**** BEGIN LOGGING AT Tue Jun 08 02:59:56 2010 Jun 08 03:00:03 03Khem Raj  07org.openembedded.dev * r48b5a0e69d 10openembedded.git/recipes/gcc/ (3 files in 2 dirs): Jun 08 03:00:03 gcc-4.5: Fix libstdc++ build for ARM Jun 08 03:00:03 * This patch makes sure that __cxa_end_cleanup is Jun 08 03:00:03 emitted in .text section. Jun 08 03:00:03 * Bump SRCREV to latest Jun 08 03:00:03 Signed-off-by: Khem Raj Jun 08 03:06:22 the url is : http://pastebin.com/CvGUpGGi Jun 08 03:07:01 this my first time to connect this server,and I am a new player,so... Jun 08 03:22:37 k Jun 08 03:22:38 lookin Jun 08 03:22:40 g Jun 08 03:23:05 joe123: Can you update again? I think that might be fixed Jun 08 03:23:28 ok,I do it now Jun 08 03:24:13 NEW URL is : http://pastebin.com/H5c1d6vq Jun 08 05:00:12 03Khem Raj  07org.openembedded.dev * raecc2961cc 10openembedded.git/recipes/uclibc/uclibc_git.bb: Jun 08 05:00:12 uclibc_git.bb: Bump SRCREV to latest master. Jun 08 05:00:12 Signed-off-by: Khem Raj Jun 08 05:39:29 khem: should I try latest gcc SRCREV with OPTSPACE enabled, or should be OPTSPACE removed forever (as you once suggested?) Jun 08 06:55:33 03Martin Jansa  07org.openembedded.dev * r14c265b107 10openembedded.git/recipes/ttf-fonts/ (ttf-dejavu_2.23.bb ttf.inc): Jun 08 06:55:33 ttf, ttf-dejavu: fix-up after RDEPNDS_PN global change Jun 08 06:55:33 Signed-off-by: Martin Jansa Jun 08 06:55:43 03Martin Jansa  07org.openembedded.dev * r55f37fa103 10openembedded.git/recipes/linux/ (2 files in 2 dirs): Jun 08 06:55:44 linux-openmoko-2.6.32: add patch for building with gcc-4.5.0 (backport from linux_2.6.34 patch) Jun 08 06:55:44 Signed-off-by: Martin Jansa Jun 08 06:55:44 03Martin Jansa  07org.openembedded.dev * rbea19a4d74 10openembedded.git/recipes/linux/ (linux-openmoko-2.6.34/pcf.patch linux-openmoko-2.6.34_git.bb): Jun 08 06:55:44 linux-openmoko-2.6.34: add slow resume fix #2337 Jun 08 06:55:44 Signed-off-by: Martin Jansa Jun 08 07:20:29 morning Jun 08 07:26:18 morning Jun 08 07:55:04 Hi Jun 08 07:55:13 Is OE supported on blackfin processors ? Jun 08 07:55:35 yes Jun 08 07:55:47 but it hasnt been maintained for a while Jun 08 07:55:54 very few people with blackfins Jun 08 07:56:12 I cant even remeber the machine name Jun 08 08:03:11 morning Jun 08 08:04:24 Im not going Jun 08 08:16:20 adsp-bf537-stamp adzs-bf538f-ezlite adzs-bf548-ezlite are bfin Jun 08 08:47:49 hey Jay7 Jun 08 08:49:07 XorA: :) Jun 08 09:12:37 raster: こんにちは、白偉大な男. Jun 08 09:14:20 ;D Jun 08 09:14:22 zecke: 日本語!!!ぐげーーーー! Jun 08 09:14:27 err Jun 08 09:14:31 すげーーーー! Jun 08 09:14:37 :) Jun 08 09:15:06 ant_work: morning :) Jun 08 09:15:27 ~hail google translate Jun 08 09:15:29 * ibot bows down to google translate and chants, "I'M NOT WORTHY!!" Jun 08 09:16:51 * ant_work thinks that unicode has still a long way to go... Jun 08 09:16:59 hrw: google translate may have problems with my street-japanese :) Jun 08 09:19:50 well.. I have troubles while compiling qemu-native-0.12.4 with -j 3 Jun 08 09:20:18 was it only my problem? :) Jun 08 09:20:30 w/o PARALLEL_MAKE was compiled fine Jun 08 09:21:22 raster: haha, google is confused.. how is life? Jun 08 09:21:52 bad gooogle Jun 08 09:21:55 bad Jun 08 09:21:56 bah Jun 08 09:22:03 i'm grumpy - i'm moving. packing up house Jun 08 09:22:05 emptying it out Jun 08 09:22:16 and now housing people fucked up royally Jun 08 09:22:23 i'll be living in limbo for a while Jun 08 09:22:23 raster: moving house, same city? Jun 08 09:22:28 no Jun 08 09:22:31 moving countries again. Jun 08 09:22:35 you know me Jun 08 09:22:37 oh... i fear... it is not japan Jun 08 09:22:44 you can play "where in the world is raster" Jun 08 09:22:48 yeah, I will move back to Germany later this year. Jun 08 09:22:53 i'm wherever carmen sandiego is :) Jun 08 09:23:02 hehe Jun 08 09:23:03 nup - not japan Jun 08 09:23:09 moving back to .de? Jun 08 09:23:13 what about your french gf? Jun 08 09:23:15 going too? Jun 08 09:23:33 raster: yupp, I try to wrap up my studies... Jun 08 09:40:45 JaMa|Wrk: have you seen error messages about missing uinput module with 2.6.34? I saw it for console-image on c7x0. Jun 08 09:41:38 ant_work: in dmesg? Jun 08 09:41:56 no, on tty1 Jun 08 09:42:40 anyone seen a build problem with libpam? Jun 08 09:42:51 ant_work: don't remember seeing it.. Jun 08 09:43:18 never seen with 2.6.26 Jun 08 09:43:22 some renaming? Jun 08 09:44:44 # CONFIG_INPUT_UINPUT is not set Jun 08 09:44:54 ah, in 2.6.34.. Jun 08 09:45:11 ok, need just some minor refinements Jun 08 09:51:27 03Henning Heinold  07org.openembedded.dev * r17bcab4c36 10openembedded.git/recipes/xorg-proto/ (12 files): inputproto: converted to BBCLASSEXTEND Jun 08 09:51:32 03Henning Heinold  07org.openembedded.dev * r8031ee4953 10openembedded.git/recipes/zziplib/ (4 files in 2 dirs): zziplib: convert to BBCLASSEXTEND and use new patch system Jun 08 09:58:07 good morning Jun 08 10:10:04 03pieterg  07org.openembedded.dev * r4711e27cc5 10openembedded.git/classes/gitpkgv.bbclass: gitpkgv: new Jun 08 11:15:20 khem, i think you should run contrib/gcc_update -r $svnrevision to get the timestamps right instead of manually calling svn update -r $svnrevision Jun 08 11:16:02 hi, how can I remove that: NOTE: the following files were installed but not shipped in any package: Jun 08 11:16:08 in a package Jun 08 11:16:12 by not installing the files? Jun 08 11:17:09 db-5.0.21.tar.gz can't be fetched Jun 08 11:18:28 GNUtoo|laptop: either ship them, or don't install them Jun 08 11:20:17 I'll not install them Jun 08 11:20:22 that's the goal Jun 08 11:20:34 I'm looking into it right now Jun 08 11:22:39 Jay7, is it permanant? Jun 08 11:22:41 -rw-r--r-- 1 gnutoo gnutoo 29300812 2010-03-31 