**** BEGIN LOGGING AT Wed Aug 18 02:59:57 2010 Aug 18 03:01:13 t4/qt4-tools-native.inc: install -m 0755 bin/qmake ${D}${bindir}/qmake2 Aug 18 03:01:51 do I just leave the include of qt4-tools-native.inc or do I rename that as well? Aug 18 03:03:16 jkridner|work: well, there are at least two references to native, so you should move them out and then rename. at least this would be what I would do Aug 18 03:54:00 khem: thanks: http://tinderbox.openembedded.net/packages/evince/ Aug 18 04:36:43 got up out of bed to check my build status and I'm failing on x11vnc now: http://tinderbox.openembedded.net/builds/83181/ (In file included from scan.c:43: pointer.h:42: error: 'pointer' redeclared as different kind of symbol...include/X11/Xdefs.h:53: error: previous declaration of 'pointer' was here) Aug 18 04:38:35 hmm... looks like a new version compared to what else shows up in tinderbox. Aug 18 04:42:21 jkridner|work: about to leave, tried building demo image, about 6 pkgs failed, had no time to look into it yet Aug 18 04:42:34 6.42 am here, but see http://tinderbox.openembedded.net/packages/ and select failed Aug 18 04:42:41 any mention of what they were? Aug 18 04:42:42 thanks. Aug 18 04:43:01 I'm just going with a preferred provider for x11vnc to hope to get to a working version. Aug 18 04:43:19 yes I have list Aug 18 04:43:51 doh, x11vnc failed for the other version for you. :( Aug 18 04:44:12 and the same error. Aug 18 04:44:44 something must have changed in Xdefs.h Aug 18 04:45:24 odd, I don't have the x-load error. Aug 18 04:45:44 jkridner|work: have to go to work now, i've sent you an email with the list Aug 18 04:45:59 thanks! Aug 18 04:46:04 back in 30-40 mins Aug 18 04:46:14 * jkridner|work will likely be asleep. Aug 18 04:48:55 heh Aug 18 04:48:57 gm eFfeM_work Aug 18 04:54:20 Crofton: why gm ? Aug 18 04:54:34 errr ... sorry, it's 7am here too, I misread it ... Aug 18 04:54:54 lack of sleep seriously does its job Aug 18 05:28:51 gm Aug 18 05:29:09 hi, I found the problem with x11vnc. Aug 18 05:29:15 cool Aug 18 05:29:16 not sure why it would have built for anyone. Aug 18 05:29:44 I mean, I found a work-around... not sure if others would have had this problem.... Aug 18 05:29:55 but, I renamed pointer and that just about has it going. Aug 18 05:30:33 digging up the log Aug 18 05:31:31 i've seen something like this in a completely different pacakge Aug 18 05:31:46 but not sure why it would build at all indeed Aug 18 05:32:04 my proposed fix: http://gitorious.org/~Jadon/angstrom/jadon-openembedded/commit/2f714577b092f46dbdabe51c01ca9202579f1131 Aug 18 05:32:04 unless others have a different version of Xdefs.h (different X?) Aug 18 05:32:15 must be. Aug 18 05:32:34 but, I can't imagine other people picking up a different X with Angstrom... so must be non-Angstrom builds. Aug 18 05:33:12 looks good to me Aug 18 05:33:40 well I normally build x11-image if I build anything with X, and I don't think I've build x11vnc Aug 18 05:33:48 before Aug 18 05:33:52 I keep forgetting to -c clean when I don't pump the PR between fixes in my patches. Aug 18 05:36:16 jkridner|work: are you building x11vnc 9.9? I was but I see koen pushed 0.9.11 16 hrs ago Aug 18 05:36:37 no, i'm building 0.9.11. Aug 18 05:37:01 argh, I typed in my clean command wrong.... issuing build yet again. Aug 18 05:37:14 ask koen why he didn't get that problem, I would have expected him to get it as well Aug 18 05:37:46 * eFfeM_work would like it if there was a way to clean then rebuild in one cmd Aug 18 05:39:32 kergoth: if you are there, can you say something on the NATIVE_INSTALL_WORKS observation from Fahad (or say something about it on the ML) Aug 18 05:40:28 now gnome-utils fails. :( Aug 18 05:40:49 actually in my log I saw that boost also failed, seems very weird Aug 18 05:40:57 * jkridner|work issues 'bitbake