**** BEGIN LOGGING AT Thu Feb 03 02:59:57 2011 Feb 03 08:17:05 good morning Feb 03 08:25:13 moin **** ENDING LOGGING AT Thu Feb 03 08:26:13 2011 **** BEGIN LOGGING AT Thu Feb 03 08:30:17 2011 Feb 03 08:41:07 03Stefan Schmidt  07master * rf5a345b1cf 10openembedded.git/recipes/wpa-supplicant/ (wpa-supplicant-0.7.inc wpa-supplicant_0.7.3.bb): Feb 03 08:41:08 wpa-supplicant_0.7: Package dbus system-service activation files an bump PR. Feb 03 08:41:08 Without these files wpa-supplicant will not lauchn on-demand when the bus Feb 03 08:41:08 name get called. Feb 03 08:41:20 03Stefan Schmidt  07master * r4a5701f2af 10openembedded.git/conf/distro/include/angstrom-2008-preferred-versions.inc: Feb 03 08:41:21 angstrom-2008-preferred-versions: Bump wpa-supplicant version to 0.7.3 Feb 03 08:41:21 This allows the use of the new DBUS API which is needed by connman 0.68 and Feb 03 08:41:21 higher. Feb 03 08:41:21 Signed-off-by: Stefan Schmidt Feb 03 08:41:21 Acked-by: Koen Kooi Feb 03 09:09:40 Tartarus: ping, weird stuff ends in my perl config.sh http://paste.pocoo.org/show/331469/ notice that /OE/tmpdir-shr/sysroots/x86_64-linux/usr/bin is later replaced with .../..pdir-shr/sysroots/x86_64-linux/usr/bin in few variables (/OE/tmpdir-shr is my TMPDIR in local.conf), noticed because libxml-parser-perl-native-2.36-r4 now fails Feb 03 09:17:15 mickey|zzZZzz: Feb 03 09:17:17 mickey|zzZZzz: ping Feb 03 09:19:00 Tartarus: before 49362da57224282f2dff462afa2e7ba9519ad7b5 it looks ok Feb 03 09:22:52 stefan_schmidt, ping? Feb 03 09:24:02 ka6sox: pong Feb 03 09:25:11 stefan_schmidt, another patch day? Feb 03 09:25:30 I have some maintainence but want to make sure its up if there will be one soon. Feb 03 09:25:58 mornin Feb 03 09:26:02 ka6sox: well, I'm a bit overloaded with other tasks right now Feb 03 09:26:15 ka6sox: But I should try to do another one... Feb 03 09:26:35 ka6sox: at least not this weekend if that helps you Feb 03 09:26:52 okay just put it on the list when you are ready I'll raise a flag if thats a problem. Feb 03 09:26:53 thanksl Feb 03 09:27:04 not this weekend means I can fix things. Feb 03 09:30:10 Tartarus: looks like only -Duserelocatableinc is the cause.. those sed calls in Configure script are replacing more then they should Feb 03 09:39:18 ping denix Feb 03 09:42:15 ka6sox: ok, thanks Feb 03 09:42:20 hi woglinde Feb 03 09:42:26 he stefan Feb 03 09:51:01 03Stefan Schmidt  07master * r0a509aeb50 10openembedded.git/recipes/ofono/ (5 files in 2 dirs): Feb 03 09:51:01 ofono: Add recipes from poky. Feb 03 09:51:01 Also add the newest, 0.40, release. Feb 03 09:51:06 some one knows how to build a hg version of a package? Feb 03 09:51:18 * stefan_schmidt has no idea Feb 03 09:52:20 ah hehe bitbake foo-hg Feb 03 09:52:43 :) Feb 03 09:53:12 * woglinde tryies hg version of cacao again Feb 03 09:53:34 stefan_schmidt: yes libnl[12] is madness, that's why I've blacklisted libnl1 + recipes depending on it for SHR Feb 03 09:53:59 libnl 1 1.2 was madness Feb 03 09:54:22 stefan_schmidt: having both libnl built in same sysroot breaks some apps (only runtime) Feb 03 09:54:43 JaMa|Wrk: urgs, even runtime? Feb 03 09:54:50 stefan_schmidt: and ie older bluez was libnl1 only Feb 03 09:55:08 stefan_schmidt: bad thing is that only in runtime (built ok) Feb 03 09:55:13 JaMa|Wrk: even worse then I thought Feb 03 09:55:29 JaMa|Wrk: yeah, build is at least visible Feb 03 09:55:48 JaMa|Wrk: besides oder bluez all recipes are ok for SHR? Feb 03 09:56:00 JaMa|Wrk: I wonder if angstrom should do the same then Feb 03 09:57:28 morning stefan_schmidt Feb 03 09:57:33 problem is that -c clean libnl1 is not enough to clean sysroot completely Feb 03 09:57:40 stefan_schmidt: can we talk for a bit on weekend? Feb 03 09:57:42 hi mickeyl Feb 03 09:57:45 moin woglinde Feb 03 09:57:54 mickey|office: sure Feb 03 09:57:56 stefan_schmidt: I had to remove all traces of libnl1 before proper rebuild of apps against libnl2 Feb 03 09:58:07 stefan_schmidt: ok, will call you on Sunday, if that's ok for ya Feb 03 09:58:12 JaMa|Wrk: urgs Feb 03 09:58:18 mickey|office: let me check Feb 03 09:58:37 mickey|office: sunday is empty right now. :) Feb 03 09:58:47 stefan_schmidt: iirc shlibs etc.. Feb 03 09:59:00 hi jama btw. Feb 03 09:59:06 hi woglinde Feb 03 09:59:32 JaMa|Wrk: hmm, I wonder if we coull avoid libnl1 completely in OE Feb 03 09:59:44 mickey|office: no FOSDEM for you? Feb 03 10:00:15 stefan_schmidt: can't make it this year. playing in a musical on saturday, friday is final rehearsal, and only for sunday it's way too expensive and stressful for me Feb 03 10:00:30 mickey|office: makes sense Feb 03 10:00:40 * stefan_schmidt will not be there as well Feb 03 10:00:46 stefan_schmidt: see http://gitorious.org/~jama/angstrom/jama-shr-experimental/commit/70b13d30b335b8fd67116d3bc0834382a77bc319 for list of all libnl1 users Feb 03 10:00:49 mickey|office: getting uni stuff done Feb 03 10:00:57 JaMa|Wrk: thx Feb 03 10:01:05 stefan_schmidt: haven't tried which ones can be also built