02:27 db-5.0.21.tar.gz Jun 08 11:23:14 gnutoo: seems yes.. Jun 08 11:23:18 will try by hands Jun 08 11:23:32 ok,else ask me for the source Jun 08 11:23:37 I'll put it online Jun 08 11:23:39 you'll fetch Jun 08 11:23:43 and put in a mirror Jun 08 11:29:01 mmm headers should be staged I think Jun 08 11:29:12 s/headers/kernel headers Jun 08 11:29:54 basicalyy here's my issue Jun 08 11:30:02 I've a recipe for out of tree kernels Jun 08 11:30:41 It deploys the kernel fine etc... Jun 08 11:30:56 but it says something like NOTE: the following files were installed but not shipped in any package: Jun 08 11:31:01 with a huge number of files Jun 08 11:31:03 .c .h etc Jun 08 11:31:15 not shure about the .c Jun 08 11:31:17 I'll check Jun 08 11:31:28 but I bet there are at least the headers Jun 08 11:31:42 should I really ship the kernel headers? Jun 08 11:31:50 I bet so Jun 08 11:32:02 for beeing able to compile on target a module Jun 08 11:35:01 May I run qt application under in Angstrom? or GTK+ based application only run in Angstrom? Jun 08 11:37:12 Guest85536, what's the application? Jun 08 11:37:21 download.oracle.com -> 404 Jun 08 11:37:21 and angstrom has QT I think Jun 08 11:37:23 hehe.. Jun 08 11:38:03 url was changed Jun 08 11:38:05 ok Jun 08 11:38:14 ah no.. Jun 08 11:38:19 seems site problems Jun 08 11:38:24 ah? Jun 08 11:38:31 http://www.oracle.com/technology/software/products/berkeley-db/index.html Jun 08 11:38:40 gnutoo : say, simple GUI based application Jun 08 11:38:44 ok Jun 08 11:38:45 can't fetch even by url from site Jun 08 11:38:55 so we mirror? Jun 08 11:39:14 seems we should at least for some time Jun 08 11:39:24 Guest85536, I think it'll work but if this application is not in openembedded/angstrom you'll need to cross-compile it Jun 08 11:39:36 s/it'll/it should Jun 08 11:40:46 Jay7, should I put the source on my server? Jun 08 11:41:13 gnutoo: yes please Jun 08 11:41:24 ok Jun 08 11:42:42 gnutoo, u mean, I can cross-compile qt application in my host & directly can run in Angstrom. Jun 08 11:42:55 yes Jun 08 11:43:16 gnutoo, any recipes i need to run related to qt? Jun 08 11:43:25 Jay7, http://gnutoo.homelinux.org/openembedded/sources/ Jun 08 11:43:54 Guest85536, search for qt in the recipes directory Jun 08 11:44:07 gnutoo, ok Jun 08 11:44:25 I didn't do GUI applications developments beside pygtk for the targets Jun 08 11:46:08 Jay7, btw do you have some experience with kenrel.bbclass,kernel recipes etc... Jun 08 11:46:24 gnutoo: ant_work have ;) Jun 08 11:47:19 ok thanks a lot Jun 08 11:47:42 because I'm unshure what to do with the image/kernel directory Jun 08 11:47:46 i have a question on some weird behavior of bitbake: when compiling an image openembedded/recipes/gtk+/gtk+_2.20.0.bb fails on the first build, but succeeds on the second build Jun 08 11:47:49 that would contain the headers Jun 08 11:48:03 with x-load it's the other way round: succeeds the first time, but fails on any rebuild attempt Jun 08 11:48:35 any of these problems known? Jun 08 11:52:16 Jay7, was the download ok? Jun 08 11:53:09 gnutoo: downloading, thanks :) Jun 08 11:53:15 ok Jun 08 11:53:32 because with the old router I had permissions issues sometime Jun 08 12:06:39 CMoH: I bet you're using multiple BB_THREADS Jun 08 12:06:52 ant_work, it is not the case Jun 08 12:07:02 actually, on one computer I am using and on another i am not Jun 08 12:07:16 both behave the same way so it should be in theory unrelated to BB_THREADS Jun 08 12:07:24 I am using PARALLEL_MAKE on both though Jun 08 12:07:29 strange then, I *always* see gtk+-native failing on first build Jun 08 12:07:55 with multiple BB_THREADS Jun 08 12:08:53 well, thi sworks: Jun 08 12:09:04 PARALLEL_MAKE = "-j5" Jun 08 12:09:06 #BB_NUMBER_THREADS = "4" Jun 08 12:09:08 BB_NUMBER_THREADS = "2" Jun 08 12:09:17 2 is fine, it seems... Jun 08 12:10:20 it's not an actual problem since the packages do get built in the end; or in the case of x-load only the first time Jun 08 12:10:54 i mean it's not a blocker - rather a hindrance Jun 08 12:12:00 fwiw http://lists.linuxtogo.org/pipermail/openembedded-devel/2010-May/019925.html Jun 08 12:12:20 discussion somehow stalled Jun 08 12:12:57 yeah it's the same error, thanks for the point Jun 08 12:15:26 who can upload db-5.0.21 to oe mirror? Jun 08 12:15:59 why? upstream SRC_URI doesn't work? Jun 08 12:16:07 at this moment - no Jun 08 12:16:18 download.oracle.com return 404 Jun 08 12:16:27 we can wait some time Jun 08 12:17:37 Jay7: http://build.shr-project.org/sources/db-5.0.21.tar.gz Jun 08 12:17:53 I've downloaded from gnutoo's site Jun 08 12:18:20 ok Jun 08 12:19:14 854/4986 tasks :( Jun 08 12:19:16 too long Jun 08 12:48:50 i've had problems build libpam_1.1.0, seems they are related to berkley db, a new db_5.0.21 recipe was just added. adding DEFAULT_PREFERENCE = "-1" to db_5.0.12.bb solves the build failure of libpam Jun 08 14:12:54 * kergoth sighs Jun 08 14:16:30 * XorA hands kergoth the coffee Jun 08 14:16:33 the variable reference tracking code missed quite a lot of variables Jun 08 14:16:41 i'm *hoping* the vast majority are due to the methodpool functions Jun 08 14:16:51 since a tweak to bitbake and i can start including those Jun 08 14:16:52 * stefan_schmidt hands kergoth a cookie for the coffee Jun 08 14:16:53 * kergoth ponders Jun 08 14:17:02 hurray, thanks ;) Jun 08 14:17:30 XorA: that was our nice job for the day :) Jun 08 14:17:40 hehe Jun 08 14:18:14 hmm, good deed would have sound less german I think :) Jun 08 14:18:42 hmm Jun 08 14:18:48 i think bitbake -k needs some work Jun 08 14:18:52 it just halted when it shouldn't have Jun 08 14:20:58 btw, bitbake interactive mode seems broken in