x11vnc' to at least verify that was fixed. Aug 18 05:41:39 boost worked for me, so I'm assuming that I built some dependency for it "in time". Aug 18 05:41:49 gnome-utils needs a dependence on scrollkeeper whatever that is Aug 18 05:43:08 boost fails because it apparently requires cmake, guess that one was not there or not there properly Aug 18 05:43:18 will dig into these tonight Aug 18 05:44:04 x11vnc compiled finally... I'll submit the patch. Aug 18 05:44:16 http://tinderbox.openembedded.net/builds/83190/ Aug 18 05:44:39 great & thanks for the patch Aug 18 05:45:10 anyone familiar with sdk? I'm wondering why sdk.bbclass has its own STAGING_DIR_TARGET definition Aug 18 05:45:28 STAGING_DIR_TARGET = "${STAGING_DIR}/${OLD_MULTIMACH_ARCH}${TARGET_VENDOR}-${TARGET_OS}" Aug 18 05:46:12 bitbake.conf:STAGING_DIR_TARGET = "${STAGING_DIR}/${BASEPKG_TARGET_SYS}" Aug 18 05:46:50 hmmm.... "--disable-scrollkeeper" should have gotten rid of that scrollkeeper dependency. Aug 18 05:47:00 did something cause EXTRA_OECONF to break? Aug 18 05:47:09 These differ causing my gcc-cross-sdk trying to pick up dirs from armv5teb-... instead of armeb-... Aug 18 05:47:14 and ofc these are not there Aug 18 06:06:56 hmmm... autoreconf doesn't seem to use EXTRA_OECONF Aug 18 06:07:08 what do you mean? Aug 18 06:07:38 EXTRA_OECONF is passed to ./configure, not autoreconf, and it's done in oe, not in the tool. EXTRA_AUTORECONF or whatever is used in the call to autoreconf, but again is done in the oe class, not the tool Aug 18 06:07:40 I mean, gnome-utils is failing in autoreconf, not autoconf. Aug 18 06:07:46 or configure. Aug 18 06:08:15 inside autotools_do_configure... Aug 18 06:08:58 there is a case statement... and i'm not sure what drives it.... Aug 18 06:09:44 ...but, configure gets thrown away in it and autoreconf is run. Aug 18 06:10:42 03Jason Kridner 07org.openembedded.dev * rd9b316bb71 10openembedded.git/recipes/vnc/ (2 files in 2 dirs): Aug 18 06:10:42 x11vnc: replace definition of pointer Aug 18 06:10:42 Avoids conflict with Xdefs.h definition of 'pointer'. Aug 18 06:10:42 Signed-off-by: Jason Kridner Aug 18 06:10:42 Acked-by: Frans Meulenbroeks Aug 18 06:12:03 k, I was wrong, gnome-utils is dying in 'intltoolize --copy --force --automake' Aug 18 06:13:00 * jkridner|work thinks he should just enable scrollkeeper and be done with this. Aug 18 06:20:44 jkridner|work: at least in that case you should add a dep to scrollkeeper Aug 18 06:20:48 gnome-applets fails in the same way. Aug 18 06:21:19 yeah, that is the patch I'm playing with: http://gitorious.org/~Jadon/angstrom/jadon-openembedded/commit/616330681379114b2ae51c59e115e7f85c9d08b2 Aug 18 06:21:27 jkridner|work: i'm worrying there is something odd with your setup; e.g. your libgcrypt issue was not reproducible here Aug 18 06:21:35 unfortunately, the build of scrollkeeper fails. :( Aug 18 06:23:57 well, I can rebase my handful of image packages, nuke my tmpdir, and start a clean build (and go to sleep). Aug 18 06:25:11 03Frans Meulenbroeks 07org.openembedded.dev * r0479b70418 10openembedded.git/recipes/bison/bison.inc: Aug 18 06:25:11 bison.inc: add dependency on flex-native Aug 18 06:25:11 needed when building (as discovered by jkridner) Aug 18 06:25:11 Signed-off-by: Frans Meulenbroeks Aug 18 06:25:24 perhaps also nuke your pstage dir if it lives outside of TMPDIR Aug 18 06:25:38 did that already. Aug 18 06:50:38 is there an easy way to find out to what bitbake vars expand Aug 18 06:53:24 nevermind, just rediscovered bitbake -e Aug 18 08:11:40 eFfeM_work: when i take git diff does it take diff with the latest recipe present in the repository? Aug 18 08:12:06 i think so, not 100% sure Aug 18 08:12:36 Noor: latest in your checkout Aug 18 08:13:08 hi JaMa Aug 18 08:13:11 Noor: so if there is newer version and you didn't "git pull" then