**** BEGIN LOGGING AT Wed Mar 09 02:59:57 2011 Mar 09 08:11:37 good morning Mar 09 08:30:05 gm Mar 09 08:32:01 likewise: hi Mar 09 08:40:34 hi mckoan Mar 09 08:40:46 mckoan: did you go to EW2011? Mar 09 08:43:43 hi Mar 09 08:52:44 likewise: not this year Mar 09 09:10:20 likewise: netherlands is playing India today at ICC Cricket World Cup :) Mar 09 09:11:36 khem: I still have to catchup on the Biathlon world championships :-) Mar 09 09:11:58 khem: The Dutch play cricket? Gnarf. Mar 09 09:12:22 heh I know India use to be good at hockey but then Dutch became better and better Mar 09 09:12:24 khem: that the best may win Mar 09 09:12:49 yes most of the team is south african/dutch dual nationals Mar 09 09:16:42 fennec is causing some trouble: doesn't compile with current recipe revision and when bumping SRC_URI revisions there's extra trouble. should I put it on bugzilla or rather ML ? Mar 09 09:20:58 fennec (in org.oe.dev HEAD) depends on autoconf213-native which fails to build. when worked around by installing host distro autoconf213 fennec fails compiling with make[5]: *** No rule to make target `-lX11', needed by `mozilla-xremote-client' Mar 09 09:22:45 morning Mar 09 09:23:33 hrw: hi Mar 09 09:25:22 mckoan: I have one more suggestion for koalapc cases - provide atx24 connector instead of atx20 one Mar 09 09:29:10 kergoth: remember as I said that something is wrong with do_install in linux-libc-headers and bitbake/master? after downgrading to bitbake-1.10 I got usable error log http://paste.pocoo.org/show/350706/ Mar 09 09:29:15 what does this command do? MACHINE=beagleboard ./oebb.sh bitbake virtual/kernel Mar 09 09:29:48 kergoth: don't know why 1.12 and master doesn't show anything.. and after adding symlinkg fakeroot -> pseudo I was able to build linux-libc-headers Mar 09 09:30:31 hrw: Thx :-D Mar 09 09:31:49 is it hanging here? http://pastebin.com/yknbkmB9 Mar 09 09:33:30 mckoan: I have to buy atx20->atx24 adapter and this will make cable mess even bigger Mar 09 09:35:21 srilatha: that will 'bitbake' the prefered kernel package for the BeagleBoard based on the recipes you have in your tree (virtual/kernel is a virtual recipe that always points ot the prefered kernel) Mar 09 09:35:45 srilatha: should build a kernel for beagleboard Mar 09 09:36:07 dcordes_, from where does it take the kernel source code? Mar 09 09:36:41 NotTooDumb3: you can check in the machine config which kernel bitbake will pick Mar 09 09:36:59 NotTooDumb3: conf/machine/beagleboard.conf Mar 09 09:37:02 NotTooDumb3: it takes it from the source defined in the recipe (in the case of most BeagleBoard kernels, Linux-OMAP GIT or the Ti BSP GIT trees). Mar 09 09:37:10 i am running the above statement, i am confused whether it's running or hanging at 0% Mar 09 09:37:19 PREFERRED_PROVIDER_virtual/kernel = "linux-omap-psp" Mar 09 09:38:24 NotTooDumb3: recipes/linux/linux-omap-psp_2.6.32.bb there you can find source code Mar 09 09:39:04 it applies quite a load of patches Mar 09 09:39:26 conf/machine/beagleboard.conf what is path for this? from which folder did you move? Mar 09 09:39:39 you might see more progress information by turning on additional verbosity via conf/local.conf (see conf/local.conf.example comentary) Mar 09 09:39:51 dcordes_, is oe easy to use with beagle board or any other boards? Mar 09 09:39:58 NotTooDumb3: from your oe main directory Mar 09 09:40:33 ok i was searching in bitbake main directory..how are these two related? Mar 09 09:41:31 bitbake is the tool that does all the work and it needs the oe metadata (recipes, conf) to do so Mar 09 09:41:43 is : recipes/linux/linux-omap-psp_2.6.32.bb from conf/machine/beagleboard.conf? i searched for recipe and gives no result Mar 09 09:42:24 NotTooDumb3: not exactly. search