**** BEGIN LOGGING AT Mon Jul 26 02:59:57 2010 Jul 26 03:37:20 hi all Jul 26 03:48:12 I was trying to make abuild of the fso-illume-image using bitbake but it returns few errors.. Jul 26 03:48:48 it returns error when trying to build the gtk Jul 26 03:51:52 i googled for the possible reason and i could run "bitbake -crebuild gtk+" succesfully but still when i run "bitbake fso-illume-image" it returns the same error Jul 26 03:53:14 here is the respective line in the log file http://pastebin.com/A8ALMrnS Jul 26 04:04:04 ranjith: you might have to disable using version script Jul 26 04:04:36 khem: actually i dint get what you mean.. Jul 26 04:04:41 -version-script -Wl,.libs/libgdk_pixbuf-2.0.ver Jul 26 04:04:47 ok Jul 26 04:04:49 this line in linker commandline Jul 26 04:05:05 you have to patch the makefile probably Jul 26 04:05:30 do you mean i ahve to edit the make file and change the respective portion Jul 26 04:05:31 ? Jul 26 04:09:30 khem: do i ahve to edit the makefile and add the line "-version-script -Wl" or do i need to do something else..? Jul 26 04:12:42 you have to remove it Jul 26 04:12:52 but it is not there.. **** BEGIN LOGGING AT Mon Jul 26 05:00:15 2010 Jul 26 05:48:17 cd Jul 26 05:49:30 gm Jul 26 06:58:31 hi guys Jul 26 06:58:59 is there a command which automatically gets rid off all old packages from the workdir? Jul 26 06:59:10 like ones which have been replaced by newer versions Jul 26 06:59:46 good morning Jul 26 07:48:42 ~seen hrw Jul 26 07:48:44 hrw <~hrw@conference/ubuntu/x-lhcifvzehjgrsubw> was last seen on IRC in channel #oe, 2d 18h 53m 40s ago, saying: 're'. Jul 26 07:51:49 morning Jul 26 07:51:53 hi xora Jul 26 07:52:06 xora are you stable maintainer? Jul 26 07:52:06 * XorA is back from Ubuntu land Jul 26 07:52:11 woglinde: no Im not Jul 26 07:52:58 ~seen likewise Jul 26 07:52:59 likewise <~likewise@82-170-243-215.ip.telfort.nl> was last seen on IRC in channel #oe, 16d 18h 22m 21s ago, saying: 'eFfeM: nope, but about to do so before monday'. Jul 26 07:56:09 woglinde: looking for an ack? Jul 26 07:56:21 mckoan right Jul 26 07:56:25 my java patches Jul 26 07:56:27 for stable Jul 26 07:56:39 hrw gave me his ack per IRC Jul 26 07:56:52 mckoan beware that I needed to resend 2 patches Jul 26 07:57:00 woglinde: hrw was in Prague, he should be back today I think Jul 26 07:57:16 xora I know Jul 26 07:57:25 * XorA got back late last night :_) Jul 26 07:58:01 woglinde: tell me exacly what patch you need to ack Jul 26 07:58:30 [oe] [PATCH 00/21] second round of patches needed for openjdk for oe-stable Jul 26 07:58:45 [oe] [PATCH 0/2] updated two patches for second round of openjdk stable transition Jul 26 07:58:53 morning hrw Jul 26 07:58:57 morning Jul 26 07:59:12 hey hrw Jul 26 07:59:39 XorA: weekend weather was afwul Jul 26 07:59:47 hrw: yeah :-) Jul 26 07:59:59 hrw would be nice to get the ack's for the stable java patches Jul 26 08:00:02 hrw: designed to cook us in conference then soak us when not Jul 26 08:00:12 XorA: btw - place where we ended on Wed has awesome food Jul 26 08:00:37 woglinde: I do not run OE builds so do not know does it builds or not Jul 26 08:01:19 hrw hm I and stefan tested it Jul 26 08:01:50 hrw: cool, I just kept choosing places at random but everywhere seemed to be good Jul 26 08:01:54 only problem are segfaults with jamvm Jul 26 08:01:59 woglinde: remind me in half hour - I should be after system refresh then Jul 26 08:02:04 but we are trying to work it out Jul 26 08:02:11 okay Jul 26 08:02:12 woglinde: I leave this burden to hrw Jul 26 08:02:20 mckoan I neead two ack's Jul 26 08:02:24 as you know Jul 26 08:02:35 mckoan: you basically prefer to avoid acking patches Jul 26 08:03:40 week without doing updates = 550MB to fetch Jul 26 08:05:05 woglinde: 21 patches + 2 patches ? Jul 26 08:05:32 mckoan 21 patches 2 updated Jul 26 08:41:30 morgen Jul 26 08:43:10 hrw: bon jour Jul 26 08:47:09 hello Jul 26 08:48:15 does anyone knows why the same patches are applied to libtool-native and libtool-cross ? Jul 26 08:48:22 s/knows/know/ Jul 26 08:56:49 hi zecke Jul 26 08:57:02 woglinde: hey Jul 26 08:57:13 woglinde: did the t-shirt fit? probably not, did you even try? Jul 26 08:57:22 sorry didnt try yet Jul 26 08:57:26 woglinde: any success with an alternative bootloader? Jul 26 08:57:35 not tested yet Jul 26 08:57:45 woglinde: I can encode one ASN1 string, but my phones don't like it. Jul 26 08:57:58 zecke haha Jul 26 09:40:37 03Koen Kooi  07org.openembedded.dev * r76da91238b 10openembedded.git/recipes/linux/linux-omap-psp-2.6.32/beagleboard/configs/cpuidle-gether: linux-omap-psp 2.6.32: add multiconfig for beagle with cpuidle and g_ether built in Jul 26 10:02:17 good morning Jul 26 10:03:36 hi florian Jul 26 10:03:41 florian: good morning Jul 26 10:03:59 khem: I just tested a simple workaround for the armv4t SDK compiler issue successfully. Jul 26 10:04:39 hi eFfeM_work, hi pb_ Jul 26 10:07:46 Which OE environment variable contains the target binary architecture? Jul 26 10:10:26 florian: I think it is HOST_ARCH Jul 26 10:10:33 (no kidding) Jul 26 10:11:47 I shoulld make a bitbake -e once this build has finished Jul 26 10:11:59 actually looked at the code and I seem to be wrong, it is HOST_SYS Jul 26 10:12:13 COMPATIBLE_HOST compares against HOST_SYS Jul 26 10:13:46 FEED_ARCH contains it too.... Jul 26 10:13:47 btw just read your native prefs email and I think indeed that the same rules should apply and 2.24.1 should win Jul 26 10:14:28 (but, as is well known, I have no objections nuking the old recipes) Jul 26 10:15:04 effem hehe find a security bug Jul 26 10:15:11 thats a good reason Jul 26 10:15:21 Someone mentioned that there was a discussion about this before but I didn't find anything... Jul 26 10:15:49 florian: on native prio? not that I am aware Jul 26 10:15:57 but for dbus-glib I converted the old recipes too Jul 26 10:16:42 eFfeM_work: yes Jul 26 10:17:26 woglinde: do i hear a volunteer to do 1100+ recipes ? Jul 26 10:17:29 :-) Jul 26 10:19:28 * florian wonders why we build gtk+-native for console-image Jul 26 10:19:34 effem hehe I started Jul 26 10:20:17 woglinde: so did I by fixing cpan.bbclass, but that one is at least not legacy Jul 26 10:20:19 florian: I found plenty of reasons ;) Jul 26 10:20:20 florin check the deps Jul 26 10:20:44 florian: why is glib build when I do a bitbake virtual/kernel ? Jul 26 10:21:14 florian: one of them was the networkmanager, which despite beeing an headless service, includes gnome dependencies Jul 26 10:21:22 bluez Jul 26 10:21:29 (actually eFfeM knows: this is because insane.bbclass has a dep to it in order to be able to check if .desktop files are ok) Jul 26 10:25:22 florian: in fact console-image is too big to be a console only Jul 26 10:26:15 heh Jul 26 10:27:01 I'm waiting for a Sim.ONE console image to finish... I bet it is too big for its flash. Jul 26 10:27:32 anyone seen bitbake making the shell quit when it terminates? Jul 26 10:29:23 mike_cw no Jul 26 10:30:49 started doing it a couple of weeks ago, but my OE hash and bitbake version are locked so it must have been something else that changed Jul 26 10:31:14 when bitbake stops on error it makes the whole shell quit. running from a text console it will log me out Jul 26 10:34:30 never happened here Jul 26 10:35:42 ok... like I guessed: its too big Jul 26 10:35:59 florian minimal-image? Jul 26 10:36:07 kexecboot-image? Jul 26 10:36:45 I want to call a phy_driver function from a user space application, how