master Jun 08 14:21:50 yes, db-native_5.0.21.bb' fails Jun 08 14:22:50 mckoan: above was two links for sources Jun 08 14:25:30 yeah i've heard about that, haven't had a chance to look at it. RP would probably be the better person to ask about it, since he did the client/server rework, but will take a look Jun 08 14:25:31 Jay7: yep, downloading by hand, thx Jun 08 14:31:07 * kergoth sighs Jun 08 14:47:03 now tiff is failed :) Jun 08 14:47:56 http://tinderbox.openembedded.org/builds/73341/ Jun 08 15:03:16 ah.. seems that was because build run out of space Jun 08 15:04:28 happens Jun 08 15:16:23 ugh, crap Jun 08 15:16:44 hmmm Jun 08 15:22:04 * kergoth just realized there's another big missing piece to the variable reference tracking :9 Jun 08 15:22:05 :( Jun 08 15:22:11 event handlers. Jun 08 15:22:27 as with the method pool, those no longer exist in the metadata Jun 08 15:39:14 kergoth: event handlers should never change anything at least? Jun 08 15:41:23 Anyone hit cmake+x8664 coming up with unsupported operating system, in oe.dev? Jun 08 15:46:02 RP: how can I halp you to debug what we presume being a race issue? Do you need ssh to inspect the rootfs? Jun 08 15:48:23 ant_work: how about a pointer to a summary of the problem for a start? :) Jun 08 15:48:25 your words were "the threading dependency code has been around for a while and Jun 08 15:48:27 most bugs like this turn out to be issues in the metadata, not bitbake." Jun 08 15:48:56 ant_work: hmm, that thread :/ Jun 08 15:49:12 updated today btw...offtopic I guess :/ Jun 08 15:49:20 ant_work: you need to concentrate on producing a small test case that shows the problem, try and isolate it Jun 08 15:50:12 th efact that same issue comes rebuilding from pstage would indicate a bad package, isn't? Jun 08 15:59:17 RP: yeah, hopefully the event handlers all either 1) modify the metadata, in which case the changed vars will be caught later when something uses them, or 2) cosmetic - display to the user, nothing with side effects. since they can't stop a task from running, or change how it runs, maybe it's okay to ignore them. Jun 08 16:06:52 problems with gettext-native-0.18-r1 too ? Jun 08 16:12:17 gettext-native-0.18-r1 build fails, is it normal? http://pastebin.com/rzT8Nb2b Jun 08 16:35:42 * Jay7 still have problems with tiff Jun 08 16:50:45 is there a shorthand like PREFERRED_VERSION_foo="git" for PREFERRED_VERSION_foo="0.8.15+gitr4711" ? Would be convenient to just track git regardless if khem changed the revision again :) Jun 08 16:51:32 iirc you can substitute % for the SRCREV Jun 08 16:55:14 it would be nice to document BB_SRCREV_POLICY in OE's local.conf.sample Jun 08 16:55:36 sounds like you just volunteered ;) Jun 08 16:57:45 kergoth, i think the recipe of that package is just not really ok; let me leave PV at it's default and set SRCREV and PR_append instead Jun 08 16:59:35 * kergoth still isn't 100% on the srcrev/autorev/srcpv/etc shit Jun 08 17:06:06 blindvt: you can set DEFAULT_PREFERENCE = "99999" in the git recipe too Jun 08 17:10:29 http://tinderbox.openembedded.net/public/logs/task/6191829.txt Jun 08 17:11:13 any thoughts? Jun 08 17:13:32 blindvt: but soon I will change the mechanism for uclibc to use new autorev Jun 08 17:19:06 when building micro distro for avr 32 the meta-toolchain task fails with: Jun 08 17:19:06 | The directory that should contain system headers does not exist: Jun 08 17:19:06 | /home/oe/atngw100/debug_build/sysroots/avr32-oe-linux-uclibc/usr/include Jun 08 17:19:10 any suggestions? Jun 08 17:21:36 nik0n: micro uses /include for locating system headers instead of /usr/include Jun 08 17:21:39 khem, ah yes, i did that too ;) Btw.. i'm currently forwarding the uClibc-configury to master. Now i'm currently filling in the features, but i'm not pleased with the networking handling for i'd prefer that to come from the user rather than only from the distro. What do you think? ( http://paste.debian.net/76638/ ) Jun 08 17:22:02 nik0n: which recipe fails like that Jun 08 17:22:11 khem: gcc-cross-sdk Jun 08 17:22:30 blindvt: ok if you refine your idea that would work Jun 08 17:22:34 khem, should ARM USE_BX come in via distro or machine? Jun 08 17:22:35 to be exact: gcc-cross-sdk_4.2.2.bb Jun 08 17:22:39 well.. Jun 08 17:22:44 * Jay7 is going to use devshell.. Jun 08 17:22:46 blindvt: via machine Jun 08 17:24:23 khem, should networking (socket, ipv4, ipv6) come from USER_FEATURES/DESIRED_FEATURES (no idea how to name it) which is ORed in to the distro networking features? Jun 08 17:24:42 khem, or how should that be handled in your POV? Jun 08 17:25:09 blindvt: here is how are see it. DISTRO_FEATURES is the supreme power and MACHINE_FEATURE is complementing it Jun 08 17:26:09 DISTRO_FEATURE may ask for THUMB but machine may override it saying oh I dont support THUMB Jun 08 17:27:00 khem, so if machine supports thumb, distro asks for it, using BX is up to whom? distro? user? Jun 08 17:27:26 blindvt: using BX is a machine feature for sure Jun 08 17:28:03 khem, yea, but it's optional, isn't it Jun 08 17:28:23 ... run.do_devshell.9216: line 500: gnome-terminal: command not found Jun 08 17:28:24 blindvt: however sometimes I think of feature blocks and then DISTRO can pick and chose to form a profile Jun 08 17:28:28 hmm.. Jun 08 17:28:49 how can I replace gnome-terminal with something other? Jun 08 17:28:59 blindvt: no its not optional when concerned machine is involved where BX is not available we have to use MOV PC LR Jun 08 17:29:35 ah.. $TERMCMD Jun 08 17:31:38 nik0n: paste complete log.do_compile somewhere Jun 08 17:32:44 khem, while i'm preparing supper.. Think of networking. micro pulls in sysvinit (at least