you won't see it Aug 18 08:13:22 OK Aug 18 08:14:17 hi eFfeM_work Aug 18 08:14:40 I thought when I do git diff it will take diff from the latest version Aug 18 08:14:42 hhhhmmm Aug 18 08:14:48 thanks for the information Aug 18 08:16:13 it diffs locally so against the latest in your copy, not the latest remote Aug 18 08:16:16 eFfeM_work: did you have time to review xorg cleanup? Aug 18 08:16:48 nope, was away most of yesterday evening Aug 18 08:17:13 this is planned for today, started off a clean build yesterday evening (but that also gave some issues) Aug 18 08:17:45 eFfeM_work: ok, if you plan it for today, I'll wait another day.. :) Aug 18 08:18:08 ok (briefly peeked into it, but saw no obvious problem) Aug 18 08:19:21 "nothing" changed in my builds - SHR without any preferred-xorg*.inc included (so all used versions are still there) Aug 18 08:19:39 and test build of micro with 7.5 include also doesn't complain about missing recipes Aug 18 08:20:07 actually tried to build beagleboard-demo-image last night but it failed on a few recipes (including boost :-(, seems a cmake dep is missing Aug 18 08:20:47 if you build from pstage suddenly you notice all kind of missing deps of packages that are there when you do a regular build Aug 18 08:22:46 * eFfeM_work curses at openoffice writer Aug 18 08:30:15 pb_: hi Aug 18 08:30:34 JaMa: the major things to watch out for is DEFAULT_PREFERENCE = "-1"; I also saw pinning of specific versions (but that seems quite rare and I haven't seen it in X, then again I did not look too carefully at X) Aug 18 08:37:21 Fahad_, Noor, really appreciate your work on removing do_stage. One small question though: when you have a native recipe that also has a non-native counterpart, could you see if they can be merged? Aug 18 08:42:12 Hello, does anybody know how to crosscompile rubygems with OE? I would like to install sqlite3-ruby on my gumstix overo. Aug 18 08:45:25 eFfeM_work: I'll try to look into this. Aug 18 08:46:01 would be nice, and if you are already looking at a recipe, it is something that is probably fairly easily done Aug 18 08:48:52 The only think I have to look is the activites that are achedule in mentor Aug 18 08:49:28 actaully Tom assigned us few recipes that we need to update Aug 18 08:49:38 ah ok Aug 18 08:49:52 so we are working on those recipes and pushing them Aug 18 08:50:14 I'll definetly try to combine some of the recipes Aug 18 08:54:37 zecke: hail zecke Aug 18 09:03:51 pb_: sorry to pester, did I miss the list of action items we need to work on? Aug 18 09:07:15 out of curiosity? have there been TSC meetings in jul/aug? there are no minutes on the wiki Aug 18 09:08:43 zecke: ah, no, I failed to generate the list. I will try to do that today. Aug 18 09:49:31 pb_: ping-ping Aug 18 09:51:55 hi jay7 Aug 18 09:52:18 pb_: XorA says that you are living around/in Cambridge, UK Aug 18 09:52:23 yes, that's correct Aug 18 09:52:45 can I ask you some questions about Cambridge hotels? :) Aug 18 09:52:46 * XorA gets blamed for all these days :-) Aug 18 09:53:14 sure, though I don't know all that much about hotels here. obviously I have my own house so I don't need the services of local hotels very often :-} Aug 18 09:58:08 pb_: are private msg acceptable? :) Aug 18 10:00:00 oh, right, sorry. I hadn't noticed those messages. Aug 18 10:13:40 03Koen Kooi 07org.openembedded.dev * r89a9abb4c6 10openembedded.git/conf/distro/ (7 files in 2 dirs): angstrom: start with merging in angstrom/angstrom-next branch Aug 18 10:13:52 03Graeme Gregory 07org.openembedded.dev * r1b5b4fd868 10openembedded.git/recipes/u-boot/ (u-boot-git/omapzoom2/inline-fix.patch u-boot_git.bb): u-boot_git.bb : newer gcc dont allow weak inlines Aug 18 10:24:50 he sneaky commits there Aug 18 11:00:01 XorA: btw: why you're using eglibc-2.11 for angstrom-next? why not 2.12.1? Aug 