do i do this ? Jul 26 10:36:45 woglinde: yes that's the next candidate Jul 26 10:36:48 in c++ Jul 26 10:37:19 melchior read the topic Jul 26 10:37:33 phy_driver sounds like kernel stuff Jul 26 10:37:47 you normaly dont have access from userspace to it Jul 26 10:38:06 that was enough over topic from me Jul 26 10:38:42 might be a topic for #edev but the idea sounds wrong anyway Jul 26 10:40:25 ok thank you Jul 26 11:09:51 hi can anyone help me to know how to get pass the error of gtk+ in buitbake fso-illume-image Jul 26 11:10:14 ranjith you got an answer on the ml Jul 26 11:10:22 I was running bitbake fso-illume-image but it returned an error that do_compile on gtk+ failed Jul 26 11:10:41 woglinde: hi i got the answer but its not working Jul 26 11:10:55 woglinde: I am already using the bash Jul 26 11:11:26 and also tried cleanign the version file but the file is created everyonce the bitbake fso-illume-image is run Jul 26 11:11:33 and the same errror comes Jul 26 11:11:58 ranjith are you using stable or dev? Jul 26 11:12:07 which one .. Jul 26 11:12:12 oe Jul 26 11:12:18 what else? Jul 26 11:12:31 i have pulled it from git Jul 26 11:13:20 type git branch Jul 26 11:13:24 and paste it Jul 26 11:13:42 * fso/milestone5.5 Jul 26 11:13:42 org.openembedded.dev Jul 26 11:14:01 aeh Jul 26 11:14:03 its fso/milestone5.5 Jul 26 11:14:49 wau thats old Jul 26 11:15:01 on which device you want to run your software? Jul 26 11:15:09 openmoko Jul 26 11:15:30 I talked about device not a company Jul 26 11:15:43 openmoko has 2 devices Jul 26 11:15:50 actually i was trying to learn openembedded .. and i got the instructions from some documentation in the openembedded site Jul 26 11:15:50 neo and freerunner Jul 26 11:16:08 who told you to checkout fso branch? Jul 26 11:16:22 it was intructed in the sute Jul 26 11:16:41 do you have a pointer? Jul 26 11:16:57 I suggest use open embedded dev branch Jul 26 11:17:03 and shr distritbution Jul 26 11:17:04 just a minute i shall get it now Jul 26 11:17:35 rm your oetmp Jul 26 11:17:41 and start a clean build Jul 26 11:18:02 http://wiki.openmoko.org/wiki/OpenEmbedded Jul 26 11:18:23 this is the link form where i got the instructions Jul 26 11:19:02 hm okay I see Jul 26 11:19:14 but there is no one left working on this branch Jul 26 11:19:25 best distribution for freerunner is now shr Jul 26 11:19:32 did you mean to clean the bitbake fso-illume-image using bitbake fso-illume-image -c clean Jul 26 11:19:45 and do the same build again..? Jul 26 11:20:05 not Jul 26 11:20:09 checkout oe-dev Jul 26 11:20:20 ok Jul 26 11:20:21 then Jul 26 11:20:23 change in local.conf distro to shr Jul 26 11:20:28 rm oetmp Jul 26 11:20:38 oetmp ..? Jul 26 11:20:40 which you set in local.conf Jul 26 11:20:50 i ahven't seen anything in that name Jul 26 11:20:54 ok Jul 26 11:30:25 which is the url for oe-dev branch Jul 26 11:33:16 woglinde: how can i change to oe-dev i just ran git checkout oe-dev and it changed to branch oe-dev but i need to give url for git pull what is the url to be given ..? Jul 26 11:35:16 none Jul 26 11:35:27 whats git branch telling you now? Jul 26 11:43:20 after git checkout oe-dev Jul 26 11:43:25 i did git pull Jul 26 11:43:31 it gave this error http://pastebin.com/v05A1Qus Jul 26 11:43:40 so which url i hvae to give here ? Jul 26 11:44:10 please type git checkout org.openembedded.dev Jul 26 11:44:15 after this Jul 26 11:44:19 git pull --rebase Jul 26 11:45:51 ok Jul 26 11:46:23 i ahd changed one file in Jul 26 11:46:30 so its showing as an error Jul 26 11:46:38 error: You have local changes to 'recipes/freesmartphone/cornucopia.inc'; cannot switch branches. Jul 26 11:46:54 this is the erro i get when i give git checkout org.openembedded.dev Jul 26 11:46:59 sure Jul 26 11:47:13 you changed something in this file Jul 26 11:47:19 either you check it in Jul 26 11:47:22 git diff to view Jul 26 11:47:27 ok Jul 26 11:47:31 or do git checkout recipes/freesmartphone/cornucopia.inc Jul 26 11:47:41 or git reset --hard :) Jul 26 11:48:04 but all changes will be lost then Jul 26 11:48:33 btw, do we have 'workflow' sample on wiki? Jul 26 11:49:01 i.e. pull, branch, work, commit, rebase, merge, push Jul 26 11:52:07 if i am going to run bitbake with distro as shr do i need to switch to origin/shr/testing2010 Jul 26 11:52:10 ? Jul 26 11:52:18 noz at the moment Jul 26 11:52:27 use plain oe-dev Jul 26 11:52:32 it should work fine Jul 26 11:53:15 ok Jul 26 11:55:16 now i ran bitbake fso-illume-image but it says there is no distro in the name "shr" Jul 26 11:55:42 also some sanity check error which i think can be resolved by switching the sanity check off Jul 26 11:57:13 whats bitbake --version saying? Jul 26 11:57:33 1.8.18 Jul 26 11:57:46 BitBake Build Tool Core version 1.8.18, bitbake version 1.8.18 Jul 26 11:59:34 okay Jul 26 12:00:04 so what about the error that says no distro named "shr" Jul 26 12:00:43 please look into conf/distro/shr.conf Jul 26 12:00:48 is that file present? Jul 26 12:00:59 ok Jul 26 12:01:38 no its not there.. Jul 26 12:02:05 please type git branch Jul 26 12:02:40 fso/milestone5.5 Jul 26 12:02:41 * oe-dev Jul 26 12:02:41 org.openembedded.dev Jul 26 12:02:46 *sigh* Jul 26 12:02:54 please type git checkout org.openembedded.dev Jul 26 12:02:59 git pull --rebase Jul 26 12:03:40 yes done it Jul 26 12:04:02 no check again if the shr.conf file is there Jul 26 12:04:11 now shr is there. Jul 26 12:04:23 so can you pls let me know what happened before..? Jul 26 12:04:27 so start your build Jul 26 12:04:31 ok Jul 26 12:04:42 you made a branch from fso/milestone-5.5 Jul 26 12:04:42 thanks for the help and patience Jul 26 12:04:50 ok Jul 26 12:04:51 called oe-dev Jul 26 12:05:18 ok now got it.. so what i did was after getting the new version again changed to the older branch Jul 26 12:05:28 thanks once again Jul 26 12:17:27 ping hrw java-patches Jul 26 12:18:19 pong Jul 26 12:46:33 03Xerxes Rånby  07org.openembedded.dev * r929ace6062 10openembedded.git/recipes/openjdk/openjdk-6-common.inc: Jul 26 12:46:33 openjdk-6-6b18: Fix cacao on soft-float systems. Jul 26 12:46:33 * openjdk-6-common.inc: Configure cacao with --enable-softfloat Jul 26 12:46:33 on systems that are not equipped with a FPU. Jul 26 12:46:33 This fixes openjdk + cacao on ARMv4 and ARMv5 systems. Jul 26 12:46:33 Bumped PR. Jul 26 12:46:36 03Xerxes Rånby  07org.openembedded.dev * ra6125ce312 10openembedded.git/recipes/openjdk/openjdk-6-common.inc: openjdk: Bump INC_PR after merge. Jul 26 12:46:38 mickey|office: just read your email on native preferences, somehow i think this is not going to convince koen Jul 26 12:50:22 heh Jul 26 12:52:03 florian: tried lzma-kernel to shrink your image? Jul 26 12:52:31 wtf Jul 26 12:52:46 I'm trying to add my patches to the opkg package Jul 26 12:52:53 the do_patch step fails Jul 26 12:53:08 your package is wrong Jul 26 12:53:19 NOTE: Applying patch 'Fix-missing-include-in-opkg-headers.patch' (openembedded/recipies/opkg/files/Fix-missing-include-in-opkg-headers.patch) Jul 26 12:53:22 ant_work: not for the simone yet but its a good idea anyway Jul 26 12:53:24 NOTE: Task failed: [Errno 17] File exists Jul 26 12:53:51 florian: as reference we went from 1,2 to 850 kb Jul 26 12:54:08 woglinde: how could he be wrong ? Jul 26 12:54:22 ant_work: that's quite impressive... for whcih machine? Jul 26 12:54:26 bzaurus Jul 26 12:55:16 just a bit slower boot Jul 26 12:55:25 tolerable Jul 26 12:55:58 ant_work: did you count no CONFIG_PRINTK also? Jul 