for "linux-omap-psp" only (Without path and version #) Mar 09 09:42:54 oh yes this line is there PREFERRED_PROVIDER_virtual/kernel = "linux-omap-psp" Mar 09 09:43:14 bitbake uses these target names only. it will pick the latest version (unless instructed otherwise) and will find the path on its own Mar 09 09:43:19 exactly Mar 09 09:43:20 using oe, can i compile the source code i want without from these standard sources? Mar 09 09:43:58 do you have any idea what is uImage.bin file? Mar 09 09:44:06 yes you could create a new recipe to do so and change the PREFERRED_PROVIDER_virtual/kernel Mar 09 09:44:37 uImage.bin file is a kernel image that is made for use with u-boot bootloader Mar 09 09:44:41 and does MACHINE=beagleboard ./oebb.sh bitbake virtual/kernel give me MLO, u-boo.bin, uImage? can i take any file system along with these files to boot my board? Mar 09 09:45:35 u-boot can also boot from uImage instead of uImage.bin right? if i want my u-boot to boot using uImage instead of uImage.bin what do i need to change? Mar 09 09:45:49 I am not sure but it will possibly emmit these things as they are needed to boot. It depends on the kenrel configuration which filesystem you can use Mar 09 09:46:55 dcordes_, i am confused whether my virtual/kernel is running or hanged? is there any way to know that? Mar 09 09:47:10 you should consult documentation specific for your machine in order to find answers for such questions. please check http://elinux.org it's rich in beagle doku Mar 09 09:47:46 I think u-boot config is written directly to beagle nand or read from sd ... Mar 09 09:48:02 but check that wiki please. it has everything to get started for bb dev Mar 09 09:48:14 10:38 < dcordes_> you might see more progress information by turning on additional verbosity via conf/local.conf (see conf/local.conf.example comentary) Mar 09 09:48:43 note that kernel does take long: checking out source code and compilation both are large processes Mar 09 09:50:48 oh..ok PREFERRED_PROVIDER_virtual/kernel = "linux-omap-psp" in this statement does it checkout linux-omap-psp?? when i give my own source code, what do i need to give here? path? no i think right Mar 09 09:52:57 go to the recipes/linux/ directory and read the linux-qsd8 kernel recipe. it's the most simple kernel recipe I know. on that basis you can create your own kernel recipes. this assumes a minimum command of the git program Mar 09 09:57:40 NotTooDumb3: http://docs.openembedded.org/usermanual/html/ Mar 09 09:59:43 ok thank you Mar 09 10:00:38 you're most welcome Mar 09 10:20:13 fennec problems documented in bugzilla: !oebug 5529 Mar 09 10:21:07 autconf213: #oebug 5527 Mar 09 10:53:40 when do_create_srcipk, it says unable to create directory, what could be the problem Mar 09 11:18:51 GNUtoo|laptop: let's discuss here Mar 09 11:19:08 GNUtoo|laptop: it's fairly easy to bump fennec srcrev. Mar 09 11:19:36 GNUtoo|laptop: if you look at the commit history you will notice koen only changed the revision hashes and bumped PR Mar 09 11:19:54 yes but you told it had additional deps Mar 09 11:20:04 so I'll try to fix the current well-known one Mar 09 11:20:08 good Mar 09 11:20:17 but my connection is too slow as I said Mar 09 11:20:33 and I use autorev Mar 09 11:20:43 bitbake -c clean autoconf-native Mar 09 11:20:47 I did that some time ago Mar 09 11:20:53 and it's still parsing Mar 09 11:20:56 because of autorev Mar 09 11:20:59 and slow connection Mar 09 11:21:14 wait, you use non static revisions for _everything_ in OE ?? Mar 09 11:21:17 so I'm afraid I won't be of great help Mar 09 11:21:27 I'll explain Mar 09 11:21:28 no Mar 09 11:21:55 basically it has to know the revision of everything at parsing time Mar 09 11:22:12 so during parsing time it goes fetch the last