18 11:01:25 I am building sgml-common_0.6.3.bb file Aug 18 11:01:36 the glibc do_package gave me an error Aug 18 11:01:38 NOTE: Task failed: localedef returned an error (command was PATH="/home/nahsan/SB/open_comm/tmp/work/armv7a-angstrom-linux-gnueabi/glibc-2.9-r36.3/image/usr/bin:/home/nahsan/SB/open_comm/tmp/sysroots/i686-linux/usr/bin/armv7a-angstrom-linux-gnueabi:/home/nahsan/SB/open_comm/tmp/sysroots/i686-linux/usr/sbin:/home/nahsan/SB/open_comm/tmp/sysroots/i686-linux/usr/bin:/home/nahsan/SB/open_comm/tmp/sysroots/i686-linux/sbin:/home/nahsan/SB/open_c Aug 18 11:01:51 can someone tell me what is the problem Aug 18 11:04:25 OK i got it resolved i have modified bitbake.conf Aug 18 11:04:39 that is why it was giving me an error Aug 18 11:22:05 ok, not much gnome-* is building for me. gnome-doc-utils-native: http://tinderbox.openembedded.net/packages/700651/ Aug 18 11:23:17 but yet, it built fine yesterday. Aug 18 11:32:31 I think gnome-doc-utils-native has a dependency on gettext-native that isn't part of the recipe Aug 18 11:40:19 JaMa: it was still not the def_pref for any other distro Aug 18 11:40:32 JaMa: I assumed eglibc dudes were still working on it Aug 18 11:41:25 XorA: SHR is using 2.12 for some time.. only problem were those alignment traps (without -ggdb3) Aug 18 11:42:33 JaMa: good to know, angstrom-next is still in devel anyway, still a lot of gcc 4.5 problems to cure Aug 18 11:42:51 for example? Aug 18 11:43:13 we're also using gcc-4.5 for some time and nothing really bad happen here Aug 18 11:45:05 it looks like libidl-native has a dependency on bison-native. Aug 18 11:50:00 JaMa: main one, u-boot for beagelboard/omapzoom2.... is broken Aug 18 11:50:29 ok Aug 18 11:51:05 * JaMa switched his spitz builds from angstrom-2008 to angstrom-2010 Aug 18 11:58:25 jkridner|work: you're still supposed to be asleep Aug 18 12:03:42 useless facts: already 46 people donwloaded the bitbake 1.10 tarball from berlios Aug 18 12:07:13 eFfeM_work: when was it released ? Aug 18 12:07:22 aug 13 Aug 18 12:07:40 was announced on the ml Aug 18 12:22:22 I don't get this... http://tinderbox.openembedded.net/builds/83371/ shows do_fetch failing for files in the downloads dir. Aug 18 12:22:45 eFfeM_work: I'm almost always up at this time. Aug 18 12:23:36 jkridner|work: is there also the same file with .md5 extension in downloads dir? Aug 18 12:24:06 hm, I recall about 6 hrs ago you talked about scrollkeeper, must be a short night Aug 18 12:26:12 03Frans Meulenbroeks 07org.openembedded.dev * r3c30b1e718 10openembedded.git/recipes/i2c/i2c.bb: Aug 18 12:26:12 i2c: added LDFLAGS to cc line to fix hash QA Aug 18 12:26:12 Signed-off-by: Frans Meulenbroeks Aug 18 12:32:08 JaMa: yeah. Aug 18 12:33:23 ok, the one failing now is due to a md5sum mismatch... Aug 18 12:33:26 I heed help regarding sgml-common_0.6.3.bb Aug 18 12:33:30 * jkridner|work looks for the patch on the md5sum. Aug 18 12:33:48 I am compiling this recipe and glibc do_package fails Aug 18 12:33:57 giving error Aug 18 12:33:58 NOTE: Task failed: localedef returned an error (command was PATH="/home/nahsan/SB/open_comm/tmp/sysroots/i686-linux/usr/bin/armv7a-angstrom-linux-gnueabi:/home/nahsan/SB/open_comm/tmp/sysroots/i686-linux/usr/sbin:/home/nahsan/SB/open_comm/tmp/sysroots/i686-linux/usr/bin:/home/nahsan/SB/open_comm/tmp/sysroots/i686-linux/sbin:/home/nahsan/SB/open_comm/tmp/sysroots/i686-linux//bin:/home/nahsan/SB/open_comm//bitbake-1.8.18/bin:/usr/local/sbin: Aug 18 12:34:17 NOTE: generating locale en_BW (UTF-8) Aug 18 12:34:23 qemu-arm: relocation error: qemu-arm: symbol unlink, version GLIBC_2.0 not defined in file libc.so.6 with link time reference Aug 18 12:34:48 last 2 lines are before the Task failed message Aug 18 12:35:02 any hint what is going wrong? Aug 18 12:48:28 eFfeM_work: JaMa: any hint on the reported problem