26 12:56:16 ah, well, the 30% saving is indipendent Jul 26 12:56:50 dj-death> NOTE: Task failed: [Errno 17] File exists Jul 26 12:57:17 woglinde: what's wrong in this : http://pastebin.com/wuVGL3xb Jul 26 12:57:58 dj-death the patch Jul 26 12:58:18 it's not Jul 26 12:58:25 try to apply it without oe Jul 26 12:58:28 I can apply it by hand Jul 26 12:59:00 even within the OE buildtree Jul 26 12:59:14 after OE has unpacked the package Jul 26 12:59:57 dj-death: try bitbake -c clean and bitbake -c patch Jul 26 13:00:46 heh... Jul 26 13:00:58 minimal <> angstrom 'war' is boring Jul 26 13:01:17 hrw nobody will fix it Jul 26 13:01:35 'my distro does not pin version and breaks' is a bug for minimal distro users and *they* should fix it Jul 26 13:01:48 because prober solution is to switch all to BBCLASSEXTENDED Jul 26 13:02:17 mickey|office: http://4mat.bandcamp.com/album/decades Jul 26 13:03:52 hrw: I don't see it like this... OE introduced a special preference for the old-style native packages. This is not what you would expect, no benefit for anyone and broken by design. Jul 26 13:04:34 OE or bitbake? Jul 26 13:04:39 hrw: apart from this is it going to break any distribution for somme compatibility reason some day unless you pin everythig that has a native package left Jul 26 13:05:06 git pull;git tag "oe-before-massive-breakup'; do ugly stuff;git tag "oe-after-massive-breakup-all-can-break-but-thats-dev";git push Jul 26 13:05:10 ant_work: better use -D Jul 26 13:05:25 ant_work: and see from where the patch is applied Jul 26 13:05:35 dj-death: just wondering you have cruft in your workdir Jul 26 13:05:42 going through discussions is what we have since BBCLASSEXTENDED was invented Jul 26 13:05:53 Poky did migration in short time Jul 26 13:06:08 OE lacks dictator which does not fear Koen Jul 26 13:06:19 dj-death: thus -c clean Jul 26 13:06:22 hrw? Jul 26 13:06:28 oe lacks manpower Jul 26 13:07:01 woglinde: we always lacked it Jul 26 13:07:10 ant_work: cruft ? Jul 26 13:07:32 yes, "File exists" Jul 26 13:07:49 woglinde: let someone kill recipes, merge blindly BBCLASSEXTEND stuff from Poky, push all that. things will break but reverts should be rejected in favour of fixing things Jul 26 13:07:50 let's try Jul 26 13:09:35 woglinde: OE lacks manpower and starts to loose place on market. During ubuntu/linaro platform sprint I heard vendors moving from OE/Poky to Ubuntu for board support Jul 26 13:10:19 does ubuntu provides cross compilation support ? Jul 26 13:10:31 dj-death: slowly but it does Jul 26 13:10:47 dj-death: few persons (including me) are working on it Jul 26 13:11:41 Poky and OE both give a chance for faster development then Ubuntu due to less maintainers with which you need to discuss changes Jul 26 13:11:42 * ant_work thinks at least .dev branch should be clean of legacy metadata Jul 26 13:11:59 thing which take 1h in OE can take days in Ubuntu Jul 26 13:12:37 usecase: "bitbake umba" fails for me. I do patch, merge it into metadata, push, build. work done in ~1h Jul 26 13:13:23 In ubuntu: "dpkg-buildpackage umba" fails for me. I do patch, submit bug, pester developer, wait for new package release. Jul 26 13:14:18 and do not forget about releaseschedule in ubuntu - patch after freeze needs to wait months Jul 26 13:14:18 I think we are discussing about two different things... Jul 26 13:14:34 sure Jul 26 13:14:57 How dev should look like and that we need to get rid of the old stuff is one thing. Jul 26 13:15:16 But my point is that we introduced a bug that needs to be fixed. Jul 26 13:15:28 and question is who will make it? Jul 26 13:15:45 i hardly dare to say it, but I will do: rm recipes/*/*native*bb Jul 26 13:15:46 :p Jul 26 13:16:13 that's a workaround but not a fix Jul 26 13:16:19 agree Jul 26 13:16:56 the question is why do we have so many versions Jul 26 13:16:58 I'm not sure where the problem is, koen mentioned it not being a bug in bb 1.10 (or did I misunderstand that) Jul 26 13:17:01 hrw so you mean I can add your ack to the java-patches and push them to oe-stable *g* Jul 26 13:17:27 *gg* Jul 26 13:17:42 ant_work: http://pastebin.com/MC7f3yRQ Jul 26 13:17:53 woglinde: for your java changeset I gave you Ack last week Jul 26 13:18:16 ant_work: -D didn't help to know from where the patch is applied Jul 26 13:18:17 hrw hehe I thought you wanted made it official Jul 26 13:18:21 okay Jul 26 13:18:41 will mail Jul 26 13:18:44 moment Jul 26 13:20:53 woglinde: sent official, gpg signed email Jul 26 13:21:49 mayn thanks Jul 26 13:24:38 dj-death: can't find file to patch at input line 5 Jul 26 13:26:00 copy the file in the recipe dir Jul 26 13:26:28 and try to patch it manually again Jul 26 13:26:39 btw you haven't posted the patch ;) Jul 26 13:28:39 ant_work: http://groups.google.com/group/opkg-devel/browse_thread/thread/1366ad9898c1f082 Jul 26 13:28:43 ant_work: here it is Jul 26 13:29:38 seems malformed patch Jul 26 13:29:57 try to rediff it in place Jul 26 13:30:01 I dropped the top of the patch Jul 26 13:30:08 diff - aburN or alike Jul 26 13:30:26 but once again, it applies nicely if I do it manually Jul 26 13:30:50 unless OE is not using patch -p1 ... Jul 26 13:31:02 dj-death go to oetmp/work/arch/blahkackage/blackpackage Jul 26 13:31:05 quilt push Jul 26 13:31:08 if that works Jul 26 13:31:45 hrw hm mail seems to be borken Jul 26 13:32:13 in /opt/OE/test-dev/work/armv7a-angstrom-linux-gnueabi/opkg-nogpg-0.1.8-r1 Jul 26 13:32:29 one deeper Jul 26 13:32:34 I have these directories : opkg-0.1.8 opkg-nogpg-0.1.8 temp Jul 26 13:32:40 ah Jul 26 13:32:42 thats it Jul 26 13:32:44 it might explain my problem ;) Jul 26 13:32:50 wrong directory-dir Jul 26 13:32:51 yes Jul 26 13:32:56 ok Jul 26 13:33:20 didnt checked that Jul 26 13:33:25 I'm not sure to understand how OE choose its work directory Jul 26 13:33:26 you have to set S Jul 26 13:33:31 hmm ok Jul 26 13:33:39 that's the line I dropped Jul 26 13:33:44 hehe Jul 26 13:34:41 what's the workdir by the way ? the directory used to unpack the sources ? or the directory used to patch ? Jul 26 13:35:29 woglinde: so it looks like you got whole series Ack from me and Koen Jul 26 13:35:50 hrw yes Jul 26 13:35:53 ;) Jul 26 13:37:05 okay I will push it now Jul 26 13:37:44 woglinde: thx, it's working fine now Jul 26 13:38:40 khem, ping Jul 26 13:38:56 hi blindvt Jul 26 13:39:08 woglinde, hi Jul 26 13:40:03 is there anything i could do to get the non-controversial parts of 'gcc-cross: fix PV and make it compile' and the other 2 patches in that series applied? Jul 26 13:40:16 aeh? Jul 26 13:40:51 woglinde, http://www.mail-archive.com/openembedded-devel@lists.openembedded.org/msg07942.html Jul 26 13:41:53 woglinde, i don't mean to sound impatient but.. :) Jul 26 13:41:55 ah this one Jul 26 13:43:00 03Koen Kooi  07org.openembedded.dev * r181244a3a6 10openembedded.git/recipes/ncurses/ncurses.inc: ncurses: proPagate is with a p, not a g Jul 26 13:44:40 you mean micro and minimal fix? Jul 26 13:46:01 woglinde, yes Jul 26 13:46:50 hrw: (4mat) pretty cool :) Jul 26 13:46:53 woglinde, plus the ppc fix Jul 26 13:47:58 (i.e. just leave out touching SRCREV and fiddling with PR_append) Jul 26 13:48:34 damn how the hell the wrong patch got in Jul 26 13:49:47 hm? one of mine? Jul 26 13:49:53 woglinde: shouldn't have been tested before ack? Jul 26 13:49:56 no mine Jul 26 13:50:05 mckoan I had Jul 26 13:50:35 but seems git fooled me Jul 26 13:52:45 woglinde: happens ;-) Jul 26 13:53:56 mckoan its only cosmetic Jul 26 13:54:02 