revision number or hash Mar 09 11:22:26 of the autorev stuff Mar 09 11:22:28 since I use SHR Mar 09 11:22:37 which has autorev for both fso and shr stuff Mar 09 11:22:51 it goes fetch the infos about what was the last commit Mar 09 11:22:58 of a lot of stuff Mar 09 11:23:07 since my connection is *very * slow Mar 09 11:23:10 it takes ages Mar 09 11:23:14 require conf/distro/include/fso-autorev.inc Mar 09 11:23:14 require conf/distro/include/shr-autorev.inc Mar 09 11:23:32 ah I get it. but that's not directly relevant to fennec right Mar 09 11:23:50 no Mar 09 11:23:55 your point is your slow ok :) Mar 09 11:24:04 it's because I included the 2 lines above Mar 09 11:24:22 ya Mar 09 11:24:24 see still not finished parsing Mar 09 11:25:41 ok thanks. it would be awesome to fix http://bugzilla.openembedded.org/show_bug.cgi?id=5529 Mar 09 11:26:06 dcordes_, how to check that it's my provider and not something else like some router down on the internet? Mar 09 11:26:24 because some time ago there was a big router provider that went down Mar 09 11:26:38 it affected europe a lot Mar 09 11:26:46 during some minutes/hours Mar 09 11:27:21 I don't know I do everything via HSDPA Mar 09 11:27:42 hmmm Mar 09 11:27:45 I could do that Mar 09 11:27:46 pci device in my laptop connecting directly. vodafone.de does all the routing for me Mar 09 11:27:54 ok Mar 09 11:28:40 since I need a server and download huge ammount of sources I can't rely on 3g Mar 09 11:28:57 I could try to catch a neighbourg wifi Mar 09 11:29:49 that's illegal in Germany even if it's non encrypted Mar 09 11:30:01 your bug report is not verbose enough Mar 09 11:30:07 ok what is missing ? Mar 09 11:30:35 mine is not encrypted either Mar 09 11:30:36 anyway Mar 09 11:31:01 more context Mar 09 11:31:18 maybe attach the whole log.do.compile Mar 09 11:31:46 ah wait Mar 09 11:31:48 it tell where Mar 09 11:31:49 `mozilla-xremote-client Mar 09 11:32:00 dcordes_, ^^^ Mar 09 11:32:27 still not parsed Mar 09 11:32:30 it's really slow Mar 09 11:33:07 I interrupt that's too much Mar 09 11:35:05 GNUtoo|laptop: is "No rule to make target `-lX11'" libx11 ? Mar 09 11:35:27 -l find the lib with Mar 09 11:35:32 for instance Mar 09 11:35:34 -lfoo Mar 09 11:35:41 will find libfoo Mar 09 11:35:48 -lz will find libz Mar 09 11:35:49 etc... Mar 09 11:36:03 so here Mar 09 11:36:06 libX11 Mar 09 11:36:10 with a big X Mar 09 11:37:57 recipes/xorg-lib/libx11_1.4.1.bb Mar 09 11:38:16 lrwxrwxrwx 1 gnutoo gnutoo 15 2011-01-15 16:10 libX11.so -> libX11.so.6.3.0 Mar 09 11:38:22 in STAGING_LIBDIR Mar 09 11:40:19 how can I find STAGING_LIBDIR ? Mar 09 12:15:15 GNUtoo|laptop: still around ? Mar 09 12:24:53 dcordes_, I was eating Mar 09 12:25:04 STAGING_LIBDIR is at: Mar 09 12:25:23 .../sysroots/armv7a-oe-linux-gnueabi/usr/lib/ Mar 09 12:55:06 dcordes_, It seem to work ok again(my internet connection) Mar 09 12:57:44 hmmm Mar 09 12:57:47 strange strange Mar 09 12:57:52 GNUtoo|laptop: GNUtoo|laptop ln: creating symbolic link `libX11.so': File exists Mar 09 12:58:04 wait a sec Mar 09 12:58:10 I've another problem before Mar 09 12:58:28 | + sed -i '-es,^SUBDIRS = .*,SUBDIRS = bin . lib,' /home/gnutoo/embedded/oe/oetmps/nokia900/work/x86_64-linux/autoconf213-native-2.13-r13.0/autoconf-2.13/Makefile.am Mar 09 12:58:34 /home/gnutoo/embedded/oe/oetmps/nokia900/work/x86_64-linux/autoconf213-native-2.13-r13.0/autoconf-2.13/Makefile.am doesn't exist Mar 09 12:58:48 yes I documented it in bugzilla. the fennec bug depends on it Mar 09 12:58:54 but the Makefile.in exists Mar 09 12:59:02 so I bet I've to modify that bb Mar 09 12:59:05 http://bugzilla.openembedded.org/show_bug.cgi?id=5527 Mar 09 12:59:33 all newer autoconf versions have Makefile.am Mar 09 12:59:49 it is not some upstream problem Mar 09 