, and couldn't see what the excitement was all about :( Feb 21 22:55:11 on all my machines ubuntu is so broken as to be unusable Feb 21 22:55:39 my ubuntu laptop isn't exactly working perfectly.... Feb 21 22:55:51 * XorA doesnt exactly have exotic hardware either Feb 21 22:56:08 does more or less work fine, but then the laptop is a good 5 years old :) Feb 21 22:57:32 I first used it on the new netbook -- I found the netbook-specific version of ubuntu to be less-than-expected, and I was rather surprised to find that Fedora 12 actually had better support for the networking hardware than did the highly-acclaimed netbook edition of ubuntu. Feb 21 22:58:32 hehe Feb 21 22:58:37 ubuntu does just fine on the eeepc 701, but yet again, it might have age on it's side :) Feb 21 22:58:47 * mwester notes that "yum install glibc-static" is going to also upgrade the other glibc components on the host, and wonders what might happen to an OE build-in-progress. Feb 21 22:59:00 jic23, eeepc 1005 Feb 21 22:59:08 * XorA has a 901 Feb 21 22:59:20 fitted with a high speed SSD Feb 21 22:59:24 nice Feb 21 22:59:38 I went for "cheap" :( Feb 21 22:59:38 high speed SSD best money Ive spent on gadget in ages Feb 21 23:00:13 XorA, what time is it there? Feb 21 23:00:19 11pm Feb 22 00:56:01 where does one set the linux kernel version in oe? **** ENDING LOGGING AT Mon Feb 22 02:59:58 2010