so I will leave it Jul 26 13:54:27 woglinde: good for you Jul 26 13:54:31 yes Jul 26 13:54:54 woglinde: I simply wonder why nobody claimed such problem Jul 26 14:02:25 mickey|office: good afternoon Jul 26 14:24:17 please tell me, when i am configuring "qt-everywhere-opensource-src-4.6.2-x11" in openembedded. then it gives error. Jul 26 14:24:55 commnad and error in http://pastebin.com/EGizJ4Yw Jul 26 14:25:10 hi stefan_schmidt Jul 26 14:25:31 hi woglinde Jul 26 14:25:40 akumar -> The EGL functionality test failed! Jul 26 14:25:50 whats so hard to understand that Jul 26 14:26:30 but i have EGL Jul 26 14:31:50 woglinde: I have also adjusted the PATH for EGL, QMAKE_INCDIR_EGL and QMAKE_LIBDIR_EGL. Jul 26 14:32:37 akumar why you dont use the recipe? Jul 26 14:33:28 * mckoan sometimes must brush up his skills of tin-soldering to create a JTAG cable ;-) Jul 26 14:34:51 good afternoon pb_ Jul 26 14:36:04 woglinde: first i used the recipes, but i try my basic opengl program then it gives error. erron at http://pastebin.com/y5RnVdy1 Jul 26 14:37:17 akumar you seems to miss the lib where _ZN11QEglContextC1Ev is in Jul 26 14:39:09 woglinde: please suggest me how to handle, and which lib have _ZN11QEglContextC1Ev Jul 26 14:39:44 akumar try to use objdump in sysroots/armvbla/usr/lib/ Jul 26 14:40:17 woglinde: fine, i try. Jul 26 14:40:49 are you familiar with objdump? Jul 26 14:44:12 woglinde: no Jul 26 14:44:37 ~hail scp Jul 26 14:44:38 * ibot bows down to scp and chants, "I'M NOT WORTHY!!" Jul 26 14:58:34 pb_: i just bought a new car... Jul 26 14:58:36 * mickey|office broke now Jul 26 14:58:53 that's why you should steal them instead Jul 26 14:58:54 mickey new car? Jul 26 14:58:55 why? Jul 26 14:58:57 no pesky depreciation Jul 26 14:59:02 "new" = used, but still Jul 26 14:59:06 woglinde: why not ;) Jul 26 14:59:19 sabine is earning quite well these days and not doing anything with her money *cough* ;) Jul 26 14:59:32 hm okay Jul 26 14:59:44 current one is 8 years old and starting to emit symptoms Jul 26 14:59:51 okay Jul 26 15:00:27 http://www.flickr.com/photos/hrwandil/4830986974/ Jul 26 15:00:58 mickey|office: which bmw this time? Jul 26 15:01:01 hrw: good pic :) Jul 26 15:01:07 hrw: 330i Cabriolet Jul 26 15:01:38 mickey|office: I think that I will cut new avatar from it Jul 26 15:01:45 hehe Jul 26 15:02:10 btw., if things go well, it looks like i might get a n900 on loan for some months Jul 26 15:02:26 will have a look at porting FSO Jul 26 15:02:41 anyone know who added the n900 to OE? Jul 26 15:02:59 rkritiiti Jul 26 15:03:11 thanx Jul 26 15:03:12 was doing some stuff for gsoc 2009 Jul 26 15:03:21 but he nerver merged back Jul 26 15:03:46 ah, too bad. i think the SHR and FSO folks will work a bit on it then Jul 26 15:03:58 mickey|office: oh, very good Jul 26 15:04:18 mickey|office: oFono iirc works on it Jul 26 15:04:32 * pb_ is spending all available money on his house at the moment Jul 26 15:04:33 eek Jul 26 15:04:34 mickey|office: and meego has 2.6.35 kernel for n900 so you can get something more recent then 2.6.28 Jul 26 15:05:02 i created a native recipe, and it was building correctly (plus its non native package); i deleted by tmp folder to try a clean build of everything; though now that recipe isnt being built as a dependancy in the image run queue (therefore it cant build the non-native package); if i try to bitbake it seperately it seems to go into a loop (hands at) 'NOTE: Resolving any missing task queue dependencies'; ive checked for circular dependencies (an Jul 26 15:05:03 mickey|office: but, I do get to drive around in a digger, so I guess this just about qualifies as a new vehicle :-} Jul 26 15:05:15 pb_: hehehe, sounds like fun Jul 26 15:05:29 yeah, it's quite entertaining Jul 26 15:05:58 construction going fairly well, though we uncovered a huge pit in the ground on friday while we were digging the foundations Jul 26 15:06:05 pb_: i thought you're actually refurbishing it rather than _constructing_ ? Jul 26 15:06:40 mickey|office: half and half. we have demolished one part of the house and are rebuilding it. the other part was less bad to start with so we are refurbishing that area. Jul 26 15:07:32 pb_: right. and it's big enough so you can actually live in the still existing part? :) Jul 26 15:07:49 mickey|office: heh, no, we are currently living in a caravan in the garden. Jul 26 15:08:05 the still-existing part would actually be large enough, but it has no kitchen or bathroom Jul 26 15:08:25 pb_: ah right, i remember. is your little one enjoying the "camping"? Jul 26 15:08:35 yes, she seems to be having a great time Jul 26 15:21:31 yeah.. Jul 26 15:21:37 re jay7 Jul 26 15:21:43 days of deliveries Jul 26 15:22:35 * Jay7 have received school linux distros prev. week, efika mx today and seems ben nanonote tomorrow :) Jul 26 15:22:53 ah hehe Jul 26 15:23:05 jay7 you are a teacher? Jul 26 15:24:24 woglinde: no, I'm authorized parther of company who make that distro Jul 26 15:24:48 that is demo-pack for school's Jul 26 15:24:56 jay7 ah cool Jul 26 15:25:34 where should i begin debugging a recipe hanging on: "NOTE: Resolving any missing task queue dependencies" Jul 26 15:28:07 today i tried building a x86_64 machine and found many inspiration/patches on the mailinglist. is there any reason why there is no generic 64bit x86 machine in there? Jul 26 15:28:43 also when building gcc-cross/glibc the libc.so.6 from glibc is not staged, so gcc fails ... anyone experienced a familiar problem? Jul 26 15:29:59 well i guess specifying the recipe manually (using -b) gets aroudn the dependancy errors o.0 Jul 26 15:30:10 err, well i guess not errors, just stops it from locking up Jul 26 15:32:07 warflyr are you using oe-dev? Jul 26 15:32:16 warflyr and which bitbake with it? Jul 26 15:32:32 oe-dev, bout 2 weeks old, bb 1.11.0 Jul 26 15:33:22 err, gumstix/overo repo based on oe-dev Jul 26 15:49:16 mickey|office, hi, I was suggested to make an install guide? that goes to shr wiki? Jul 26 15:49:22 to openmoko-wiki? Jul 26 15:49:36 to fso-wiki? (unlikely because fso is distro-agnostic) Jul 26 15:49:43 I mean it can run on debian Jul 26 15:49:55 on htc-linux wiki(dev wiki=>unlikely) Jul 26 15:50:11 GNUtoo|laptop: SHR might make most sense Jul 26 15:50:21 ok Jul 26 15:50:29 I was hesitating between openmoko and SHR Jul 26 15:50:33 openmoko has the shr manual Jul 26 15:50:35 that's why Jul 26 15:54:21 hmm, i don't have a strong opinion on that, but i think stuff should rather move away from the Openmoko wiki into seperate projects. We should support the learning experience that Openmoko is a dead platform, but the subprojects are alive and kicking Jul 26 16:02:38 03Chris Larson  07org.openembedded.dev * rbbc3d48274 10openembedded.git/classes/sanity.bbclass: (log message trimmed) Jul 26 16:02:38 sanity.bbclass: Run our checking at BuildStarted, not ConfigParsed Jul 26 16:02:38 ConfigParsed is often utilized to do programmatic changes to the configuration Jul 26 16:02:38 metadata. If these other event handlers set any variables which Jul 26 16:02:38 sanity.bbclass needs to check, there's a problem, since we can't ensure those Jul 26 16:02:38 handlers run before this one. So, we move the sanity checking to when the Jul 26 16:02:39 build is about to start, this ensures that the configuration metadata is Jul 26 16:06:11 okay, I think I may finally have my siteinfo revamp ready to go in Jul 26 16:06:17 * kergoth_ does a quick sanity check Jul 26 16:06:17 