and runs ok with libnl2 Feb 03 10:01:07 mickey|office: cebit? Feb 03 10:01:20 mickey|office: I might come around for a day or two Feb 03 10:01:21 stefan_schmidt: yes, cebit i want to make, not all the time, but at least two days Feb 03 10:01:35 nice, lets try to sync that then Feb 03 10:01:37 mickey|office: ok, then we should try to take the same days :) Feb 03 10:01:40 yep Feb 03 10:01:51 morphis wants to come as well Feb 03 10:02:22 JaMa|Wrk: hostap does work with libnl2 with the right config option. Same codebase as wpa_supplicant Feb 03 10:02:27 mickey|office: cool Feb 03 10:02:38 mickey|office: raster will be there as well. At the E booth Feb 03 10:02:48 ya, heard about that. almost a family meeting then :) Feb 03 10:03:10 * mickey|office still remembers the cool times over in Taipeh Feb 03 10:03:19 JaMa|Wrk: for the other I would need to check. some smaller thinks like iw or lowpan-tools should not be thjat difficult Feb 03 10:03:25 mickey|office: :) Feb 03 10:06:23 JaMa|Wrk: Could you reply to the mail and bring in some of your experience? Feb 03 10:07:59 yup doing it atm :) Feb 03 10:08:24 JaMa|Wrk: great Feb 03 10:10:14 I will be at cebit from thur-sat Feb 03 10:10:31 woglinde: cool Feb 03 10:11:05 mostly on the navit booth Feb 03 10:11:27 woglinde: oh, did not know you are active at the navit project Feb 03 10:12:03 yeah small things Feb 03 10:14:02 hi florian Feb 03 10:16:04 args Feb 03 10:16:06 please Feb 03 10:16:15 no errors now on my harddisk Feb 03 10:17:37 good morning Feb 03 10:29:05 morning florian Feb 03 10:29:54 hi ka6sox Feb 03 10:30:15 morning woglinde Feb 03 10:31:59 hm in the phoronix tests ext4 hat better performace than xfs in all cases Feb 03 10:37:36 ah nice the icedtea webplugin can now be build seperatly Feb 03 10:44:57 03Stefan Schmidt  07master * r18a238b620 10openembedded.git/recipes/tasks/task-bug.bb: task-bug.bb: Add bug core tasks used in our images Feb 03 10:45:07 03Stefan Schmidt  07master * ra32e9b2026 10openembedded.git/recipes/buglabs-apps/ (183 files): Feb 03 10:45:07 buglabs-apps: BugLabs java applications. Feb 03 10:45:07 Tons of small apps that live on buglabs.net and are mostly arch independent. These Feb 03 10:45:07 recipes make them available for images and feeds. Feb 03 10:45:08 03Stefan Schmidt  07master * r951c432cf1 10openembedded.git/recipes/images/bug-ramdisk.bb: bug-ramdisk: Add ramdisk image for QA testing Feb 03 10:45:09 03Stefan Schmidt  07master * r2ffe5731b1 10openembedded.git/recipes/tasks/task-poky.bb: task-poky.bb: Add poky and sato tasks we are using in our images Feb 03 10:45:10 03Stefan Schmidt  07master * re7246bd151 10openembedded.git/recipes/tasks/task-bug-feed-extra.bb: task-bug-feed-extra.bb: Add tasks for our feed Feb 03 10:45:12 03Stefan Schmidt  07master * rbe758d6589 10openembedded.git/recipes/tasks/task-bug-apps.bb: task-bug-apps.bb: Add tasks for BugLabs applications Feb 03 10:45:14 03Stefan Schmidt  07master * rb0170b91b7 10openembedded.git/recipes/images/bug-image-production.bb: bug-image-production.bb: Add BugLabs production image shipped on our devices. Feb 03 11:11:49 wtf Feb 03 11:12:14 anfrage von messehostessen.info Feb 03 11:12:17 ups Feb 03 11:12:21 email from Feb 03 11:12:44 woglinde: don't tell your wife... Feb 03 11:12:49 :)) Feb 03 11:25:13 what I need is an inline translator for IRC. Feb 03 11:27:25 lol Feb 03 11:30:03 ka6sox: /exec sometranslationscript "your text" :) Feb 03 11:30:14 irssi should know that Feb 03 11:30:26 maybe znc does Feb 03 11:31:10 znc may be scripted with perl Feb 03 11:31:16 I've tried once even :) Feb 03 11:31:43 Jay7, today @ 00:00 there were a LOT of inserts from oe_stats but things didn't fall down and go boom. Feb 03 11:32:02 1/3 the load of what we have been seeing. Feb 03 11:32:12 ka6sox: cool! Feb 03 11:32:53 so once you get done with the new code we can put up a test vhost and see that it works. Feb 03 11:33:48 no rush...just reporting. Feb 03 11:39:12 genau Feb 03 11:39:14 oops wrong window Feb 03 11:39:40 stimmt :) Feb 03 11:42:27 meistens Feb 03 11:43:15 true! Feb 03 11:43:29 I agree? Feb 03 11:44:07 mostly Feb 03 11:44:17 okay...will find something. Feb 03 11:58:59 ka6sox: I need more hours per day :) Feb 03 12:00:23 Jay7, so do I...been at this for almost 20hrs Feb 03 12:01:17 * Jay7 remembers old'n'good times when he was unmarried :) Feb 03 12:23:34 boing Feb 03 12:23:46 so who is going to FOSDEM then? Feb 03 12:37:27 rschus Feb 03 12:37:30 koen Feb 03 12:37:35 effem Feb 03 12:37:39 florian Feb 03 12:39:58 * dm8tbr is going Feb 03 12:40:14 and crofton Feb 03 12:44:08 * like|away will probably visit one day, no hotel Feb 03 12:44:24 hi likewise Feb 03 12:44:26 hi Feb 03 12:54:56 hey florian Feb 03 12:55:21 hi XorA Feb 03 12:56:15 florian: cya sat morning like normal I guess :-) Feb 03 12:57:07 * XorA is flying to FOSDEM tonight Feb 03 12:58:01 xora hm Feb 03 12:58:04 ah right Feb 03 12:58:18 woglinde: ?? Feb 03 13:02:56 xora I thought it would be to early Feb 03 13:03:03 but tomorrow