per default, didn't look yet and in the end i don't want it) which in turn seems to require inet_ntoa, i.e. ipv4. Does sysvinit in this scheme require ipv4 -- i.e. would not be a legit provider of whatever it provides -- or impose libc to turn on ipv4? Jun 08 17:32:54 bbl Jun 08 17:32:55 khem: sure wait a mom Jun 08 17:43:09 Why is ophono being preferred over fso by other mobile initiatives other than angstrom and shr? Jun 08 17:52:49 shazkhan: ophono stole fso design didnt they :) Jun 08 17:53:25 shazkhan: Because ofono is backed by Nokia and Intel? Jun 08 17:53:44 And because it's (to be) used in actual shipping devices? Jun 08 17:55:39 Plus, the architectural design is not too bad... Jun 08 17:57:26 they think FSO is slow and only for mobile phones Jun 08 17:57:41 and does not care for standards compliance Jun 08 17:58:00 at the end of the day, it's all about control Jun 08 17:58:09 yep Jun 08 17:58:25 and they think that it does not support multiple backends Jun 08 17:59:11 does ofono use dbus at core ? Jun 08 17:59:20 yes Jun 08 17:59:30 but its more modular Jun 08 17:59:49 and people can licence client applications in whatever way they want Jun 08 17:59:51 I see Jun 08 17:59:56 thats the catch Jun 08 18:00:10 as long as you go DBus, you're pretty much safe Jun 08 18:00:17 dbus is crossing the license barrier Jun 08 18:00:39 so you can do the same with ofono, fsogsmd, pyneo, etc. Jun 08 18:00:54 no license problems until you hack the middleware Jun 08 18:01:22 mholtmann is notoriously known for not being a teamplayer... Jun 08 18:01:37 he's a bright guy, but that's one of his weaknesses Jun 08 18:02:03 so theheh Jun 08 18:02:12 I dont know that much Jun 08 18:02:48 well, the mere existance of connman tells a story Jun 08 18:02:58 so they think its going to scale across mobility devices Jun 08 18:03:37 they hope, yes. but the primary goal was to have something for their own produts Jun 08 18:03:46 yeah Jun 08 18:04:00 mickeyl: is it and offshoot of meego ? Jun 08 18:04:05 in some way Jun 08 18:04:13 lets say meego is embracing it Jun 08 18:04:22 oh I see Jun 08 18:04:28 it wasn't planned, but it's now an important part of meego Jun 08 18:04:34 since meego is a lot about phones Jun 08 18:04:35 i see Jun 08 18:04:39 they need the phone middleware Jun 08 18:04:47 intel is more interested in selling atom Jun 08 18:04:56 in MIDs Jun 08 18:05:30 ultimately intel wants smartphones Jun 08 18:05:34 infact they are pretty much scared of ARM Jun 08 18:05:37 they approached us years ago Jun 08 18:05:43 with a CPU design for a phone Jun 08 18:05:53 but back then it was ridicolously high power consumption Jun 08 18:05:56 with their cortex a9 there were rumours of arm coming with desktop chip Jun 08 18:07:37 mickeyl: what does ofono offers that fso does not Jun 08 18:07:45 in your opinion Jun 08 18:08:02 well, first, you really need to compare fsogsmd w/ ofono Jun 08 18:08:06 FSO is so much more than just GSM middleware Jun 08 18:08:21 ofono offers manpower and likely standardization Jun 08 18:08:28 but that's it Jun 08 18:08:35 yeah standardization is key in industry isnt it Jun 08 18:08:41 fsogsmd has the better core architecture and language Jun 08 18:08:45 (IMO) Jun 08 18:09:09 yes, it is Jun 08 18:09:19 I thought FSO was based on standards ...? Jun 08 18:09:21 who is backing fsogsmd atm Jun 08 18:09:23 and that's the reason why ofono will likely win Jun 08 18:09:40 shazkhan: standardization in the sense of "many people will use that, because it's from Intel and Nokia" Jun 08 18:09:45 de-factor standardization Jun 08 18:09:47 not de-jure Jun 08 18:10:04 khem: only the community, as there are no longer any commercial products shipping with FSO Jun 08 18:10:13 so if linaro also choses it then it gets another scape goat Jun 08 18:10:26 *nod* Jun 08 18:10:34 it has already chosen ofono Jun 08 18:10:43 hmm Jun 08 18:10:58 if I can't find any commercial vendor for FSO this year, it's doomed to remain a hobby Jun 08 18:11:07 and as such i will put much less energy in it then Jun 08 18:11:14 shazkhan: are you part of linaro decision makes then ask them to support FSO Jun 08 18:14:14 Couldn't ofono be extended? Jun 08 18:14:33 I'm assuming you have disagreements with their architectural choices. Jun 08 18:15:14 khem can't find the link to linaro mailing archive .... this has already been discussed. Being new to embedded world I tried my best to promote FSO Jun 08 18:16:03 shazkhan: oh well. I guess ideally intel and nokia should have backed fso Jun 08 18:16:19 and influenced the standardization parts and made it a reality Jun 08 18:16:29 instead Jun 08 18:17:48 that would have been nice, yes. Jun 08 18:17:52 for many of us Jun 08 18:18:04 khem u are on the linaro-dev! Jun 08 18:18:55 haven't you checked you inbox? You might be more sensible then me to reply based on your experience .... Jun 08 18:19:15 shazkhan: now I am Jun 08 18:19:21 yes I am reading the posts Jun 08 18:19:38 And I am reading your reply as well ;p Jun 08 18:20:57 meh Jun 08 18:21:00 khem asac clearly said they are using ofono .... Jun 08 18:21:19 yeah I see that Jun 08 18:21:26 oh well well Jun 08 18:21:42 * khem needs to eat Jun 08 18:24:21 kergoth: I tried to filter our *-dbg from module RDEPENDS but I am failing atm Jun 08 18:24:46 I tried to use oe_filter_out Jun 08 18:24:54 in kernel.bbclass Jun 08 18:31:56 well Jun 08 18:32:05 about tiff error Jun 08 18:32:18 mv -f .deps/dummy.Tpo .deps/dummy.Plo Jun 08 18:32:44 this fails because there are no *.Tpo but some *.Plo Jun 08 18:36:53 under devshell all is compiled ok.. Jun 08 18:41:35 hmmm... Jun 08 18:41:41 now it's bitbaked ok Jun 08 18:41:47 * Jay7 wonders Jun 08 18:42:53 