very good Jul 26 16:06:35 * kergoth_ always has so many branches with things that never get quite polished up to be merged Jul 26 16:06:40 s/up/up enough/ Jul 26 16:15:23 think I may just push this one. cleanups that don't change behavior really don't require review in my opinion, as long as you ensure it didn't make anything explode Jul 26 16:15:26 heh Jul 26 17:07:30 03Frans Meulenbroeks  07org.openembedded.dev * r8d209c6d13 10openembedded.git/ (3 files in 2 dirs): Jul 26 17:07:30 cpan.bbclass: removed legacy staging Jul 26 17:07:30 bumped PR for the main perl recipes so this change takes effect Jul 26 17:07:30 Signed-off-by: Frans Meulenbroeks Jul 26 17:07:30 Acked-by: Koen Kooi Jul 26 17:07:31 Acked-by: Roman I Khimov Jul 26 17:28:46 btw - what stops OE from moving to bitbake 1.10? (except lack of release of such one) Jul 26 17:30:35 hrw, dunno, I recently moved (but I would still welcome a baselined 1.10 release; as it is more clear which version in case there are issues) Jul 26 17:31:07 kergoth_: can we release 1.10.0 finally? Jul 26 17:31:09 hrw, I don't know of anything, seems to work fine mostly. There does appear to be a bug with -k though, it doesn't seem to make it continue past certain errors, but that's just a bug that we'll get fixed Jul 26 17:31:41 1.8.18 is over half year old and looks like 1.10 is more and more used by people Jul 26 17:33:50 I branched it, expecting RP would release it, but he hasn't. maybe I should go ahead and do it, or zecke. we can always release new minor versions to fix issues, but at least itd be out there Jul 26 17:34:28 I know that I will not release 1.10.0 - no rights Jul 26 17:35:06 course, where to do the release -- code isn't at SF anymore, but I guess we could do the file release there Jul 26 17:35:15 otherwise the cgit interface lets you download archives of tags :) Jul 26 17:35:48 kergoth_: s/SF/Berlios/ Jul 26 17:36:00 meh Jul 26 17:36:09 berlios sucks Jul 26 17:36:12 kergoth_: make a tarball, tag git, release on berlios Jul 26 17:36:28 tempting to ditch it at some pointa nd just host the manual elsewhere Jul 26 17:36:30 kergoth_: but it is still home of BitBake as a project Jul 26 17:36:31 but for now yeah Jul 26 17:37:19 elo mickeyl Jul 26 17:37:31 heya Jul 26 17:41:56 * hrw -> off Jul 26 17:55:58 can someone explain me how insane.bbclass hooks into the build system ? Jul 26 17:56:07 what do you mean? Jul 26 17:56:24 how is it loaded/when is it invoked Jul 26 17:56:28 read the class. Jul 26 17:56:37 "addtask qa_staging after do_populate_staging before do_package_write" Jul 26 17:56:41 seems pretty obvious to me Jul 26 17:56:43 * kergoth_ shrugs Jul 26 17:57:49 kergoth_: ok, that explains part of my question, missed it Jul 26 17:58:16 what's the other part? Jul 26 17:58:22 but are the files in the classes dir always processed? or should I have found an inherit insane somewhere Jul 26 17:58:34 you missed an INHERIT += insane in the distro Jul 26 17:58:36 look harder Jul 26 17:58:50 kergoth_: :-) Jul 26 17:58:59 INHERIT variable contains a space separated list of classes which are always inherited, they go into the configuration metadata Jul 26 17:59:08 thanks alot, only looked for an invocation in the classes dir Jul 26 17:59:10 the only class that's implicitly loaded is base Jul 26 17:59:15 ofc Jul 26 17:59:35 :) Jul 26 18:00:11 03Evgeniy Dushistov  07org.openembedded.dev * rc47c6611be 10openembedded.git/lib/oe/ (patch.py path.py): Jul 26 18:00:11 oe.patch: don't error when the symlink already exists and is correct Jul 26 18:00:11 Adds oe.path.symlink convenience function. Jul 26 18:00:11 Signed-off-by: Evgeniy Dushistov Jul 26 18:00:11 Signed-off-by: Chris Larson Jul 26 18:00:15 03Chris Larson  07org.openembedded.dev * r87dc166bad 10openembedded.git/ (23 files in 16 dirs): (log message trimmed) Jul 26 18:00:15 Siteinfo cleanup Jul 26 18:00:15 - Don't supply site files for native Jul 26 18:00:15 - Split up the site information, so we don't need to add every single Jul 26 18:00:16 combination of os and architecture to the siteinfo.bbclass Jul 26 18:00:16 - Instead of specifically looking for '${FILE_DIRNAME}/site-${PV}/', Jul 26 18:00:16 leverage FILESPATH. Now it searches the paths in both BBPATH and FILESPATH Jul 26 18:01:10 actually what I am really interested in is seeing if insane can be split, I build non gfx systems and I would like to reduce build time when building from scratch by excluding the .desktop check (or put it in a separate class that is only executed whenever one is building a desktop file) Jul 26 18:01:35 actually it is not the check itself that i have problems with, it is the dpeendency on desktop-file-utils-native Jul 26 18:01:40 * kergoth_ doubts the impact of one little check is going to be statistically significant Jul 26 18:01:43 ah Jul 26 18:02:28 i noticed that when I do a bitbake virtual/kernel from scratch it goes to build glib etc etc; I don't mind about the cycles or the disk space but I do mind about the time Jul 26 18:03:19 check -g Jul 26 18:03:22 see what's pulling it in Jul 26 18:03:34 that's understandable Jul 26 18:03:44 splitting up insane seems like a decent idea, its pretty huge Jul 26 18:03:46 heh Jul 26 18:04:17 effeM: have you looked into adding appropriate -natives to ASSUME_PROVIDED, based on what you have on the machine? Jul 26 18:04:21 that can reduce build time a fair bit Jul 26 18:04:36 *g* Angstrom DOES NOT support ipkg because ipkg has been superseded by opkg Jul 26 18:05:02 * kergoth_ thinks angstrom's "we don't support this" needs to be moved somewhere other than parse time Jul 26 18:05:27 it could leverage StampUpdate to see if bitbake intends to actually *build* the things it doesn't support Jul 26 18:05:30 rather than just parsing it Jul 26 18:05:31 perhaps Jul 26 18:06:08 kergoth feeling better today? Jul 26 18:06:53 i can concentrate pretty well today, but now i feel like i'm starting to get sick. Jul 26 18:06:57 this year is annoying :) Jul 26 18:08:05 * woglinde is testing the dbus-glib stuff Jul 26 18:19:31 kergoth_: Moving theis angstrom warnings is a good idea. We had wuite some questions by users who got really confused by these. Jul 26 18:20:54 * florian can't type any more... time to go home Jul 26 18:20:59 bbiab Jul 26 18:21:02 ka6sox-work: hi Jul 26 18:21:08 hehe Jul 26 18:21:11 bye florian Jul 26 18:21:13 hi khem Jul 26 18:21:20 hello woglinde Jul 26 18:28:05 kergoth_: thanks for your answers, was called away for a while Jul 26 18:28:09 np Jul 26 18:28:48 ASSUME_PROVIDED could help as well, partial checkout also helps reducing the bb parsing time Jul 26 18:29:04 didn't think about ASSUME_PROVIDED before Jul 26 18:30:03 m4-native, etc Jul 26 18:30:22 btw glib is pulled in by insane.bbclass when building virtual/kernel (through desktop-file-utils) and actually it is glib-native Jul 26 18:30:24 just can't do it for 1) ones we patch for us, and 2) ones which are needed for specific versions, like python-native adn perl-native Jul 26 18:31:28 of course Jul 26 18:31:56 course, we don't flag those in any way, so you kind of need to figure it out on a case by case basis Jul 26 18:32:09 for 1) we should really add -oe to the PN, indicating its *ours* Jul 26 18:34:06 good plan Jul 26 18:35:45 actually at the moment I am building libpcap and tcpdump and for the last 30 minutes I've mostly see x related stuff pass by Jul 26 18:37:24 and there are native recipes that apparently are not used by any other recipe, for instance I just checked