its friday Feb 03 13:03:06 hi chase Feb 03 13:03:21 and friday == beer :-) Feb 03 13:03:37 was too overcorwded for me last year Feb 03 13:03:40 and to smoky Feb 03 13:04:01 chase I have one question about the last patch denix checked in Feb 03 13:04:04 they said it was private this year so we can hope Feb 03 13:04:44 you scared him :) Feb 03 13:34:13 * RP__ will be at FOSDEM Feb 03 13:34:35 hi rp Feb 03 13:34:37 ah right Feb 03 13:44:15 XorA: right Feb 03 13:59:53 03Stefan Schmidt  07master * r1206aa00bf 10openembedded.git/recipes/linux/linux-bug20-linaro_2.6.35.bb: linux-bug20-linaro_2.6.35: Add experimental recipe Feb 03 14:00:03 03Stefan Schmidt  07master * r75f081b040 10openembedded.git/recipes/buglabs/bug2v4l2_0.01.bb: bug2v4l2: Add recipe bug20 v4l2 interface Feb 03 14:00:04 03Stefan Schmidt  07master * rbeba6b5291 10openembedded.git/conf/distro/include/buglabs-autorev.inc: buglabs-autorev.inc: Add include file for BugLabs AUTOREV builds Feb 03 14:00:06 03Stefan Schmidt  07master * r88609ddb59 10openembedded.git/recipes/buglabs-apps/com.buglabs.app.bugdash2.bb: com.buglabs.app.bugdash2: Bump for newer version Feb 03 14:30:59 backreading, actually I'll probably only be at fosdem on sunday Feb 03 14:46:20 03Stefan Schmidt  07master * rd0173ea4bd 10openembedded.git/conf/distro/include/angstrom-2008-preferred-versions.inc: Feb 03 14:46:20 xserver-nodm-init: switch to ndm-init 2.0 to get a working matchbox session under Xorg Feb 03 14:46:20 Signed-off-by: Stefan Schmidt Feb 03 14:46:20 Acked-by: Graeme Gregory Feb 03 14:46:20 Acked-by: Koen Kooi Feb 03 14:46:32 03Stefan Schmidt  07master * r1dffad26a4 10openembedded.git/recipes/connman/connman_git.bb: Feb 03 14:46:32 connman: Don't prefer connman git version over the latest release in angstrom Feb 03 14:46:32 Signed-off-by: Stefan Schmidt Feb 03 14:46:32 Acked-by: Graeme Gregory Feb 03 14:46:32 Acked-by: Koen Kooi Feb 03 14:48:34 likewise: are you going on saturday or sunday ? Feb 03 15:01:28 03Steffen Sledz  07org.openembedded.dev * r93686d8afb 10openembedded.git/recipes/linux/linux-2.6.24/hipox/defconfig: Feb 03 15:01:28 linux-2.6.24: enable oprofile support/modules for hipox machine Feb 03 15:01:28 Signed-off-by: Steffen Sledz Feb 03 15:19:58 * Tartarus aims some hate at perl Feb 03 16:04:31 Hi, I'm using squashfs as my rootfs. But when I boot the image I get "opkg_conf_load: Could not create lock file /usr/lib/opkg/lock: Read-only file system.". How can I tell opkg to work in a different directory which will be mounted as rw (i.e. opt) ? Feb 03 16:39:07 re Feb 03 16:40:53 what are viable ext3 options on an embedded device, if normally run as a block dev (no mtd and the like) Feb 03 16:40:57 ? Feb 03 16:42:40 besides using data=journal or ordered and noatime (but latter is more for performance, lifetime, etc.) Feb 03 16:43:12 use ext4 Feb 03 16:43:18 without journal Feb 03 16:44:13 Helio_2000, read-only rootfs requires you to do some hacking to the various recipes Feb 03 16:44:21 and make a writable /var/run and such exist via tmpfs Feb 03 16:44:33 we really need to find a cleaner method of implementing that Feb 03 16:44:40 hi kergoth Feb 03 16:44:43 Yeah Feb 03 16:44:43 woglinde: is 2.6.32 recent enough being production ready? Feb 03 16:45:08 blindman why you are stucked at .32? Feb 03 16:45:18 woglinde: we got some issues with mass, timer based, hard power off tests and ext3 breaks usually Feb 03 16:45:36 woglinde: to say the least, it's all very complicated :) Feb 03 16:45:55 Tartarus: yes, I have mounted tmp, var/volatile and /etc as a tempfs Feb 03 16:46:06 Helio_2000, need more than that Feb 03 16:46:08 blindman get better hw Feb 03 16:46:10 /var/log and /var/run at least Feb 03 16:46:23 yes, well, those are linked into /var/volatile Feb 03 16:46:25 k Feb 03 16:46:46 hi sakoman Feb 03 16:46:47 but, however, I'm worried about opkf writing here: /usr/lib/opkg/lock Feb 03 16:46:53 *opkg Feb 03 16:47:00 Kick opkg into using /var/volatile/ or whatever :) Feb 03 16:47:21 woglinde: besides, I wasn't aware, that ext4 offers so much more in this respect, or rather, whats the minimum kernel version which should be used then? Feb 03 16:47:24 also, check the log.do_rootfs for your image Feb 03 16:47:30 hacking its recipe? isn't there a cleaner way? Feb 03 16:47:32 aha Feb 03 16:47:38 See if there's anything that's trying to be done in terms of setting things up there that could be done on the rootfs Feb 03 16:47:45 blindman sorry I cannt say if .32 is stable enough Feb 03 16:47:49 but maybee it is Feb 03 16:47:53 And then kill off on device package management Feb 03 16:47:55 I forget the var Feb 03 16:47:59 but check classes/image.bbclass Feb 03 16:48:10 blindman do you have heavy write ops on the fs? Feb 03 16:48:24 ok Tartarus I'll take a look, thank you Feb 03 16:48:30 woglinde: I meant whats a reasonably new enough kernel version Feb 03 16:48:37 woglinde: in this use case where it breaks, yes Feb 03 16:49:33 hm Feb 03 16:49:37 you could try it Feb 03 16:49:43 