13:00:07 I tried to locate a change that could affect this in the git log Mar 09 13:00:11 but no luck Mar 09 13:00:53 workaround: install autoconf213 package in your host and do Mar 09 13:01:19 require conf/distro/include/fso-autorev.inc Mar 09 13:01:20 require conf/distro/include/shr-autorev.inc Mar 09 13:01:45 sorry, wrong lines Mar 09 13:01:53 ASSUME_PROVIDED += "autoconf213" Mar 09 13:01:53 ASSUME_PROVIDED += "autoconf213-native" Mar 09 13:02:04 why? Mar 09 13:02:25 that just skips the dep Mar 09 13:02:33 and assume you provide it in your host Mar 09 13:02:39 and use the host autoconf Mar 09 13:02:55 => not a good idea Mar 09 13:03:09 GNUtoo|laptop: as I said "workaround:" Mar 09 13:03:27 I don't know how to solve this problem in the metadata Mar 09 13:03:36 http://www.pastie.org/1651307 Mar 09 13:03:39 better workarround Mar 09 13:03:45 bump PR also Mar 09 13:03:48 if you use that Mar 09 13:04:06 should I confirm if it works ? Mar 09 13:04:29 I'll confirm it Mar 09 13:04:34 if you wait a sec Mar 09 13:04:36 as it's parsing Mar 09 13:05:10 ouch Mar 09 13:05:13 another error Mar 09 13:05:49 I forgott the then Mar 09 13:05:52 I'll add it Mar 09 13:07:21 only versions <= 2.13 do lack Makefile.am . i.e. maybe you should not add the workaround in the .inc but in the autoconf213_2.13.bb ? Mar 09 13:07:47 it's fine in the .inc but it need some ack to push Mar 09 13:08:17 ok Mar 09 13:09:37 can you paste the working diff ? Mar 09 13:12:20 not working yet Mar 09 13:12:22 one sec Mar 09 13:13:49 http://www.pastie.org/1651345 Mar 09 13:19:37 dcordes_, it's compiling....must wait for it to finish Mar 09 13:20:20 what's next? openoffice?(just joking) Mar 09 13:22:27 kgilmer, hi, you're in japan? Mar 09 13:22:43 buglabs have offices in japan? Mar 09 13:23:19 hi GNUtoo|laptop Mar 09 13:23:41 yes, i'm in japan. if you count my small home office, then yes, we have offices in japan :) Mar 09 13:24:30 lol ok Mar 09 13:24:32 GNUtoo|laptop: Yes, as a matter of fact :) Libreoffice or whatever it is! Mar 09 13:24:32 GNUtoo|laptop: you mean wait for it to reach the libx11 bug Mar 09 13:24:44 GNUtoo|laptop: that's the second problem Mar 09 13:24:57 autconf patch doesn't apply. are you in org.oe.dev HEAD ? Mar 09 13:25:13 dcordes_, can't say it's still compiling Mar 09 13:25:15 for the patch Mar 09 13:25:24 well pastebin eat patches Mar 09 13:25:32 maybe we should do a patchbin? Mar 09 13:25:56 or you upload it Mar 09 13:26:11 kgilmer, the advantage is that you get GPS fix easily from your home Mar 09 13:26:15 dcordes_, ok Mar 09 13:28:33 ouch my internet connection is slow again....hmmmm Mar 09 13:30:16 GNUtoo|laptop: next... maybe test rmnet Mar 09 13:30:34 GNUtoo|laptop: I noticed I have /dev/qmi0 in my current test kernel so no modules needed Mar 09 13:30:44 but not sure how to debug Mar 09 13:32:29 dcordes_, | make[5]: *** No rule to make target `-lgconf-2', needed by `libmozgnome.so'. Stop. Mar 09 13:33:16 dcordes_, qmi0 worked for me Mar 09 13:33:27 sigh it's very slow Mar 09 13:33:38 it's impossible to navigate Mar 09 13:33:56 altough the website I am looking at is a bit heavy(buglabs's blog) Mar 09 13:38:12 GNUtoo|laptop: -lgconf-2 ? I didn't see that error. so you overcame libx11 err ? Mar 09 13:38:42 no idea Mar 09 13:39:10 dcordes_, try re-bitbaking without cleaning Mar 09 13:39:14 because it seem to go on Mar 09 13:39:17 when re-bitbaking Mar 09 13:39:32 a bit like the java Mar 09 13:39:34 I am having trouble with TIDmaiVideoSink Mar 09 13:39:50 I don't get it to play Mar 09 13:39:52 whats's Dmai? Mar 09 13:40:09 GNUtoo|laptop: you mean you do "bitbake fennec" it errors out, you run "bitbake fennec" and it works ? Mar 09 13:40:17 I bet your ti dmai video sink is a gstreamer thing Mar 09 13:40:21 that is