I've only bitbake devshell before Jun 08 19:07:41 how to use LEAD_SONAME to specify it only for subpackage? (ie libgpsd in this case?) is LEAD_SONAME_subpackageName right override? Jun 08 19:08:00 DEBUG: LIBNAMES: pkg libgps libs 1 bins 0 sonames ['libgpsd.so.0', 'libgps.so.19'] Jun 08 19:08:00 NOTE: Multiple libraries (libgpsd.so.0, libgps.so.19) found and LEAD_SONAME not defined Jun 08 19:08:13 yes Jun 08 19:08:35 or we can package libgpsd to separate package (it probably wasn't part of libgps before gpsd_2.94) Jun 08 19:09:27 yes thats possible too Jun 08 19:12:31 khem: looks lik LEAD_SONAME_libgps doesn't work, still says "NOTE: Multiple libraries (libgpsd.so.0, libgps.so.19) found and LEAD_SONAME not defined" but with LEAD_SONAME = "libgps.so" it works Jun 08 19:16:16 I think you'd be better off putting the libs in separate packages. Jun 08 19:16:42 LEAD_SONAME is something of a historical mistake and it's almost never the right answer to any question. Jun 08 19:16:49 * kergoth grumbles and tries to sort out the NoProvider mess in bitbake Jun 08 19:17:09 pb__: ok.. build confirmed that libgpsd is new in 2.94 and that's why libgps19 was renamed to libgps during upgrade from 2.91 to 2.94 Jun 08 19:17:50 now I have to check what is libgpsd for and if some other recipe rdepend on it Jun 08 19:20:10 kergoth: proposed oe_popen is available also in older bitbake-1.18 right? Jun 08 19:20:16 it's not in bitbake Jun 08 19:20:18 it's in OE Jun 08 19:20:21 hence oe_ Jun 08 19:20:34 ah, ok, thanks Jun 08 19:21:17 look at openembedded/lib/oe/utils.py Jun 08 19:21:28 and openembedded/classes/utils.bbclass Jun 08 19:21:31 * kergoth doesn't remember which for that offhand Jun 08 19:25:30 not so successfull on first try, but I'll check http://paste.pocoo.org/show/223323/ Jun 08 19:27:38 that pasteydoesn't actually show anything useful Jun 08 19:27:50 it's erroring about there being no log file, but it doesn't include the cause of the task failing Jun 08 19:28:03 khem: log-do_compile for gcc-cross-sdk with micro/avr32 is here: http://www2.poessinger.org/log.do_compile.24099 Jun 08 19:29:20 part of that is a bug in bitbake, of course, it shouldn't freak out and die when a python function doesn't emit anything to the log Jun 08 19:29:27 kergoth: yes :/, wasn't there some patch for better error messages recently? maybe only for some types of errors Jun 08 19:29:46 there are many, many areas where the exception handling needs to be reworked Jun 08 19:29:50 one of them has been improved Jun 08 19:29:52 :) Jun 08 19:37:12 * kergoth moves the printing of no provider and multiple provider messages into the UI Jun 08 19:41:00 damnit, trying to kill the traceback from the NoProvider exception Jun 08 19:41:02 lost in a maze Jun 08 19:50:26 Has anybody built vlc with mozilla support on OE? Jun 08 19:54:37 damnit Jun 08 20:00:54 the most annoying part about trying to improve the exception handling in bitbake is exceptions can be raised in any of 3 different contexts, and how they're caught varies from one to the next Jun 08 20:00:59 (server, client, worker process) Jun 08 20:01:21 its unclear looking at any given piece of code which contexts it is/could be called in Jun 08 20:01:25 RP: thoughts? Jun 08 20:03:43 hmm, also appears there's a bug if you build with -k and get a NoProvider Jun 08 20:03:55 seems to hang, like the server process doesn't realize there's nothing left to be built Jun 08 20:03:56 hmmm Jun 08 20:22:45 now cmake Jun 08 20:22:53 * kergoth dies Jun 08 20:23:15 http://tinderbox.openembedded.net/public/logs/task/6193478.txt Jun 08 20:23:32 changing this behavior should NOT be this difficult Jun 08 20:23:57 seems FORTIFY_SOURCE or vice versa Jun 08 20:24:13 khem, http://paste.debian.net/76666/ Jun 08 20:26:49 http://public.kitware.com/mantis/view.php?id=9267 Jun 08 20:26:50 hehe.. Jun 08 20:26:57 That's because string literals in lines 1899..1901 are longer than CHIPNAME_STRING_LENGTH. Jun 08 20:27:12 last update 2009-11-16 11:47 Jun 08 20:27:16 oh.. Jun 08 20:27:27 AttributeError: RunQueue instance has no attribute 'build_pipes' Jun 08 20:27:29 argh Jun 08 20:29:00 looks like an interrupt during the runqueue generation causes this one, as the final states assume it's been entered when it tries to clean up Jun 08 20:29:03 * kergoth grumbles Jun 08 20:35:20 03Henning Heinold  07org.openembedded.dev * r851601ff57 10openembedded.git/recipes/xorg-proto/ (5 files): xf86bigfontproto: converted to BBCLASSEXTEND Jun 08 20:35:22 03Henning Heinold  07org.openembedded.dev * r22d5890c87 10openembedded.git/recipes/xorg-proto/ (10 files): xextproto: converted to BBCLASSEXTEND Jun 08 20:35:22 03Henning Heinold  07org.openembedded.dev * rcd5f214042 10openembedded.git/recipes/xorg-proto/ (5 files): bigreqsproto: converted to BBCLASSEXTEND Jun 08 20:35:23 03Henning Heinold  07org.openembedded.dev * r35af214fdf 10openembedded.git/recipes/xorg-proto/ (5 files): xcmiscproto: converted to BBCLASSEXTEND Jun 08 20:36:39 jo Jun 08 20:36:51 dman the db5 download is broken by oracle Jun 08 20:39:53 * kergoth sighs Jun 08 20:45:23 he kergoth whats up? Jun 08 20:45:48 trying to do some quite simple improvements to bitbake and its proving to be a gigantic pain in the ass Jun 08 20:45:55 :( Jun 08 20:46:00 as an example, sick of seeing a traceback for NoProvider Jun 08 21:09:05 hm Jun 08 21:09:07 NOTE: Couldn't find shared library provider for libXinerama.so.1 Jun 08 21:09:19 bitbake gtk+ Jun 08 21:16:30 jo ant Jun 08 21:17:17 03Martin Jansa  07org.openembedded.dev * r2aef1cb15d 10openembedded.git/classes/debian.bbclass: Jun 08 21:17:17 debian.bbclass: call objdump from binutils-cross Jun 08 21:17:17 * host objdump can support