zziplib-native, but can't find why it exists, there seems to be no user Jul 26 18:44:10 hm Jul 26 18:44:12 | checking for growing stack pointer... configure: error: in `/devel/arm/oetmp-ang/work/armv7a-angstrom-linux-uclibceabi/glib-2.0-2.24.1-r0/glib-2.24.1': Jul 26 18:44:15 | configure: error: cannot run test program while cross compiling Jul 26 18:44:18 | See `config.log' for more details Jul 26 18:56:32 seems the wrong site-config is loaded but how comes this? Jul 26 18:57:25 have you checked siteinfo.bbclass ? Jul 26 18:57:55 khem ping Jul 26 18:57:59 its uclibc Jul 26 18:58:58 export CONFIG_SITE="/devel/arm/git/openembedded/site/endian-little" Jul 26 18:59:02 nothing more Jul 26 18:59:14 effem can you please look up what is on normal build? Jul 26 18:59:39 siteinfo.bbclass says: "arm-linux-uclibceabi": "endian-little bit-32 common-linux common-uclibc arm-common arm-linux-uclibc",\ Jul 26 19:00:15 armv7a-angstrom-linux-uclibceabi Jul 26 19:00:16 hm Jul 26 19:00:17 arm-linux-uclibc has the same minus the arm-linux-uclibc entry Jul 26 19:00:40 effem sure but I want to see what is the run_foo Jul 26 19:00:40 there is no armv7a entry Jul 26 19:01:19 seems we need armv* Jul 26 19:01:22 or soemthing Jul 26 19:02:03 target = bb.data.getVar('HOST_ARCH', d, 1) + "-" + bb.data.getVar('HOST_OS', d, 1) Jul 26 19:02:07 target is used as index Jul 26 19:02:31 ah found it Jul 26 19:02:32 targetinfo = { Jul 26 19:02:34 not sure how this part actually works Jul 26 19:02:46 look at this Jul 26 19:02:52 seems khem overlooked it Jul 26 19:03:34 hm damn now we cannt work with regexp Jul 26 19:03:38 because of armeb Jul 26 19:04:05 khem ping ping ping Jul 26 19:06:59 hm I have an idea to fix it Jul 26 19:08:33 hm regexp Jul 26 19:22:58 hmm need a little help here with cmake.bbclass Jul 26 19:23:24 i need to set the CMAKE_SYSTEM_PROCESSOR according to the target CPU Jul 26 19:23:39 cmake itself fails with that Jul 26 19:23:45 http://public.kitware.com/Bug/view.php?id=9065 Jul 26 19:24:11 (last comment on that bug) Jul 26 19:24:21 what would be the best way to set that var? Jul 26 19:25:37 something as it is done for CMAKE_SYSTEM_NAME Jul 26 19:44:04 woglinde_: pong*3 Jul 26 19:44:47 khem we need to alter siteinfo.bbclass Jul 26 19:45:19 khem a nearly working regexp is Jul 26 19:45:21 re.sub('\A(arm.*)(eb|\B\w+)-','arm\\2-', target Jul 26 19:46:20 oh for toolchain canonical name you mean Jul 26 19:46:41 yes Jul 26 19:46:57 woglinde_: wait Jul 26 19:47:04 otherwise the false site or incomplete siteconfig is loaded Jul 26 19:47:05 I am coming up with alternatives Jul 26 19:47:17 which cpu name is not going to be touched Jul 26 19:47:46 toolchain will be something like arm-v5te_angstrom-linux-gnueabi-* Jul 26 19:48:25 o.O? Jul 26 19:53:59 khem? Jul 26 19:58:22 that sounds a bit unwholesome. can't you just set a custom program-prefix in gcc and adjust ${CC} to match, rather than messing with the vendor string? Jul 26 20:00:31 *sigh* I am to dump with regexep Jul 26 20:01:01 why is re.sub('\A(arm)(\w*)(eb-|-)','\\1\\3', 'armv7aeb-moo') nor working Jul 26 20:01:40 for armv7a its working Jul 26 20:01:48 but why not for eb Jul 26 20:03:07 pb__: we need something like that we already have MULTIMACH* stuff with this you dont need Jul 26 20:03:12 MULTIMACH Jul 26 20:03:22 and its more consistent Jul 26 20:03:58 hm how do I match foo or nothing Jul 26 20:04:04 with regexp Jul 26 20:04:25 so I think what I have done is leavaraged MULTIMACH stuff to work for toolchain too Jul 26 20:04:35 I think it looks better now Jul 26 20:05:11 just sent the revised patches Jul 26 20:09:13 woglinde_: why should the above work? Jul 26 20:09:44 * zecke hates cellphones Jul 26 20:09:51 zecke enlight me Jul 26 20:09:59 zecke: heh they are not for IRCing :) Jul 26 20:10:26 how do I match foo or nothing with regexp? Jul 26 20:10:34 if a legacy staging recipe has a do_stage calling just autotools_stage_all - the fix is to just remove it? Jul 26 20:10:36 woglinde_: pickup the latest 7 patches I posted you dont need that then hopefully Jul 26 20:10:45 mrmoku: yes Jul 26 20:10:48 khem I dont want to rebuild now Jul 26 20:10:51 khem: thanks :) Jul 26 20:10:58 woglinde_: grrr Jul 26 20:11:11 woglinde_: what exactly do you want from the armstring? Jul 26 20:11:30 woglinde_: what is \w supposed to mean in your regexp language? Jul 26 20:11:43 its python regexp Jul 26 20:11:53 http://docs.python.org/library/re.html#search-vs-match Jul 26 20:11:56 woglinde_: okay so any charachter. :) Jul 26 20:12:04 yes Jul 26 20:12:49 zecke the crux os to delete all btw arm and the - Jul 26 20:13:05 or delete all btw arm and eb- Jul 26 20:13:24 woglinde_: aeh, are you sure it is not work? Jul 26 20:14:29 yes Jul 26 20:14:46 >>> re.sub('\A(arm)(\w*)(eb-|-)','\\1\\3', 'armv7aeb-moo') Jul 26 20:14:46 'arm-moo' Jul 26 20:14:48 this is wrong? Jul 26 20:15:26 yes Jul 26 20:15:31 it needs to be armeb Jul 26 20:18:33 woglinde_: ofc you can also go for the easy dirty way, just add armv7a and armv7aeb or so to siteinfo Jul 26 20:18:44 armeb-moo is the right answer? Jul 26 20:18:55 effem bah Jul 26 20:19:11 effem there is armv6 armv5t Jul 26 20:19:14 and so on Jul 26 20:19:48 zecke for armv7a-moo it should be arm-moo and for armv7aeb-moo armeb-moo Jul 26 20:20:45 woglinde_: hmm Jul 26 20:20:59 woglinde_: I suck at regex too, I could draw a DFA. :) Jul 26 20:22:20 woglinde_: we need to put the or somehow else... (\w*eb-) is fine. Jul 26 20:22:49 (\web-) would be wrong Jul 26 20:23:09 ups (\w*eb) groups the whole match before eb Jul 26 20:24:15 is problem with failing compilation fo prismstumbler (while bitbaking e-image) known to anyone? Jul 26 20:24:54 i am using dev branch and was few commits behind. i'll try to pull and check if it is fixed Jul 26 20:25:20 woglinde_: I would use my lovel split('-', 1) Jul 26 20:26:59 zecke split gains you nothing here Jul 26 20:27:25 woglinde_: I know, that's why I called it the easy dirty way Jul 26 20:28:08 actually what also could be done is having a func or table that maps from specific to generic Jul 26 20:28:53 crap. after git pull I am stuck with changed ABI (4 to 5). is there a way to rebuild what's needed, or do i have to remove my tmp dir? Jul 26 20:29:48 * mrmoku has the same problem on shr buildhost :P Jul 26 20:30:49 03Thomas Zimmermann  07org.openembedded.dev * rf392e7f75c 10openembedded.git/recipes/aceofpenguins/ (3 files in 2 dirs): Jul 26 20:30:49 aceofpenguins: update to version 1.3 Jul 26 20:30:49 * new game: spider Jul 26 20:30:49 * some fixes Jul 26 20:30:49 * renamed files folder to aceofpenguins Jul 26 20:30:49 Signed-off-by: Thomas Zimmermann Jul 26 20:30:52 03Thomas Zimmermann  07org.openembedded.dev * r6303c5602c 10openembedded.git/recipes/aceofpenguins/aceofpenguins-launcher_0.4.bb: Jul 26 20:30:52 aceofpenguins-launcher: update to 0.5 to add button for spider game Jul 26 20:30:52 Signed-off-by: Thomas Zimmermann Jul 26 20:31:49 effem I dont like the easy dirty way Jul 26 20:35:29 woglinde_: i understand that Jul 26 20:35:59 nazgee, either remove tmp or revert to before the ABI bump Jul 26 20:36:46 actually i think a less drastic rm is possible, guess there is no need to remove most of the native recipes (except perhaps the ones that are affected by the change requiring the abi change) Jul 26 20:38:11 btw my bitbake libpcap tcpdump is now packaging the gst-plugins-base; yuk Jul 26 20:39:05 my build/tmp/sysroots/ppc440-angstrom-linux directory is not complete enough