Aug 18 12:52:32 iirc someone said qemu-native was failing 10-15 percent of the times during locales build Aug 18 12:52:41 and he wanted assume_provided one Aug 18 12:52:51 check in the ML Aug 18 12:52:59 Noor: not really, I got similar error yesterday from eve on target :/ eve: relocation error: /usr/lib/libwebcore_efl.so.0: symbol floorf, version GLIBCXX_3.4.3 not defined in file libstdc++.so.6 with link time reference Aug 18 12:53:04 rather strange, though Aug 18 13:31:53 noor, no idea either, you're working with qemu ? Aug 18 13:41:20 no Aug 18 13:51:26 If I want to impact the mirrors (to add my own), do I have to edit base.bbclass -- or is there a way for me to put it in a conf file? Aug 18 14:22:33 jkridner|work: in local.conf you can say something like: Aug 18 14:22:34 PREMIRRORS_prepend = "(ftp|https?)$://.*/.* ftp://172.16.1.20/mirror/" Aug 18 14:23:03 thanks! Aug 18 14:23:07 yw Aug 18 14:31:35 eFfeM_work: is there anything similar for PSTAGE? Aug 18 14:32:07 morning Aug 18 14:32:14 morning kergoth_ Aug 18 14:32:22 jkridner, no idea (i only use pstage locally and then you can always just mount an nfs partition or so) Aug 18 14:32:24 hi kergoth Aug 18 14:33:19 anything similar to what? Aug 18 14:33:35 to the way you can setup downloads mirrors in local.conf. Aug 18 14:34:00 I'm wanting to setup a pstage mirror via http--look for the pstage file before attempting to build. Aug 18 14:34:26 I have been doing rsync, but I want to move it over to HTTP so others can use it to accelerate builds. Aug 18 14:34:51 I was checking out zsync. Aug 18 14:35:17 I didn't want to waste bandwidth with wget. also, I might want to avoid overwriting. Aug 18 14:35:23 If you set PSTAGE_MIRROR, it will try to fetch it from there Aug 18 14:35:41 and it will use bb.fetch.go to do it, which means it should attempt the normal mirrors stuff from that Aug 18 14:38:11 jkridner|work: what is the best full featured OE image target for the beagleboard? Aug 18 14:38:57 cbrake: today, beagleboard-linuxtag-image, but I have a patch pending to make beagleboard-demo-image be a superset of that. Aug 18 14:39:08 jkridner|work: I'm looking at doing weekly clean builds and then merging to a "testing" branch once it completes so people can always go to a known working version of OE (at least for a small number machines/images) Aug 18 14:39:41 sounds like a good plan. I don't think its particularly feasable to keep oe stable for everything, there's just too many damn combinations. but specific subsets.. Aug 18 14:39:49 if I get that patch accepted, then I'd say beagleboard-demo-image is the one I want to encourage to demonstrate Angstrom's abilities. Aug 18 14:40:01 * kergoth_ mutters Aug 18 14:40:08 sounds good, thanks Aug 18 14:40:21 does "testing" sound like a good branch name? Aug 18 14:40:28 we are even planning to ship a snapshot of Angstrom with the xM boards. (koen's influence at work) Aug 18 14:40:37 :-) Aug 18 14:41:19 * jkridner|work can't really comment on the branch name. Aug 18 14:50:34 03Lukas D. Gorris 07org.openembedded.dev * rfd13cdbe0b 10openembedded.git/conf/machine/include/htc-qsd8.inc: htc-qsd8: include xf86-input-tslib module for all machines Aug 18 15:14:38 oe will only be "stable" for stuff people use :) Aug 18 15:24:52 Crofton|work: agreed, and it it would be nice if it was at least stable for that :-) Aug 18 16:06:40 "what people use" depends on what subset of the oe userbase you're catering to. it's not feasable to even figure out what all our users are using, much less making sure it all works Aug 18 16:13:48 right Aug 18 16:25:35 03Frans Meulenbroeks 07org.openembedded.dev * r5cd41c31a6 10openembedded.git/recipes/iputils/iputils_20020927.bb: Aug 18 16:25:35 iputils, removed 2002 version Aug 18 16:25:35 Signed-off-by: Frans Meulenbroeks Aug 18 17:01:17 do we have enough users of