multiple targets and instead of output it shows list of targets to choose from Jun 08 21:17:17 see http://thread.gmane.org/gmane.comp.handhelds.openembedded/33294/focus=33357 Jun 08 21:17:17 * PATH set in cmd as suggested by kergoth Jun 08 21:17:17 Signed-off-by: Martin Jansa Jun 08 21:17:19 Acked-by: Khem Raj Jun 08 21:17:19 03Martin Jansa  07org.openembedded.dev * r2d0589a1de 10openembedded.git/recipes/linux/ (linux-openmoko-2.6.34/pcf.patch linux-openmoko-2.6.34_git.bb): Jun 08 21:17:20 linux-openmoko-2.6.34: bump SRCREV, remove applied pcf patch Jun 08 21:17:20 Signed-off-by: Martin Jansa Jun 08 21:17:20 03Martin Jansa  07org.openembedded.dev * rcd5497ac36 10openembedded.git/recipes/linux/ (2 files in 2 dirs): Jun 08 21:17:21 linux-openmoko-2.6.29: fix for lost touchscreen bug #2328 Jun 08 21:17:22 Signed-off-by: Martin Jansa Jun 08 21:17:22 03Martin Jansa  07org.openembedded.dev * r3132122132 10openembedded.git/recipes/postgresql/ (3 files in 2 dirs): Jun 08 21:20:49 03Martin Jansa  07org.openembedded.dev * r65af999427 10openembedded.git/classes/debian.bbclass: debian.bbclass: remove unwanted inherit leftover (from trying oe_popen instead setting PATH) Jun 08 21:23:13 jo pb Jun 08 21:23:26 woglinde: hey...playing with gtk+-native? Jun 08 21:23:36 no Jun 08 21:24:05 oh I see, normal gtk+ Jun 08 21:25:05 03Khem Raj  07org.openembedded.dev * r9d886cd149 10openembedded.git/recipes/pam/ (libpam-1.1.0/define-HAVE_DBM.patch libpam_1.1.0.bb): Jun 08 21:25:05 libpam_1.1.0.bb: Add patch to fix running with DB5 Jun 08 21:25:05 * Imported from gentoo bug #319831 Jun 08 21:25:05 Signed-off-by: Khem Raj Jun 08 21:36:02 03Martin Jansa  07org.openembedded.dev * r66c5bc0721 10openembedded.git/classes/insane.bbclass: Jun 08 21:36:02 insane.bbclass: set PATH before calling desktop-file-validate Jun 08 21:36:02 * otherwise fails on hosts without desktop-file-utils installed, because cannot find binaries from desktop-file-utils-native Jun 08 21:36:02 Signed-off-by: Martin Jansa Jun 08 21:38:38 nik0n: ok the problem is that gcc assumes SYSTEM_HEADER DIR to be /usr/include and appends sysroot to it Jun 08 21:39:00 nik0n: and in micro's case /usr is omitted like traditional GNU systems Jun 08 21:39:35 nik0n: therefore gcc configury must be tought to use /include here when configuring for micro Jun 08 21:39:46 now its fixable but not without a hack. Jun 08 21:39:52 hm someone has a recpie at hand where configure is run in a buildir= Jun 08 21:40:01 03Martin Jansa  07org.openembedded.dev * r4b371c21f8 10openembedded.git/recipes/ (20 files in 10 dirs): Jun 08 21:40:01 packages: PR bump after EFL SRCREV change resulting to different .so names. Jun 08 21:40:01 Signed-off-by: Martin Jansa Jun 08 21:40:02 03Martin Jansa  07org.openembedded.dev * r59eadc34d1 10openembedded.git/recipes/efl1/ecore_svn.bb: Jun 08 21:40:02 ecore: remove no longer available --enable-ecore-txt Jun 08 21:40:02 Signed-off-by: Martin Jansa Jun 08 21:40:02 not the sourcedir Jun 08 21:40:03 03Martin Jansa  07org.openembedded.dev * r879c8e4433 10openembedded.git/recipes/efl1/evas.inc: Jun 08 21:40:03 evas: don't use SRCNAME in FILES and RREPLACES as efl.bbclass is using PN Jun 08 21:40:04 * it results to not packaging /usr/lib/.debug/libevas-ver-svn-06.so.0.9.9 to evas-dbg Jun 08 21:40:04 Signed-off-by: Martin Jansa Jun 08 21:40:12 03Martin Jansa  07org.openembedded.dev * r1789f88e45 10openembedded.git/ (conf/distro/include/sane-srcrevs.inc recipes/efl1/eet_svn.bb): Jun 08 21:40:12 EFL: bump SRCREV to new snapshots Jun 08 21:40:12 Signed-off-by: Martin Jansa Jun 08 21:40:52 nik0n: if you have options could you use minimal distro or some other distro which does not flatten the tree like micro ? Jun 08 21:41:39 khem I am trying now to build openjdk with uclibc-git Jun 08 21:42:02 woglinde: hmm too stressful for it I guess :) Jun 08 21:42:19 03Martin Jansa  07org.openembedded.dev * r3aaad67878 10openembedded.git/conf/distro/include/sane-srcrevs.inc: Jun 08 21:42:19 opkg: bump SRCREV for fix preventing overeager removing of files which were after opkg upgrade owned by different package Jun 08 21:42:19 Signed-off-by: Martin Jansa Jun 08 21:42:52 khem we will see Jun 08 21:43:09 but I need a way to use a builddir for configure Jun 08 21:43:19 which is not the sourcedir Jun 08 21:43:29 any hints for doing it with autotools? Jun 08 21:44:50 hmm you mean ${B} Jun 08 21:45:07 you can customize do_configure Jun 08 21:45:08 hm I meant Jun 08 21:45:11 mkdir build Jun 08 21:45:13 cd build Jun 08 21:45:23 ../configure Jun 08 21:45:24 B = ${S}/build Jun 08 21:45:25 ta da Jun 08 21:46:08 has anyone used shadow with uclibc lately ? Jun 08 21:46:18 kergoth do I need to make build? Jun 08 21:46:23 no Jun 08 21:46:25 or is it done automagicly? Jun 08 21:46:26 its in the 'dirs' flag Jun 08 21:46:26 okay Jun 08 21:46:28 thanks Jun 08 21:47:12 tomorrow I will be at the linuxtag Jun 08 21:47:16 could be fun at some point to try enabling it globally and add a way to explicitly disable it for recipes for projects that can't do out of tree builds Jun 08 21:47:17 but on the navit booth Jun 08 21:47:22 this year Jun 08 21:47:37 gcc alters B as well Jun 08 21:48:02 woglinde: enjoy the linuxtag Jun 08 21:48:19 ah.. linuxtag is opening today.. Jun 08 21:48:39 khem: what about killing that OPTSPACE? Jun 08 21:50:07 JaMa: I thought about it. have to RFC it I have patch lined up here did not do yet. Jun 08 21:50:16 JaMa: I will send out an email tonight Jun 08 21:50:41 great, thanks Jun 08 21:51:24 JaMa: actually someone here is measuring perf loss due to -Os Jun 08 21:51:44 he is using OE build X system and it performs slow Jun 08 21:52:04 where as he has some other