to use as an NFS root dir -- it's missing /sbin for example. is there a bitbake command I'm supposed to use to install everything in there? Jul 26 20:39:16 nothing in tmp is *ever* right for nfs Jul 26 20:39:17 ever. Jul 26 20:39:26 unless you're building as root, which you shouldn't be doing Jul 26 20:39:30 kergoth_: ok, how do I get something that's right for NFS? Jul 26 20:39:35 what you want to do is IMAGE_FSTYPES += "tar.gz" Jul 26 20:39:37 bake an image Jul 26 20:39:41 and unpack that as root somewhere Jul 26 20:41:27 kergoth_: my cross-compiler uses that sysroots directory by default. if I unpack elsewhere, I'll need to build my (non-reciped) projects with extra cflags/ldflags/destdir options, which I'd like to avoid Jul 26 20:41:45 hmm? Jul 26 20:42:09 if you want a toolchain, bitbake meta-toolchain Jul 26 20:42:11 toolchain != rootfs Jul 26 20:42:28 I have a toolchain; oe made it for me Jul 26 20:42:35 cpp -v says this: Jul 26 20:42:38 yes, and a toolchain is not a filesystem. Jul 26 20:42:41 /home/hollisb/work/openembedded/build/tmp/cross/ppc440/lib/gcc/powerpc-angstrom-linux/4.3.3/include Jul 26 20:42:41 /home/hollisb/work/openembedded/build/tmp/cross/ppc440/lib/gcc/powerpc-angstrom-linux/4.3.3/include-fixed Jul 26 20:42:41 /home/hollisb/work/openembedded/build/tmp/sysroots/ppc440-angstrom-linux/usr/include Jul 26 20:42:55 for #include search directories Jul 26 20:43:04 if you want a toolchain you can use outside of oe without hassles, use meta-toolchain. Jul 26 20:43:08 it hands you a tarball Jul 26 20:43:17 with an environment setup script Jul 26 20:43:19 so my gcc will automatically find anything I put in that sysroot Jul 26 20:43:43 I don't want another toolchain; I am happy with the oe-built one Jul 26 20:43:48 again, a root filesystem is completely independent of the toolchain. if you want an nfs root, bake an image and unpack that somewhere and host it Jul 26 20:44:30 they're not *completely* independent, since I just showed you the search paths built in to my toolchain... Jul 26 20:45:24 what about it? Jul 26 20:45:44 you aren't running the crosscompiler from the target. Jul 26 20:45:50 so i fail to see what those paths have to do with a rootfs Jul 26 20:46:36 kergoth_: if, on the host, I manually install files into the cross-compiler's sysroot directory, will those files be included in the image.tar.gz file? Jul 26 20:46:48 if you bake a library with oe, say, as a part of building an image, the libs and headers for that end up in staging, and available to the crosscompiler. so you have it boht in the image and in the toolchain Jul 26 20:46:50 eFfeM: i've tried deleting, to let my PC payback some money I invested in it ;] Jul 26 20:46:50 this isn't that hard Jul 26 20:46:56 you shouldn't be manually installing files there Jul 26 20:46:59 hollisb: no Jul 26 20:47:02 if you do, you're on your own Jul 26 20:47:23 nazgee, deleting is the simplest, just takes some time to rebuild things Jul 26 20:47:38 khem: right, ok Jul 26 20:48:15 so basically there's no way for me to make use of my cross-compiler's sysroot (unless I recipe-ize my projects, I guess) Jul 26 20:48:23 files should belong to some package in order to go into rfs Jul 26 20:49:26 your crosscompiler's sysroot isn't a filesystem. it doesn't have /etc, it doesn't magically run the postinsts of the packages to ensure everything works, and it certainly doesn't have the right permissions, unless you're building as root, which again you should never be doing Jul 26 20:49:43 kergoth_: I understand it's not a filesystem Jul 26 20:49:53 kergoth_: I really do know what I'm talking about; I just don't know OE :) Jul 26 20:49:54 then why are you wanting to make it available as an nfs root? Jul 26 20:50:45 * kergoth_ gives up and wanders off Jul 26 20:50:49 kergoth_: really? Jul 26 21:23:36 damn taxes Jul 26 21:24:51 yep :( Jul 26 21:25:40 hm lets see what kergoth changed in siteinfo Jul 26 21:28:54 jineld hm I made my irs today Jul 26 21:29:33 I had to do it about a month ago or so Jul 26 21:29:37 I really hate it Jul 26 21:30:06 especially when I see how much is taken away and how they use the money Jul 26 21:30:10 :P Jul 26 21:30:31 hehe.. Jul 26 21:30:33 ugh so it was kergoth changes Jul 26 21:30:39 which broke the stuff Jul 26 21:30:41 hms Jul 26 21:31:12 what broke? i did a number of builds to test it Jul 26 21:31:31 armv7a Jul 26 21:31:40 instead of arm Jul 26 21:32:05 uh, that errored in siteinfo even before my change :P Jul 26 21:32:13 at least for me, and some others on the mailing list thread Jul 26 21:32:14 hm okay Jul 26 21:32:32 okay so still I need the regexp Jul 26 21:32:39 which cuts out something btw. Jul 26 21:32:48 something Jul 26 21:41:33 woglinde_: can you explain again about regexp? Jul 26 21:41:39 kergoth_: if I have _${PN} = "something" in override when will it evaluate Jul 26 21:42:23 woglinde_: did you need to replace armv7eb-moo to armeb-moo? Jul 26 21:42:27 I am trying to use anon python in bitbake.conf and when the var is using _${PN} override it does not get the overridden value Jul 26 21:42:41 jay7 either you get armbla-foo or armblaeb-foo Jul 26 21:42:52 khem, keys are expanded at the end of the parse, same as overrides Jul 26 21:42:53 in both cases you want to cut out the bla Jul 26 21:44:04 kergoth_: hmmm MYVAR_qemuppc = "a" and MYVAR_${PN} = "b" in a recipe Jul 26 21:44:12 I get first one right Jul 26 21:44:13 kergith bb.note or bb.info? Jul 26 21:44:45 kergoth_: but second one seems is not evaluated when my little anon python is executed Jul 26 21:46:26 hm okay Jul 26 21:46:31 it works Jul 26 21:46:45 okay that looks good Jul 26 21:46:47 siteinfo target string is arm-linux-uclibceabi Jul 26 21:46:51 kergoth_: OVERRIDES=local:${MACHINE}:${DISTRO}:${TARGET_OS}:${TARGET_ARCH}:build-${BUILD_OS}:fail-fast:pn-${PN}:${FEED_ARCH}:libc-uclibc Jul 26 21:47:54 kergoth_: so it seems it should be _pn-${PN} ? Jul 26 21:49:07 but why the hell the export of the SITEINFO FILE is wrong Jul 26 21:49:26 kergoth_: yep seems to be the case Jul 26 21:50:19 kergoth_: there are overrides like PACKAGE_ARCH_${PN} = "${MACHINE_ARCH}" I think they are wrong unless we have ${PN} in default overrides Jul 26 21:55:31 hm files are ['endian-little', 'bit-32', 'arm-common', 'common-linux', 'common-uclibc', 'arm-linux-uclibc', 'arm-linux-uclibceabi', 'common'] looks good too Jul 26 21:56:14 export CONFIG_SITE="/devel/arm/git/openembedded/site/endian-little" Jul 26 21:56:15 not Jul 26 21:57:50 khem, that's not an override. Jul 26 21:57:54 khem, at least, not a normal one. and it works fine. Jul 26 21:58:10 khem, in package.bbclass, the pkg in PACKAGES is added to it Jul 26 21:58:18 which is how RDEPENDS_ works Jul 26 21:58:22 kergoth_: which one Jul 26 21:58:25 ? Jul 26 21:58:26 ah Jul 26 21:58:29 got it Jul 26 21:58:34 PAcKAGE_ARCH_${PN} works fine because pkg is in overrides at package time Jul 26 21:58:38 and that var is only used at package time Jul 26 21:59:15 kergoth_: hmmm actually its used to determine the dir where stuff should go Jul 26 21:59:25 what do you mean? Jul 26 21:59:35 kergoth_: but there is another smear on top so it gets hidden Jul 26 21:59:53 hmm Jul 26 21:59:59 evening Jul 26 22:00:03 hi zub Jul 26 22:00:40 re Jul 26 22:00:46 re florian Jul 26 22:01:05 kergoth_: I am looking at PACKAGE_ARCH_${PN} = "${MACHINE_ARCH}" Jul 26 22:01:08 khem, yeah, I think that should probably just be PACKAGE_ARCH, for now, given we need it for multimachine Jul 26 22:01:20 