you have nothing to loose Feb 03 16:49:52 maybee its working better than ext3 Feb 03 16:50:05 or you can try backportin nilfs Feb 03 16:50:23 some one did it for the .29 ac100 kernel Feb 03 16:50:38 hehe, nice idea, but let's see what they think of possible earth shattering suggestions :) Feb 03 16:50:42 thanks! Feb 03 16:50:58 backporting isnt that hard Feb 03 16:51:00 and possible Feb 03 16:51:04 mom Feb 03 16:51:15 try gitouroius.org/ac100 Feb 03 17:08:33 kergoth: we could start by writing down what needs to be done each recipe to deal with read-only rootfs. I was about to start that effort for what I need. Feb 03 17:08:48 see you on fosdem (for those who will be there) Feb 03 17:08:53 kergoth: I also think things like update-alternatives can also be done build-host-side Feb 03 17:08:54 hrw: see you Feb 03 17:09:01 bye hrw Feb 03 17:09:32 Hmm, khem at FOSDEM? Feb 03 17:09:48 domnt thinkso Feb 03 17:09:55 he said he is busy this week Feb 03 17:10:45 Ah Feb 03 17:24:34 Tartarus: did you notice my ping? Feb 03 17:26:06 nope, sorry Feb 03 17:26:07 shoot Feb 03 17:27:01 ok Feb 03 17:27:04 So you see that too Feb 03 17:27:06 Debian/testing? Feb 03 17:27:40 Tartarus: http://paste.pocoo.org/show/331713/ Feb 03 17:27:47 Tartarus: gentoo Feb 03 17:27:54 What perl version? Feb 03 17:27:59 on host? Feb 03 17:28:00 Yeah Feb 03 17:28:12 And do you possibly have some other overlays going on or something? Feb 03 17:28:18 5.12.3 Feb 03 17:28:32 what machine? Feb 03 17:28:35 (target MACHINE) Feb 03 17:28:37 x86_64 Feb 03 17:28:40 and om-gta02 Feb 03 17:28:45 but it's from -native Feb 03 17:28:46 ok Feb 03 17:28:49 Yeah, I know Feb 03 17:28:50 so target shouldn't matter Feb 03 17:28:52 But I can't reproduce this Feb 03 17:29:05 I wonder if /OE/whatever is magicially a bad path? Feb 03 17:29:07 on second host with std TMPDIR I also doesn't have it Feb 03 17:29:31 I'll wrap those sed calls to see what it is trying to replace Feb 03 17:30:09 And I'll get a /OE locally Feb 03 17:45:06 hi ericben Feb 03 17:45:47 JaMa|Wrk, can you pastebin || email me your local.conf? Feb 03 17:46:09 $ ls /OE/mytmpdir/work/x86_64-linux/libxml-parser-perl-native-2.36-r4/ Feb 03 17:46:09 XML-Parser-2.36 image temp Feb 03 17:46:35 JaMa|Wrk, I'll try to push ubifs patches for n900 Feb 03 17:47:53 ericben , you take the train to go to fosdem,right? Feb 03 17:53:37 Tartarus: http://paste.pocoo.org/show/331726/ Feb 03 17:54:06 Tartarus: and it's not libxml-parser-perl-native fault (as it sources config.sh badly prepared by perl-native) Feb 03 17:54:22 Tartarus: after partial revert I in perl-native I can build libxml-parser-perl-native fine again Feb 03 17:55:10 but somehow my "echo" calls in Configure script are ignored.. more tee should help me (real tee :)) Feb 03 17:55:42 JaMa|Wrk, right, but building libxml-parser-perl-native would show a busted image directory Feb 03 17:55:51 So I'll try your local.conf in a minute Feb 03 17:55:58 (doing a full build rather than just -c install now) Feb 03 17:56:55 populate_sysroot was failing in libxml-parser-perl-native Feb 03 17:57:01 right Feb 03 18:10:06 Tartarus: found it http://paste.pocoo.org/show/331730/ every path with 4th character '/' will be broken... Feb 03 18:11:41 I hate perl so much Feb 03 18:11:45 good job! Feb 03 18:11:59 not sure about proper fix Feb 03 18:12:35 gm all Feb 03 18:13:25 Well, we can either fix userelocatableinc or go the poky wrapper script way Feb 03 18:13:39 I'm leaning towards the latter given perl is also broken with userelocatbaleinc and 5.12.x Feb 03 18:13:54 and poky has 5.12.x and updated to not have legacy staging (which I tried doing to 5.10 last night and it sucked) Feb 03 18:14:22 I don't use perl on target.. so no opinion on that, do whatever is better/easier for you Feb 03 18:14:47 hi gnutoo Feb 03 18:15:53 hi woglinde_ Feb 03 18:16:00 are you going to the FOSDEM? Feb 03 18:16:44 no Feb 03 18:17:24 it seem that there are not a lot of people going to it this year Feb 03 18:17:29 fortunately the talks are good Feb 03 18:17:41 eben moglen, teleco village, etc... Feb 03 18:20:04 I have to learn for oral exams Feb 03 18:20:34 ok Feb 03 18:20:39 I've exams too after fosdem Feb 03 18:20:54 but the italian system permit me to go to the fosdem: Feb 03 18:21:08 if I fail at the one right after fosdem there is another one Feb 03 18:21:17 if I fail to the other one there is another one later.... Feb 03 18:21:18 etc... Feb 03 18:21:27 sure Feb 03 18:21:32 but they are my 3 last one Feb 03 18:21:57 and I dont want to fail the first try Feb 03 18:22:19 but I will go to cebit Feb 03 18:22:27 ok Feb 03 18:22:54 btw is it possible to donate to openembedded with a specific goal in mind, giving a specific machine to a specific developer Feb 03 18:23:02 the developer has commit access Feb 03 18:23:35 s/mind,/mind:/ Feb 03 18:23:42 I think so, but a question for the eV? Feb 03 18:23:53 maybe I'll ask at the fosdem Feb 03 18:24:10 because I'd like morphis to get a nokia n900 to help on modem