root file system works like 30% faster Jun 08 21:52:04 khem on which machine? Jun 08 21:52:12 coretex a9 Jun 08 21:52:17 cortex Jun 08 21:52:31 khem: only with gcc-4.5 or every version with -Os? Jun 08 21:52:44 I think its 4.4.4 with Os Jun 08 21:52:53 it doesnt matter Jun 08 21:53:03 CSL toolchain plugged into OE same result Jun 08 21:53:15 khem intressting Jun 08 21:53:29 hm that's what I have in SHR (-Os set from sane-toolchain-eglibc..) Jun 08 21:53:31 so I gave him some tips Jun 08 21:53:34 what graphic chipt? Jun 08 21:53:40 I am waiting his response Jun 08 21:53:48 woglinde: no idea its secret :) Jun 08 21:54:05 hm okay Jun 08 21:54:11 JaMa: We need to compare the size bloat with -O2 Jun 08 21:54:13 maybee sgx Jun 08 21:54:22 than is cleat Jun 08 21:54:26 clear Jun 08 21:54:34 JaMa: if same image bloats with say 20% and perf improves by 5% Jun 08 21:54:41 prolly we are ok with Os Jun 08 21:55:05 khem: today I launched that byte benchmark on my spitz (OE gcc-4.5 build and debian - some gcc-4.4x) and both results are quite the same again Jun 08 21:55:28 when I will provide patch for cmake-2.6.4 and for cmake-native-2.6.4 where should I place this patch? Jun 08 21:55:38 JaMa: I would suggest to do add -flto when building with gcc 4.5 Jun 08 21:55:49 and you should see like 15% gain Jun 08 21:55:49 and how should I add it to SRC_URI? Jun 08 21:56:26 khem whats flto? Jun 08 21:56:29 Jay7: if the patch is specific to cmake-2.6.4 make a dir cmake-2.6.4 under recipe/cmake and put it there Jun 08 21:56:44 Jay7: otherwise dump it in files dir Jun 08 21:56:57 khem: it specific for both cmake and cmake-native but only for v.2.6.4 Jun 08 21:56:57 woglinde: link time optimizations Jun 08 21:57:17 khem did you enabled it per default on gcc-4.5? Jun 08 21:57:23 is it using BBCLASSEXTEND Jun 08 21:57:29 woglinde: no its a killer Jun 08 21:57:36 woglinde: its not ready for primetime yet Jun 08 21:57:39 ok, rebuilding benchmark (on target) Jun 08 21:57:51 khem hm Jun 08 21:58:11 so its configured Jun 08 21:58:16 woglinde: it is very agressive in optimization and people will crib about debugging info loss Jun 08 21:58:18 now lets see if it compiles Jun 08 21:58:38 woglinde: but eventually it could be used where horsepower is needed Jun 08 21:58:48 will -native recipes look into 'common' recipes dir for patches? Jun 08 21:58:59 I have seen reports of 5% to 30% improvements with -flto Jun 08 21:59:03 common == non-native Jun 08 22:00:01 Jay7: look what it has in FILESPATHPKG Jun 08 22:00:16 hrms Jun 08 22:00:18 utilities/globalDefinitions_gcc.hpp:239: error: 'finite' was not declared in this scope Jun 08 22:00:18 khem: thanks, I'll Jun 08 22:00:40 khem: cc1: error: LTO support has not been enabled in this configuration Jun 08 22:00:50 khem do you have a hint? Jun 08 22:01:02 woglinde: this article will inform you more http://lwn.net/Articles/387122/ Jun 08 22:01:07 dont we have finite with uclibc? Jun 08 22:02:33 hm Jun 08 22:02:34 http://bugs.gentoo.org/show_bug.cgi?id=289522 Jun 08 22:02:53 ah isfinite Jun 08 22:02:54 okay Jun 08 22:05:13 hm now isnanf Jun 08 22:05:18 JaMa: hmmm Jun 08 22:05:23 JaMa: let me see. Jun 08 22:06:26 hm Jun 08 22:06:28 -D_C99_SOURCE Jun 08 22:18:38 JaMa: I am pushing a patch to enable -flto Jun 08 22:18:52 JaMa: you have to rebuild your target gcc if you building on target Jun 08 22:18:58 then it should be ok Jun 08 22:19:04 to use it Jun 08 22:19:23 03Khem Raj  07org.openembedded.dev * raea549ddb2 10openembedded.git/recipes/gcc/ (2 files in 2 dirs): Jun 08 22:19:23 gcc-4.5.inc: Enable -flto on 4.5 so it can be used but its not on by default. Jun 08 22:19:23 Signed-off-by: Khem Raj Jun 08 22:20:48 this means gcc now will need libelf in DEPENDS Jun 08 22:21:09 hi mwester-laptop Jun 08 22:21:19 how is it going Jun 08 22:21:27 greetings from a far and distant land. Jun 08 22:21:38 heh are you in nova scotia Jun 08 22:21:51 or honululu Jun 08 22:21:58 Almost :) Waterloo, Ontario Jun 08 22:22:32 oh atleast its rhyming too Jun 08 22:22:45 waterloo Jun 08 22:22:50 at RIM ? Jun 08 22:23:11 Yep Jun 08 22:23:46 do they know you do linux too ;) Jun 08 22:25:23 :D Some do... one of the guys I work with is a huge Debian fan; he has his obligatory Blackberry, but also carries his personal HTC Android device about. Jun 08 22:26:27 heh Jun 08 22:26:50 There's been a bit of a culture shock in some parts here, recently -- the use of webkit has brought (O the horror!) -- git into the organization. :D Jun 08 22:26:54 khem: thanks, will rebuild and try tomorrow Jun 08 22:27:06 JaMa: ok Jun 08 22:27:16 gnight all Jun 08 22:27:46 mwester-laptop: so from cvs to git ? Jun 08 22:28:04 I deal with people who have this mental block Jun 08 22:28:06 Commercial (Perforce) to git. Jun 08 22:28:14 even worse Jun 08 22:28:28 hehe.. p4 Jun 08 22:28:30 cvs guys atleast trust open source Jun 08 22:28:41 Yep, that's exactly the issue: distrust. Jun 08 22:29:08 are you feeding webkit to them ? Jun 08 22:29:16 Not me. Jun 08 22:29:24 ok Jun 08 22:29:30 I'm just working with them on tools to organize their builds. Jun 08 22:29:44 my bike has arrived at store cant wait to get there and pick it up Jun 08 22:29:58 mwester-laptop: ok Jun 08 22:29:59 motorbike, or pedals? Jun 08 22:30:13 pedal Jun 08 22:30:31 I dont have license for motorbikes here in US yet Jun 08 22:30:47 but I am a avid two wheel driver :) Jun 08 22:31:13 I prefer 4; don't feel safe on something that needs to be moving to be stable. :) Jun 08 22:31:54 mwester-laptop: yeah hehe Jun 08 22:32:13 I usually dont ride where cars are driving Jun 08 22:32:19 usually on dedicated trails Jun 08 22:32:57 Chicago is renowned as a bicycle-UNfriendly