deal with supporting per binary package package_arch = machine_arch later Jul 26 22:01:22 * kergoth_ shrugs Jul 26 22:01:22 got a dilemma... I updated (locally so far) webkit-efl recipe. But it seems there's nothing to be put in PV, apart from svn rev Jul 26 22:01:37 yet older recipe used 1.1.11, I don't know where did it come from Jul 26 22:01:42 kergoth_: orchange it Jul 26 22:01:47 to PACKAGE_ARCH_${PN} = "${MACHINE_ARCH}" Jul 26 22:01:50 errr Jul 26 22:01:59 PACKAGE_ARCH_pn-${PN} = "${MACHINE_ARCH}" Jul 26 22:02:31 webkit-efl pkg config file claims version 0.1, but if I used that, my new recipe would have PV < old recipe PV Jul 26 22:02:35 that's wrong I assume Jul 26 22:03:35 kergoth sure you want to search in all thes paths for sitefiles? Jul 26 22:03:37 paths = itertools.chain(d.getVar("BBPATH", True).split(":"), d.getVar("FILESPATH", True).split(":")) Jul 26 22:03:38 khem, that's pointless. Jul 26 22:03:52 woglinde_, if I wasn't sure, it wouldn't be there. Jul 26 22:03:59 in fact, its *in the commit message* Jul 26 22:04:12 kergoth that dont works as expteced Jul 26 22:04:15 khem, the whole point of the _${PN} override was to apply to only to the main binary package Jul 26 22:04:19 woglinde_, ? Jul 26 22:04:31 khem, if your'e going to do it recipe wide, you shouldn't need the override at all Jul 26 22:05:13 http://pastebin.com/mD7BAq3T Jul 26 22:05:35 no wonder bitbake is so slow Jul 26 22:06:20 ? Jul 26 22:06:58 well, you're welcome to revert it for now Jul 26 22:07:00 its checking all the dirs if the files exist Jul 26 22:07:05 no shit. Jul 26 22:07:44 just revert it for now, I don't have time to deal with it Jul 26 22:08:27 I have a way to fix it that'll avoid the performance issue Jul 26 22:09:07 hm Jul 26 22:09:21 it breaks here with endian-little Jul 26 22:09:33 and bit-32 Jul 26 22:09:41 03Chris Larson  07org.openembedded.dev * r155f1ff7f2 10openembedded.git/ (23 files in 16 dirs): Jul 26 22:09:41 Revert "Siteinfo cleanup" Jul 26 22:09:41 Will revisit. Jul 26 22:09:41 This reverts commit 87dc166bade8722af4076c8644cac6a0a437826f. Jul 26 22:09:51 well, if you can tell me how to see the breakage, I'll look into it Jul 26 22:09:57 as i said, i did a number of builds with no problem Jul 26 22:10:16 use uclibc Jul 26 22:10:19 and glib-2.0 Jul 26 22:11:51 what machine/distro? Jul 26 22:12:02 every other combination i try recently is broken in one way or another Jul 26 22:12:02 angstroem Jul 26 22:12:10 was a lot of work to get a successful build to manage to actually test Jul 26 22:12:29 I am trying too to understand whats going wrong Jul 26 22:13:38 thats the outcome from siteinfo_get_filenames ['endian-little', 'bit-32', 'arm-common', 'common-linux', Jul 26 22:14:06 for endian-little for p in paths is done Jul 26 22:14:24 for all other files after endian-little not Jul 26 22:20:24 kergoth maybee the problem is that paths is a Jul 26 22:20:33 why would that be a problem? Jul 26 22:20:35 instead of string Jul 26 22:20:38 that's what it should be Jul 26 22:20:50 note: for p in paths: Jul 26 22:20:55 you wouldn't want paths to be a string Jul 26 22:20:57 you want it to be an iterable Jul 26 22:21:07 (well, stirngs are iterable, but it hands you one character a time, which is not what you want) Jul 26 22:22:06 args right Jul 26 22:22:10 list of strings Jul 26 22:22:40 for the next file it isnt iterable Jul 26 22:22:44 anymore Jul 26 22:25:23 oh, I see. yep, you're right, its because it hits the end of the iterable, need to wrap it in list() or so Jul 26 22:25:29 regardless, I'm revamping it as we speak Jul 26 22:25:50 yes Jul 26 22:25:59 * woglinde_ read about it Jul 26 22:26:04 in itertools Jul 26 22:26:11 aside: why the hell is lmbench using CONFIG_SITE when it apparently isn't autotools? Jul 26 22:26:32 okay Jul 26 22:26:36 good we find it Jul 26 22:26:40 good nite Jul 26 22:26:45 night Jul 26 22:26:57 I think you will have fixed it tomorrow Jul 26 22:26:58 *g* Jul 26 22:27:40 got one lama guestion - if there's a series of changes that depend on each other - that's what [PATCH m/n] is for? i.e. sending whole bunch? Jul 26 22:28:03 zub git-send-email Jul 26 22:28:10 git format-patch Jul 26 22:28:26 yes, I noticed that git will by default do this, I'm just asking if this implies it's a sequence Jul 26 22:28:34 with dependencies Jul 26 22:28:51 o.O? Jul 26 22:28:59 its just how you checked it in Jul 26 22:29:10 nah, maybe asking about the obvious and the answer is yes :) Jul 26 22:29:26 each commit Jul 26 22:29:35 in that order Jul 26 22:29:42 okay Jul 26 22:44:51 zub, it's a sequence yes. that doesn't necessarily imply dependence, however Jul 26 22:44:58 if they all touch different files.. ;) Jul 26 22:45:33 seems you understoog what my question was :) Jul 26 22:46:11 so, if there is a dependence, I guess I can mention it in the commit message (?) Jul 26 22:46:20 or write a cover mail Jul 26 22:47:41 often people will summarize the forthcoming patchset in a [0/n] mail Jul 26 22:47:46 I think that's the best way to go Jul 26 22:47:51 ok, thanks Jul 26 22:47:53 np Jul 26 22:57:08 hey khem? Jul 26 23:16:08 Tartarus: hi Jul 26 23:18:21 incoming :) Jul 27 00:35:38 03Chris Larson  07org.openembedded.dev * r45fc36a32f 10openembedded.git/ (24 files in 16 dirs): (log message trimmed) Jul 27 00:35:38 Siteinfo cleanup v2 Jul 27 00:35:38 - Don't supply site files for native. Jul 27 00:35:38 - Split up the site information, so we don't need to add every single Jul 27 00:35:38 combination of os and architecture to the siteinfo.bbclass. Jul 27 00:35:38 - Drop the ${FILE_DIRNAME}/site- stuff, there are other ways to do it. Jul 27 00:35:40 - SITEINFO_ENDIANESS -> SITEINFO_ENDIANNESS. Jul 27 00:48:45 hi, all Jul 27 00:50:25 how much disk space do I need to create a rootfs? I have 4G available on hard drive.... Jul 27 00:51:03 that depends on the rootfs, but I doubt that will be enough. You'll definitely want to INHERIT += "rm_work", and perhaps BB_SCHEDULER = "completion" as well Jul 27 00:56:51 kergoth_: sorry "df -h" reports 39G as available memory :) Jul 27 01:07:13 I just didn't count the digits correctly Jul 27 02:35:16 hmm, should rename siteinfo. It's not named correctly. The 'site' is generally the build machine, the common bits between builds. CONFIG_SITE is a list of bits that are common to multiple ./configure's. siteinfo.bbclass doesn't control site information, it holds information about the *target* Jul 27 02:35:26 then autotools.bbclass can take that and use it to produce a list of site files Jul 27 02:37:49 kergoth_: I have 40G of free space. How much do I need to generate rootfs? I am trying to build x11-gpe-image..... Jul 27 02:37:55 tia! Jul 27 02:38:17 IgorK: you need around 30G with minimal distro Jul 27 02:38:51 kergoth: target-siteinfo ? Jul 27 02:40:41 "site" should be removed entirely, it doesn't have anything to do with it Jul 27 02:40:44 imo anyway Jul 27 02:40:54 i'm thinking targetinfo or targetdata or platformdata or platforminfo or something Jul 27 02:40:55 * kergoth shrugs Jul 27 02:41:08 target's kind of vague, platform might be clearer Jul 27 02:41:09 dunno Jul 27 02:43:30 khem: which means I have a lot of space ;-) Jul 27 02:45:55 http://paste.pocoo.org/show/242070 and http://paste.pocoo.org/show/242071 Jul 27 02:46:02 any idea? Jul 27 02:46:40 IgorK: it isn't running out of space on your desktop Jul 27 02:46:47 its running out of space inside the generated ext2 filesystem **** ENDING LOGGING AT Tue Jul 27 02:59:56 2010