gcc 4.5 ? I was planning to add some armv7 patch backports and linaro gcc patches Aug 18 17:01:31 may be hard-float abi :) Aug 18 17:02:11 but with day time job that will take a bit long Aug 18 17:02:23 JaMa is using it afaik Aug 18 17:03:11 cool so I have found a scape goat :) Aug 18 17:04:36 Runtime target 'virtual/arm-oe-linux-gnueabi-depmod-1.0' is unbuildable, removing Aug 18 17:07:08 found a pstaging problem with cmake, report is at ML Aug 18 17:08:38 seems get_kernelmajorversion('${PV}') is returning bogus values in this case Aug 18 17:08:59 virtual/${TARGET_PREFIX}depmod-${@get_kernelmajorversion('${PV}')} is in DEPENDS in kernel.bbclass Aug 18 17:10:57 does OESTATS slow my build down a lot? Aug 18 17:12:29 jkridner|work1: I don't think so, it just sends your log files to tinderbox Aug 18 17:13:25 check if tinderbox is up Aug 18 17:30:52 it is, it just seems ever since I added OESTATS that the build is slower. just perception I suppose. Aug 18 17:38:14 03Frans Meulenbroeks 07org.openembedded.dev * r1409e393ee 10openembedded.git/recipes/boost/boost.inc: Aug 18 17:38:14 boost.inc: added dependency on cmake-native Aug 18 17:38:14 Signed-off-by: Frans Meulenbroeks Aug 18 17:39:11 eFfeM, does the cmake depends for boost only matter for some versions? Aug 18 17:39:34 03Frans Meulenbroeks 07org.openembedded.dev * r2bef960ae3 10openembedded.git/recipes/cdparanoia/cdparanoia_svn.bb: Aug 18 17:39:34 cdparanoia: oe-stylized Aug 18 17:39:34 Signed-off-by: Frans Meulenbroeks Aug 18 17:40:49 Crofton, I'd say it is for all versions (although I must say I haven't really verified all versions, I found it by incident, building boost failed because cmake-native from pstage seems to be no good, I kinda assumed all versions of boost use cmake Aug 18 17:41:06 they used to use bjam to build boost Aug 18 17:41:12 they have cmake now? Aug 18 17:41:15 ah ok Aug 18 17:41:23 well I got a cmake error Aug 18 17:41:30 will verify Aug 18 17:41:34 be sure to see the comments in the older boost files Aug 18 17:41:46 don't worry too much, just wanted to make sure you new Aug 18 17:41:49 knew Aug 18 17:43:08 well the file in my downloads dir is called ../boost-1.41.0.cmake0.tar.gz Aug 18 17:43:27 but there is also ../boost-jam-3.1.16.tgz Aug 18 17:44:51 i did a build of beagleboard-demo-image after rm-ing tmp but keeping pstage which had some of the packages in it Aug 18 17:46:46 eFfeM: yours is finished?!? Aug 18 17:46:50 nope Aug 18 17:46:54 oh, ok. Aug 18 17:47:02 I want to know how to make my builds faster... Aug 18 17:47:09 i'm fixing the things that failed on me last night Aug 18 17:47:15 I'm renting a fast machine from Amazon right now to try to make my builds fast. Aug 18 17:47:57 http://tinderbox.openembedded.net/builds/83376/ has been running for about 4-1/2 hours! Aug 18 17:47:58 jkridner|work1: octa-core, 16 GB ram and maybe an SSD Aug 18 17:48:35 I'm using hyperthreading quadcore and 17GB RAM. Aug 18 17:48:43 I started mine yesterday 17.30 my time it was probably halfway at 23.00, not sure when it failed Aug 18 17:48:47 m1.xlarge EC2 instance. Aug 18 17:48:49 That's probably fine Aug 18 17:49:02 We don't scale that well right now Aug 18 17:49:06 jkridner|work1: that does not sound too bad. Aug 18 17:49:10 past 8 threads and 20 make -j's Aug 18 17:49:33 jkridner|work1: you could consider putting your tmp dir in ram, but then be sure to add rm-work Aug 18 17:49:54 Won't help Aug 18 17:49:59 We're just too bottleneckable Aug 18 17:50:14 ah ok Aug 18 17:50:39 Tartarus: will you be reviewing/pushing the patches from your collegues? Aug 18 17:50:41 hm? of course building in ram will help, at least for disk intensive tasks. not sure how significant the improvement would be, but to say it wouldn't help is a bit much Aug 18 17:51:03 eFfEM: kergoth and I gave them an initial going over and then