Feb 03 18:26:29 wow lots of buglabs recipes landed in oe Feb 03 18:28:43 gnutoo yes stefan does it Feb 03 18:28:49 ah crofton is landed Feb 03 18:28:51 see it now Feb 03 18:29:14 ok Feb 03 18:29:29 I landed today too but I'm not in belguim yet Feb 03 18:29:34 I'm in france currently Feb 03 18:29:43 *Belgium *France Feb 03 18:30:06 woglinde_, btw do you have a bug 2.0 at work? Feb 03 18:30:31 no Feb 03 18:30:45 ok Feb 03 18:30:51 I wanted to know how good it was Feb 03 18:32:06 hmmm Feb 03 18:32:07 | Checking for C++ library boost_signals-mt... no Feb 03 18:32:09 pingus Feb 03 18:32:27 I'm not sure it's worth to fix because it has a memory leak Feb 03 18:32:32 (pingus) Feb 03 18:32:53 maybe I should spend some time to track that memory leak Feb 03 18:33:07 is there a quick way to do that? Feb 03 18:33:09 on arm Feb 03 18:33:24 valingrid arm support has landed upstream? Feb 03 18:33:47 I remember vaguely some announcement on their website Feb 03 18:34:00 I know using it slows down the execution Feb 03 18:34:09 but does it increase memory usage drastically? Feb 03 18:34:18 because I've only 256M on my best machine Feb 03 18:34:23 *target machine Feb 03 18:37:28 kergoth ping Feb 03 18:38:50 ok seem easy to fix Feb 03 18:38:55 I'll try to remove the mt Feb 03 18:39:05 which is for multithread if I remember well Feb 03 18:39:20 woglinde_: pong Feb 03 18:40:22 kergoth hm could please take a look into the jikes_1.22.bb Feb 03 18:40:56 seems they RDENDS in it influences the native recipe too Feb 03 18:41:28 ah okay Feb 03 18:41:32 first bug fixed Feb 03 18:42:01 woglinde_: ${PN} versions of things end up obeyed even if PACKAGES is empty right now Feb 03 18:42:16 woglinde_: which means those can affect native/cross unless you use virtclass overrides to blow it away Feb 03 18:42:20 for now, anyway Feb 03 18:42:21 (afaik) Feb 03 18:42:29 runtime bits for native/cross is a bit of a mess atm Feb 03 18:42:40 yes I changed it to _PN_virt Feb 03 18:42:42 and its gone Feb 03 18:42:48 might be good if the classes also adjusted the names of the rdepends the way it does for the depends Feb 03 18:42:54 though thats slightly more complex Feb 03 18:42:59 (due to -dev, etc postfixes) Feb 03 18:45:52 at this point (almost a year later) do we need the openembedded.org/snapshots ? Feb 03 18:47:31 * kergoth doubts it, but shrugs Feb 03 18:48:03 why the hell help2man-native depends on perl? Feb 03 18:48:26 should dep on perl-native i'd think Feb 03 18:49:00 ah another RDEPENDS Feb 03 18:50:31 so looks good now Feb 03 18:50:52 auditing the .dot's a bit? i've done that a bit in the past, its a mess :\ Feb 03 18:51:06 hehe Feb 03 18:51:14 with pn- its easy Feb 03 18:52:00 and I am now familiar with what to look for Feb 03 18:53:00 the native runtime problem needs a real solution at some point, but not sure what the right answer is Feb 03 18:54:43 I will commit it later Feb 03 18:55:17 i can't remember why _${PN} is handled specially Feb 03 18:55:30 there's a couple loops in bitbake that do packages + [pn] Feb 03 18:55:30 dont know either Feb 03 18:55:36 which seems strange Feb 03 18:55:47 but adding _virtclass = "" helps Feb 03 18:55:51 * kergoth nods Feb 03 18:56:03 this should be an faq or something Feb 03 18:56:04 heh Feb 03 18:58:29 03Tom Rini  07org.openembedded.dev * r4c82344174 10openembedded.git/recipes/perl/perl-native_5.10.1.bb: Feb 03 18:58:29 perl-native: Switch to using create_wrapper to make perl relocatable. Feb 03 18:58:29 This is taken from poky. In addition to problems in upstream Feb 03 18:58:29 perl 5.12.x, Martin Jansa discovered that the regex perl uses during Feb 03 18:58:29 install is broken for things like /OE/whatever (due to not escapipng the Feb 03 18:58:30 periods in the regex perl uses). Feb 03 18:58:31 Signed-off-by: Tom Rini Feb 03 18:58:40 03Tom Rini  07org.openembedded.dev * r2ba8864662 10openembedded.git/recipes/perl/perl_5.10.1.bb: Feb 03 18:58:40 perl: Make this not be MACHINE-specific in pstaging+signatures Feb 03 18:58:40 Based on d332f1881f362cef88f2b9f40186482ac612c2d0 and Feb 03 18:58:40 b02a4addca8a4727767e24953b98f30399e3592b in poky. Feb 03 18:58:40 Signed-off-by: Tom Rini Feb 03 19:00:03 JaMa|Wrk, confirmed that those changes still work on my slow laptop, pulling and building on my faster box w/ /OE/mystuff being used for TMPDIR now Feb 03 19:01:15 tartarus didnt you made some patches to help3man? Feb 03 19:01:28 can I add that in mkfs.ubifs args "-x zlib" Feb 03 19:01:33 JaMa|Wrk, ^^^ Feb 03 19:01:39 woglinde_, yes Feb 03 19:01:55 Making it happier about using our perl-native Feb 03 19:01:58 What's broken now? Feb 03 19:02:33 Tartarus: thanks.. trying Feb 03 19:02:52 GNUtoo|laptop: I guess so Feb 03 19:03:14 ok Feb 03 19:03:16 GNUtoo|laptop: iirc I've used exactly that when trying different comp alg ratio Feb 03 19:03:21 ok Feb 03 19:03:25 so I hadrdcode it Feb 03 19:03:29 *hardcode Feb 03 19:06:17 Tartarus: configured and config.sh looks sane now Feb 03 19:10:23 NOTE: package libxml-parser-perl-native-2.36-r4: task do_populate_sysroot: Succeeded Feb 03 19:10:24 NOTE: Tasks Summary: Attempted 108 tasks of which 0 didn't need to be rerun and 0 failed. Feb 03 19:10:26 too :) Feb 03 19:21:57 Tartarus: I have seen one perl hater on #yocto as well Feb 03 19:22:09 so, you may hate it together :) Feb 03 19:22:35 fray? :) Feb 03 19:22:41 yes? Feb 03 19:22:43 may be :) Feb 03 19:22:44 known him since the 90s :) Feb 03 19:22:47 Perl, how to make something so obscure that even the author can't maintain it. Feb 03 19:22:50 can't remember :) Feb 03 19:23:05 I've hated perl with a passion since early 2000's.. when I had to cross compile it for the first time Feb 03 19:23:20 I only had mild hatred in the mid/late 90's Feb 03 19:23:23 btw, is python better for cross-compiling? Feb 03 19:23:28 much much better Feb 03 19:23:34 ya, Feb 03 19:23:37 hm.. Feb 03 19:23:41 on the slug we had to build perl native Feb 03 19:23:46 it was a mess Feb 03 19:23:51 I think everything that was done to perl was to make it as difficult as possible to cross comile Feb 03 19:24:03 the test suite kills it from building cross Feb 03 19:24:08 (or did anyways) Feb 03 19:25:25 it's configure.. going from ia32 to mips.. it picks up all of the signals and system configuration information on the IA32 host.. Feb 03 19:25:33 you then cross compile, and if you are lucky you get a inary.. Feb 03 19:25:43 eww Feb 03 19:25:48 which then proceeds not to work, because system capabilities and signal definitions are all COMPLETELY BROKEN Feb 03 19:26:12 I'm no fan of the GNU autotools, but good lord, they already solved this problem.. perl has taken the approach of screw everyone.. Feb 03 19:26:28 Tartarus, is meta- stable enough to mirror? Feb 03 19:26:46 fray: try cmake, after that you become a huge fan of GNU autotools ;> Feb 03 19:26:54 cmake is definetely on my shitlist :> Feb 03 19:27:13 ka6sox, a question for Koen primarily, since he's the biggest worker atm Feb 03 19:27:16 and he's at FOSDEM Feb 03 19:27:23 We could always add it later, too Feb 03 19:27:34 Tartarus, true Feb 03 19:27:39 Jin^eLD hahah ya, cmake is pretty bad Feb 03 19:27:40 okay so I'm setting this up as RO Feb 03 19:28:08 I dunno perl just burned me so many times that the hatred burns strong.. ALMOST as bad as the hatred of Solaris.. Feb 03 19:28:41 I'm sure Tartarus can tell you stories of me getting enough GNU components built on solaris to create an environment to cross compile Linux binaries in.. and the house of pain it took to do it, and then maintain it.. Feb 03 19:29:03 you'd thing (this was back in 2002 or so) that in this day Sun could at LEAST use a POSIX /bin/sh Feb 03 19:29:50 :)) Feb 03 19:33:15 Tartarus is stiring up bad and painful memory from my previous employment.. ;) Feb 03 19:34:18 Tartarus, I don't see a way for github to pull so I'll work up a post-commit hook for git.oe.org to push up changes. Feb 03 19:35:03 and we will need something similar for bitbake. Feb 03 19:35:54 ka6sox, Well, maybe they just cron a pull every N hours? Feb 03 19:36:29 on github can't do that... Feb 03 19:36:31 I was gonna follow up and see what's next, heh Feb 03 19:36:42 Well, maybe the mirror staff can? Feb 03 19:36:55 ya, I will contact them about this Feb 03 19:37:00 k Feb 03 19:37:40 I thought the purpose was to have it sync'd pretty close so that .eu could pull out of github but push back to OE directly Feb 03 19:38:55 Tartarus: libxml-parser-perl-native also rebuilt here now just fine, thanks again Feb 03 19:43:22 fray: seems I'll repeat your Solaris experience but with FreeBSD.. Feb 03 19:43:30 hahah Feb 03 19:43:45 * fray oddly enough passed a car on the highway the other day with "FreeBSD" (MN) plates.. Feb 03 19:44:13 the guy looked early 20's late teens.. (Mazda 3 wagon)... Feb 03 19:44:25 not what I was expecting.. was thinking more someone who looked like Stallman would be driving.. ;) Feb 03 19:45:49 Tartarus, I see how to do it pretty easily...I can make it a post-commit hook on oe's git. Feb 03 19:46:32 yeah Feb 03 19:47:18 github becomes a push-mirror so the primary needs to push up changes. Feb 03 19:48:32 03Denis 'GNUtoo' Carikli  07org.openembedded.dev * r08d47fa786 10openembedded.git/conf/machine/nokia900.conf: Feb 03 19:48:32 nokia900.conf: added ubifs arguments Feb 03 19:48:32 Thanks JaMa on #openmoko-cdevel on Freenode on irc for help on the sub-pages: Feb 03 19:48:32 Feb 02 11:41:58 GNUtoo|laptop: also for gta02 I had to sacrifise sub-pages, -s same as -m otherwise it always failed Feb 03 19:48:32 Signed-off-by: Denis 'GNUtoo' Carikli Feb 03 19:48:35 github? Feb 03 19:48:45 why not gitorious too? Feb 03 19:49:26 at least gitorious publish its source code, I think github doesn't Feb 03 19:53:35 ka6sox, hi, what about gitorious? Feb 03 19:54:02 gitorious seem better than github Feb 03 19:54:17 GNUtoo|laptop, someone needs to see about doing it Feb 03 19:54:25 ok Feb 03 19:54:35 but we will stay on openembedded.org? Feb 03 19:54:41 or will we move at the end Feb 