city... so I use that as my excuse! Jun 08 22:33:17 heh Jun 08 22:33:32 bay area is in sense better Jun 08 22:33:45 still I hear cars hitting bikers Jun 08 22:33:52 so have to be careful Jun 08 22:42:14 well.. I have quick-and-dirty fix for cmake and fortify source enabled Jun 08 22:42:31 will cleanup and commit tomorrow Jun 08 22:42:47 03Andrea Adami  07org.openembedded.dev * r0ccccf729a 10openembedded.git/recipes/linux/linux-2.6.34/ (akita/defconfig c7x0/defconfig spitz/defconfig): Jun 08 22:42:47 linux_2.6.34: add defconfig for c7x0 and akita Jun 08 22:42:47 * minor fixes to spitz config, which is reused for akita Jun 08 22:42:47 * added uinput module Jun 08 22:51:46 * Jay7 -> sleep Jun 08 22:55:43 hmmm Jun 08 23:15:36 wtf Jun 08 23:15:42 this worked 15 minutes ago Jun 08 23:15:43 grr Jun 08 23:15:55 doing a bitbake -c foo and it just parses and exits, doesn't run the runqueue Jun 08 23:15:57 hmm Jun 08 23:16:36 kergoth: it happens to me too sometimes Jun 08 23:16:48 with some sdk recipes in the past Jun 08 23:16:55 but dont remember exactly Jun 08 23:19:09 very stange Jun 08 23:19:45 about strange things..the md5 check here http://fr.pastebin.ca/1879365 Jun 08 23:20:23 seems Expected ang Got are swapped Jun 08 23:21:23 ant__: hmmm Jun 08 23:21:32 that repo was down Jun 08 23:22:04 I'm copying by hand Jun 08 23:22:28 I just wanted to reproduce this odd thing (perhaps I do understand wrongly the message) Jun 08 23:23:08 he..two md5.. Jun 08 23:31:57 ok, found a valid mirror finally... Jun 08 23:33:03 good night Jun 09 00:04:36 there we go, got it Jun 09 00:04:52 khem: there was a bitbake bug that resulted in the silent exit when the specified task didn't exist Jun 09 00:05:11 ah Jun 09 00:05:16 <3 bisect P Jun 09 00:05:17 is it fixed ? Jun 09 00:05:17 :P Jun 09 00:05:20 yep Jun 09 00:05:24 ook Jun 09 00:05:30 1.10 wasn't affected Jun 09 00:05:49 03Khem Raj  07org.openembedded.dev * r0298bf1934 10openembedded.git/recipes/pam/ (5 files in 2 dirs): Jun 09 00:05:50 libpam-1.1.1: Add recipe. Jun 09 00:05:50 * Tested build on arm*uclibc Jun 09 00:05:50 Signed-off-by: Khem Raj Jun 09 00:05:59 03Khem Raj  07org.openembedded.dev * r0abf974adf 10openembedded.git/recipes/libelf/libelf_0.8.13.bb: Jun 09 00:05:59 libelf_0.8.13.bb: Add recipe Jun 09 00:05:59 Signed-off-by: Khem Raj Jun 09 00:05:59 03Khem Raj  07org.openembedded.dev * rec7e940083 10openembedded.git/recipes/gcc/gcc-4.5.inc: Jun 09 00:05:59 gcc-4.5: flto requires libelf as dependency Jun 09 00:06:00 Signed-off-by: Khem Raj Jun 09 00:07:44 kergoth: if you remember the fix to get dbg filtered out of module deps in kernel.bbclass then let me know Jun 09 00:07:54 I want to debug this shadow thing and its PITA Jun 09 00:08:05 to copy binaries with debugging Jun 09 00:08:10 ah Jun 09 00:08:24 could always disable the split of the .debug stuff entirely Jun 09 00:08:41 how ? Jun 09 00:08:46 that will work too Jun 09 00:09:12 not sure :) would have to check package.bbclass probably Jun 09 00:09:36 #INHIBIT_PACKAGE_STRIP = "1" Jun 09 00:09:36 #PACKAGE_STRIP = "no" Jun 09 00:09:47 I tried those last time but no success Jun 09 00:10:35 PACKAGE_STRIP = "no" should do it, going by the code in package.bbclass Jun 09 00:10:40 it doesn't run its runstrip function if its no Jun 09 00:11:13 ok let me try it Jun 09 00:11:51 what is INHIBIT_PACKAGE_STRIP Jun 09 00:11:56 * kergoth shrugs Jun 09 00:15:04 there is OLD_INHIBIT_PACKAGE_STRIP used in recipes/gcc/gcc-package-cross.inc Jun 09 00:15:08 but set nowhere Jun 09 00:16:30 there is DISTRO_TYPE which could be one of debug and release and it sets the flags accordingly Jun 09 00:17:03 and then there is another var DEBUG_BUILD which when set to 1 should enable debug info Jun 09 00:17:13 I guess they should be fused together Jun 09 00:18:42 bitbake.conf:SELECTED_OPTIMIZATION = "${@bb.data.getVar(['FULL_OPTIMIZATION', 'DEBUG_OPTIMIZATION'][bb.data.getVar('DEBUG_BUILD', d, 1) == '1'], d, 1)}" Jun 09 00:19:01 that stuff is pretty useless Jun 09 00:19:10 remnants from before the debug info could be built and then split out Jun 09 00:19:18 afaik, anyway Jun 09 00:20:29 kergoth: it still kicks in Jun 09 00:20:40 yes, but what's the point? Jun 09 00:20:48 of building without debug in the first place, i mean Jun 09 00:21:08 kergoth: so DISTRO_TYPE=debug|release sets the packages one wants to have for debugging like strace gdb etc. so semantically its different Jun 09 00:21:16 i know Jun 09 00:21:21 i was referring to SELECTED_OPTIMIZATION Jun 09 00:21:24 and DEBUG_BUILD Jun 09 00:21:33 well you can speed up compilation Jun 09 00:21:42 if you dont emit debug info needlessl Jun 09 00:21:49 y Jun 09 00:21:58 and it could save you disk space Jun 09 00:22:15 so why not Jun 09 00:24:55 angstrom uses -ggdb3 in FULL_OPTIMIZATION :) Jun 09 00:38:21 hmmm Jun 09 00:38:28 where the hell are these variable references coming from Jun 09 00:38:38 do_package_write doesn't do shit, yet somehow i'm picking up it doing something Jun 09 00:38:39 hmmm Jun 09 00:38:48 maybe its task interleaving Jun 09 00:43:57 damnit Jun 09 00:52:49 hmmm Jun 09 01:38:40 damnit, where the hell .. Jun 09 01:38:41 grrr Jun 09 01:39:37 MAINTAINER sure as hell isn't used by do_distribute_sources Jun 09 01:39:40 this boggles the mind Jun 09 01:39:47 * kergoth scratches head Jun 09 01:59:54 * mwester-laptop suddenly realizes that he copied his big server's local.conf to his laptop without changing -j 5 and bb_number_threads 4, which explains what's happened to his laptop that only has a single GB of RAM, dual-core CPU... :D Jun 09 02:02:16 hehe, oops Jun 09 02:38:05 hiyo Jun 09 02:38:12 greetings fellow OE nerds **** ENDING LOGGING AT Wed Jun 09 02:59:57 2010