told them to start working with the community Aug 18 17:51:04 I asked them to look at merging native and nonnative too while they were at it Aug 18 17:51:11 Is there more you'd like to see them do first before publishing? Aug 18 17:51:29 eFfeM: What happened was I gave them a list of every -native with a legacy do_stage Aug 18 17:52:00 ah ok, well I've been coaching them a little bit, and earlier today I got kind of a reply like "Tom gave us this list of recipes to do" Aug 18 17:52:22 eFfeM: I had done that, but ran out of space when doing the full beagleboard-demo-image... Aug 18 17:52:27 so thought perhaps those were recipes special for a project in mentor Aug 18 17:52:33 Nope Aug 18 17:52:36 bitbake -k world :) Aug 18 17:52:43 also, the speedup wasn't what I expected... I still seemed to get blocked on I/O (according to vmstat) and I have no idea how. Aug 18 17:52:57 Tartarus: sst. koen may read the logs Aug 18 17:53:19 jkridner disabling insane also will help a little bit Aug 18 17:53:33 what is 'insane'? Aug 18 17:53:33 * eFfeM got blamed by mailing results of bitbake -k world to the ML Aug 18 17:53:38 jkridner|work1: Yeah, it's really annoying that we just stop scaling at some point :( Aug 18 17:53:44 insane does the qa checking Aug 18 17:53:54 eFfeM: the qa tasks don't take long Aug 18 17:54:14 * jkridner|work1 would love to figure out how to cluster-build Angstrom. Aug 18 17:54:17 My SWAG is that python threading sucks Aug 18 17:54:21 from some other stuff I was reading Aug 18 17:54:39 we aren't using python threading Aug 18 17:54:41 jkridner|work1: I haven't personally scaled the icecc stuff, Roman did Aug 18 17:54:46 And that seems to be a good solution for him Aug 18 17:54:59 Not sure if it really wins out over a single machine tho Aug 18 17:55:09 Tartarus: true, but every penny counts :-) actually what I dislike about insane is that even for console only builds it requires all kind of glib stuff being build for .desktop testing Aug 18 17:55:31 it probably wouldn't be that difficult to teach bitbake to offload task executions to other machines, assuming a shared tmpdir Aug 18 17:55:42 eFfeM: I'll give you the glib stuff, but no, the actual tasks are noise :) Aug 18 17:55:44 now if you wanted to offload entire sections of the task graph, slightly more complex, but still not too bad Aug 18 17:55:50 (depending on what portions you want to offload) Aug 18 17:56:20 yeah the task is not too exciting Aug 18 17:56:20 RP's commit in poky that splits out task execution into a separate script would make it easier Aug 18 17:56:54 kergoth haven't seen it but if it does what I think it does, it definitely would help Aug 18 17:57:10 * eFfeM expects that the temp scripts are more disjunct and simpler Aug 18 17:57:25 http://git.pokylinux.org/cgit.cgi/poky-contrib/commit/?h=rpurdie/scenequeue&id=c5cddecd20a9d6610bcede096381629487115e54 Aug 18 17:59:05 looks interesting, but I have too limited python knowledge and internals knowledge to be able to say this is ok Aug 18 17:59:22 its a step in the right direction Aug 18 18:00:47 afk for 30 mins Aug 18 18:29:42 03Frans Meulenbroeks 07org.openembedded.dev * r1f4978b06b 10openembedded.git/recipes/gnome/libgnomecanvas_2.26.0.bb: Aug 18 18:29:42 libgnomecanvas: added dependency on xineramaproto Aug 18 18:29:42 Signed-off-by: Frans Meulenbroeks Aug 18 18:29:44 03Frans Meulenbroeks 07org.openembedded.dev * r3857d5d3f5 10openembedded.git/recipes/librsvg/librsvg_2.26.2.bb: Aug 18 18:29:44 livrsvg: added dependency on xineramaproto Aug 18 18:29:44 Signed-off-by: Frans Meulenbroeks Aug 18 18:29:46 03Frans Meulenbroeks 07org.openembedded.dev * r0f9da77ece 10openembedded.git/recipes/libgpod/usbmuxd_1.0.3.bb: Aug 18 18:29:46 usbmuxd: added dependency on cmake-native Aug 18 18:29:46 Signed-off-by: Frans Meulenbroeks Aug 18 18:39:35 anybody tried