03 19:54:42 I can do it but I just need focused direction. Feb 03 19:54:48 ok Feb 03 19:54:53 GNUtoo|laptop, what I'm doing is setting up a mirror Feb 03 19:54:54 RO Feb 03 19:54:59 so master is still ours Feb 03 19:55:09 my concern is rather that I don't like to depend on a service that we have no control over Feb 03 19:55:15 ok Feb 03 19:55:18 but in the future Feb 03 19:55:24 are there any plans to move? Feb 03 19:55:30 GNUtoo|laptop, dunno Feb 03 19:55:36 because it could start with a mirror Feb 03 19:55:45 and at the end we could end up on beeing on github Feb 03 19:55:56 right now all I know is that we need better service for pulls into .eu Feb 03 19:56:02 It's just a mirror Feb 03 19:56:08 since there's 20 clones on github atm Feb 03 19:56:09 ok Feb 03 19:56:12 and another bunch on gitoriousis Feb 03 19:56:21 then go on Feb 03 19:56:40 JaMa|Wrk, uses gitorious for instance Feb 03 19:56:50 but I use only the main mirror Feb 03 19:56:55 s/mirror/repo Feb 03 19:57:05 so I won't object for github or gitorious Feb 03 19:57:19 it was just to know where we were going Feb 03 19:57:26 GNUtoo|laptop, if we get them synced up we should be able to make it so that you pull from github and only go to git.oe.org to push Feb 03 19:57:37 ok nice Feb 03 19:59:32 I'm using it only, because there already was angstrom mirror and gitorious allows fast clone from existing projects Feb 03 19:59:41 ok Feb 03 20:00:04 but I can move my branch to github too Feb 03 20:00:46 or oe could come to gitorious too if it's not too much work Feb 03 20:00:55 thanks ka6sox for your work Feb 03 20:01:37 re Feb 03 20:01:39 I think the point is to make it is easier for github/gitorious to support people with OE repos on them Feb 03 20:01:42 re crofton Feb 03 20:01:46 hi woglinde_ Feb 03 20:03:10 crofton how was the flight? Feb 03 20:04:11 good Feb 03 20:04:17 Crofton so do we want a mirror on gitorious too? Feb 03 20:04:21 plenty of space on the plane Feb 03 20:04:30 ka6sox, if it is easy Feb 03 20:04:35 kk Feb 03 20:04:40 and assuming we have people using the service Feb 03 20:05:50 yes Feb 03 20:05:52 we do Feb 03 20:06:00 okay I just setup a new user for us there. Feb 03 20:06:09 and will work on adding that hook too. Feb 03 20:16:13 03Frans Meulenbroeks  07org.openembedded.dev * r4d1320406f 10openembedded.git/conf/distro/minimal.conf: Feb 03 20:16:13 minimal.conf: removed pinning of perl 5.10.1 now it has become the default Feb 03 20:16:13 as discussed with and agreed by Mickey. Feb 03 20:16:13 Signed-off-by: Frans Meulenbroeks Feb 03 20:19:55 Wow. recipes/images/initramfs-minimal-image.bb as target and 1415 packages (or steps?) and few GB of space. Doesn't exist something _more_ minimalistic? Feb 03 20:21:22 less than an initramfs? Feb 03 20:21:40 less than 1415 packages Feb 03 20:23:09 that's steps IIRC Feb 03 20:24:09 at least in my case I think I didn't get lower than 900 or 1200 when building the initramfs with a clean slate Feb 03 20:24:14 each package has diffrent stages Feb 03 20:24:19 the toolchains etc take up the most in this case Feb 03 20:24:33 yes complete toolchain is build Feb 03 20:24:45 you could speed it up with an external toolchain Feb 03 20:24:52 but thats not really recommended Feb 03 20:26:26 dm8tbr: Ok, then steps. I just need some cross-compile environment, uClibc, busybox and one specific package. 5 or 6 steps? Even kernel I've got my own. Feb 03 20:27:04 you can make bitbake show you what it's going to do Feb 03 20:27:14 that could help you to understand Feb 03 20:29:57 dm8tbr: Yup, openssl, mtd-utils, squashfs-tools-native, usw. Doesn't sounds too minimal to me. Feb 03 20:29:59 petrvlasic: 1415 tasks, not packages. every recipe has 5+, for fetch/unpack/patch/etc. and it has to build some build machine tools since we can't rely on everyone having everything Feb 03 20:36:49 (e.g. squashfs-tools-native to be able to construct a squashfs filesystem) Feb 03 20:37:00 kergoth: Ok, but more than 10GB for minimal build? I don't rather want to know how maximal build looks like. Feb 03 20:37:24 it doesn't get much much more Feb 03 20:37:35 dm8tbr: Perfect :-) Feb 03 20:37:58 for my archos devices the whole directory never exceeded 20GB IIRC Feb 03 20:38:07 kergoth, where is bitbake's master repo? Feb 03 20:38:14 (where is it housed) Feb 03 20:38:18 ka6sox: git.openembedded.org Feb 03 20:38:20 where else? Feb 03 20:38:27 petrvlasic you can inherit rm_work so the sources will be deleted after compiling Feb 03 20:38:31 oh...sorry Feb 03 20:38:32 nm Feb 03 20:38:41 but adds some taks Feb 03 20:38:47 okay I'll add the hooks there too. Feb 03 20:38:55 :) Feb 03 20:39:05 kergoth: I don't want squashfs. In initramfs? For what? Feb 03 20:39:18 petrvlasic: run bitbake -g and see what's pulling it in Feb 03 20:39:31 -g generates .dot files for graphviz, so you can see what pulls in what Feb 03 20:40:21 kergoth: No more. Last time, Graphviz froze to death. Feb 03 20:40:34 I don't use graphviz personally, i just edit the .dot files ;) Feb 