to build firefox lately ? Aug 18 18:39:38 | configure: error: Enchant library not found or too old. Use --disable-spell to build without spell plugin. Aug 18 18:40:24 sounds like a missing dep Aug 18 18:40:31 FF has those from time to time :( Aug 18 18:42:55 just saying Aug 18 18:44:27 yeah, me too :) Aug 18 18:45:36 re Aug 18 18:47:06 * khem tries bitbake world Aug 18 18:49:08 khem, -k, i hope :) Aug 18 18:49:35 kergoth_: I am being brave I know :) Aug 18 18:49:48 kergoth_: that RP commit is nice Aug 18 18:49:53 khem, good luck (and have patience) Aug 18 18:49:59 subprocess for forking tasks Aug 18 18:50:32 eFfeM: I am getting a monster machine soon so preping for that :) Aug 18 18:50:39 than I will have regular bitbake world runs Aug 18 18:50:44 i was thinking about playing with the multiprocessing module in 2.6 for managing our pool of processes, but I don't know how beneficial itd be if we do the bitbake-runtask approach Aug 18 18:50:55 because bitbake x11-image will only take 90 mins Aug 18 18:51:05 this box will need serious stuff to buiild Aug 18 18:51:53 heh Aug 18 18:51:56 So speaking of world -k Aug 18 18:52:06 Can we just go and assume the new checksums for various things are right Aug 18 18:52:09 for silly or annoying reasons Aug 18 18:52:31 kergoth_: does multiprocessing module use some scheduling Aug 18 18:52:52 Tartarus: you mean avoid the cut copy paste ?? Aug 18 18:53:02 its a one time op Aug 18 18:53:51 khem: No, i mean the recipes JaMa, iirc, added the new checkums to, commented out Aug 18 18:54:12 eFfeM: we have reduced recipes to ~7600 from ~8500 Aug 18 18:54:31 is that true ? or something wrong with my filters Aug 18 18:55:17 Tartarus: IIUC new checksum is commented out ? Aug 18 18:55:43 ERROR: Required build target 'evolve' has no buildable providers. Aug 18 18:55:51 Missing or unbuildable dependency chain was: ['evolve', 'etk'] Aug 18 18:55:54 khem: git grep "#CHECKSUMS.INI MISMATCH" recipes/ Aug 18 18:56:10 since those will kill bitbake -k, even w/ loose checksum variable set Aug 18 18:56:16 I modified that locally but wasn't sure if it was right Aug 18 18:57:14 I think commented ones should be punted out Aug 18 18:57:45 punted how? Aug 18 18:58:03 remove those Aug 18 18:59:21 switched to or removed? Aug 18 18:59:30 I found them to be the now correct sums, as of maybe 2 months ago Aug 18 18:59:34 enchant library needs bump. current version breaks firefox and gedit Aug 18 19:01:00 we need some gnome quality team :) Aug 18 19:01:15 03Tom Rini 07org.openembedded.dev * rbcd0324047 10openembedded.git/recipes/dt/dt_15.14.bb: Aug 18 19:01:15 dt: Update checksums Aug 18 19:01:15 Upstream confirmed to me that they just overwrote the old one Aug 18 19:01:15 and didn't change version (but did make some actual code changes) Aug 18 19:01:15 and in the future just might put a version string on the archive. Aug 18 19:01:15 Signed-off-by: Tom Rini Aug 18 19:07:11 khem i think your figures are about right Aug 18 19:08:06 7675 recipes says mine, thought i started with 8415 or so but could also be 8504 or so Aug 18 19:08:19 and the # will reduce with the cleanup of jama Aug 18 19:09:11 Tartarus: that is very nasty. because everyone with the old file in downloads will get an error Aug 18 19:09:20 we had something before with a python module Aug 18 19:09:51 Yes Aug 18 19:09:55 But the # of dt users is small Aug 18 19:09:56 for that one koen made a mirror of the old one Aug 18 19:10:01 And it's been wrong for about a year Aug 18 19:10:06 ah ok Aug 18 19:10:22 * eFfeM admits now to know what dt is Aug 18 19:10:39 It's a really nice tool for verifying your storage driver is sane Aug 18 19:12:56 ah cool Aug 18 19:54:29 eFfeM: nice job. how many of native recipes are there which can be converted to BBCLASSEXTEND Aug 18 19:54:36 will reduce the number as well Aug 18 20:09:46