03 20:40:36 its just text Feb 03 20:40:48 but its still a decent way to inspect the dependency graph Feb 03 20:40:50 kergoth: I know :-) Feb 03 20:40:57 if you're using master, you can also use bitbake -u depexp Feb 03 20:41:02 but its pretty basic Feb 03 20:41:07 s/master/bitbake master/ Feb 03 20:41:36 kergoth: Just to illustrate. Feb 03 20:42:02 kergoth: Ok, thanks much. Feb 03 20:42:15 everything is pulled in for a reason. if you're not happy with what initramfs-minimal does, copy it and modify it to only include what you want Feb 03 20:42:23 i'd suggest doing taht Feb 03 20:42:25 blah, can't type Feb 03 20:42:54 i agree we need to do some house cleaning, some things need to be made optional in the dependency graph, and there are some cases where deps exist accidentally, not intentionally Feb 03 20:47:02 hm as I said help2man drags in perl Feb 03 20:47:11 ups help2man-native Feb 03 20:50:01 hi guys, any quick suggestions about: Perl lib version (5.10.1) doesn't match executable version (v5.10.0) at /home/builds/jornada7xx/tmp/sysroots/x86_64-linux/usr/lib/perl/5.10.1/Config.pm line 50. ? Feb 03 20:50:12 I've started with a clean build from a recent git pull Feb 03 20:52:03 filip hm no Feb 03 20:52:16 maybe jama or tartarus knows Feb 03 20:52:36 it seems like it's using hosts' perl Feb 03 20:53:18 hm could be tartarus was complaining about something Feb 03 20:53:28 but I think it was 5.12 perl Feb 03 20:54:08 the gnu-configize script starts with: #! /usr/bin/perl -w Feb 03 20:54:49 should it be altered with /usr/bin/env perl -w? Feb 03 20:55:03 or oe has some nice hack for absolute paths in shebangs? Feb 03 20:55:15 Just /usr/bin/env perl Feb 03 20:55:15 hm I though it was checked in already Feb 03 20:55:16 drop the -w Feb 03 20:55:35 so somebody has fixed it already? Feb 03 20:55:46 Maybe... Feb 03 20:55:54 What's the last commit in your tree? Feb 03 20:56:25 08d47fa786ee535345118a2680d64b5e1ccd97a3 from Thu Feb 3 20:42:59 2011 +0100 Feb 03 20:56:29 hmmm Feb 03 20:56:51 I had this problem with pingus Feb 03 20:57:02 there is a patch switching to -mt version of boost Feb 03 20:57:06 if I drop it it works fine Feb 03 20:57:19 gnutoo hm Feb 03 20:58:05 woglinde, and it was you that added that patch according to git log Feb 03 20:58:12 basically for SHR Feb 03 20:58:17 on nokia900 Feb 03 20:58:22 in my sysroot Feb 03 20:58:26 the libs are not -mt Feb 03 20:58:31 maybe because of newer boost Feb 03 20:58:41 I don't know enough the mt/non-mt distro coverage Feb 03 20:58:45 so I don't know what to do Feb 03 20:58:51 Tartarus: are you about to push the fix maybe? Feb 03 20:59:05 I'll look into boost recipes Feb 03 21:00:37 boost 1.45 with bjam seem to use mt Feb 03 21:00:41 I'll doubble check Feb 03 21:01:38 ah no it's only for threads Feb 03 21:01:41 not for signal Feb 03 21:01:49 and pingus depend on sigals right? Feb 03 21:02:18 ouch there is a do_stage Feb 03 21:02:50 wikipedia says lastest release is 1.45 Feb 03 21:03:09 2 month ago Feb 03 21:03:15 so why is there a do_stage Feb 03 21:03:17 * GNUtoo|laptop git blame Feb 03 21:03:57 hmmm it's khem....strnage Feb 03 21:04:00 *strange Feb 03 21:04:21 as far as I understood we should not use do_stage but maybe this bjam is special? Feb 03 21:04:46 nevertheless it's so great to have a recent boost Feb 03 21:06:55 (Author: Ben Gamari but Signed-off-by: Khem Raj ) Feb 03 21:16:53 Tartarus: do you want to ack the patch for insane.bbclass? Feb 03 21:17:00 ah denix Feb 03 21:18:20 okay Feb 03 21:18:23 its now typo Feb 03 21:19:54 woglinde: btw, did you have a question? Feb 03 21:20:33 whoa, eval 'case $# in 0) exec /usr/bin/perl -S "$0";; *) exec /usr/bin/perl -S "$0" "$@";; esac' Feb 03 21:20:36 what is THIS? Feb 03 21:22:06 denix nope I tought there was a spelling error in your last commit from chase Feb 03 21:22:08 but it isnt Feb 03 21:22:53 ok Feb 03 21:23:39 sourceipk is quite novel and not many use it, so it was safe to push it anyway... Feb 03 21:24:05 no kernelones Feb 03 21:24:14 with mulitple targets Feb 03 21:26:18 woglinde: so was there a spelling error in my patch? Feb 03 21:26:27 no Feb 03 21:26:31 I thought there was Feb 03 21:26:35 but isnt Feb 03 21:26:38 OK Feb 03 21:26:47 just wanted to make sure I didn't mess something up :) Feb 03 21:26:53 scripk Feb 03 21:27:02 looks strange Feb 03 21:30:56 quite strange :) Feb 03 21:31:07 kinda like .src.rpm :) Feb 03 21:31:43 note script Feb 03 21:31:53 looks really similiar Feb 03 21:34:36 03Henning Heinold  07org.openembedded.dev * rf06b04c2b5 10openembedded.git/recipes/help2man/help2man_1.38.2.bb: help2man: fix drag in of target perl for the native package Feb 03 21:56:38 good nite Feb 03 23:14:14 anyone know how big the Angstrom-distribution package repo is? Feb 03 23:30:44 ka6sox: XorA|gone or Koen should know :) Feb 03 23:31:21 Jay7, Koen told me where to find out. Feb 03 23:32:44 wow, Koen is talking with you! You are choosen! ;) Feb 03 23:32:51 * Jay7 hides **** ENDING LOGGING AT Fri Feb 04 02:59:57 2011