some hardware accel tech on TI boards Mar 09 13:40:29 sure Mar 09 13:40:46 dcordes_, I mean it continues Mar 09 13:40:55 not sure if it won't stop with the exact same error later Mar 09 13:40:58 I must look Mar 09 13:41:35 GNUtoo|laptop: there was no way I could get paste libx11 err. no matter how often I would restart the bitbake fennec Mar 09 13:41:51 then look in staging_libdir Mar 09 13:41:53 as I told you Mar 09 13:41:55 and actually I did not see anything fix by blackmagick yet Mar 09 13:42:35 * GNUtoo|laptop doesn't like magick Mar 09 13:42:41 *magic Mar 09 13:42:57 magic is stuff you don't/can't understand Mar 09 13:43:35 as I told you: 13:56 < dcordes_> GNUtoo|laptop: GNUtoo|laptop ln: creating symbolic link `libX11.so': File exists Mar 09 13:43:53 yes I just said to look Mar 09 13:43:59 not to create a symlink Mar 09 13:44:07 it's a hint Mar 09 13:44:08 ok I looked. didn't fix it Mar 09 13:44:12 look if ith's there Mar 09 13:44:22 if it's not there depend on libx11 Mar 09 13:44:37 else pass the correct arguments to configure to make it find libx11 Mar 09 13:44:56 and that's what I can't do cause I don't understand the whole path confusion Mar 09 13:45:12 so you came across that error as well ? Mar 09 13:48:15 dcordes_, still compiling.... Mar 09 13:48:18 I must wait Mar 09 13:48:22 it's rather long Mar 09 13:50:16 ah ok :) here it takes like half an hour to get to the libx11 err Mar 09 13:53:43 dcordes_, same error Mar 09 13:53:59 ok Mar 09 13:54:25 same means this one: Mar 09 13:54:31 | make[5]: *** No rule to make target `-lgconf-2', needed by `libmozgnome.so'. Stop. Mar 09 13:55:53 dcordes_, should I bugreport in oe ml Mar 09 13:55:59 because I've not the time to solve it Mar 09 13:56:04 (it could take days fulltime) Mar 09 13:56:25 if I don't want to touch mozilla stuff there is a good reason Mar 09 13:56:32 I touched it in the past Mar 09 13:56:36 and it was very hard Mar 09 13:57:36 for the libx11 follow my advises Mar 09 14:00:29 GNUtoo|laptop: I did not see lgconf-2 error and I don't know how to follow your advise and yes it is a good idea to post it on the ML . if you are already at it you can refer to the libx11 bug I posted in bugzilla Mar 09 14:00:47 GNUtoo|laptop: thanks a lot for trying to help ! Mar 09 14:01:54 I think it's important to have all the programs constantly working that are well known among end users Mar 09 14:01:55 I'll look a bit for -lgconf-2 and post to the mailing list Mar 09 14:02:02 I never succedded to compile fennec Mar 09 14:02:03 it makes OE more attractive to new developers Mar 09 14:02:28 maybe we have some mozilla magicians in OE who are not aware of the problem Mar 09 14:02:39 iirc I also tried to compile firefox in the paste and failed Mar 09 14:02:52 once I succedded for firefox Mar 09 14:02:56 and I will need it Mar 09 14:03:09 for when I get a bug 2.0 Mar 09 14:03:18 because it can be used as desktop computer Mar 09 14:03:38 or maybe I could just use midori like on my eee701 Mar 09 14:03:45 it might be better to bump pr to a new beta "tag" revision before we solve these things Mar 09 14:04:28 yea fennec also has the desktop like advantages with less screen waste. I tried it in ubuntu on hd2 Mar 09 14:04:48 http://vimeo.com/14630263 Mar 09 14:08:18 dcordes_, I'll tell you asked me to bugreport Mar 09 14:10:50 03Andreas Oberritter  07master * r40d80ab44f 10openembedded.git/recipes/mrouted/mrouted_3.9.4.bb: Mar 09 14:10:50 mrouted-3.9.4: intitial recipe Mar 09 14:10:50 Signed-off-by: Andreas Oberritter Mar 09 14:10:50 Acked-by: Khem Raj Mar 09 14:10:50 Acked-by: Paul Menzel Mar 09 14:13:02 GNUtoo|laptop: don't forget to ask for ack of your autconf213 patch Mar 09 14:28:39 hmmm Mar 09 14:28:45 my patch is bogus Mar 09 14:28:52 + if [ -e ${S}/bin/*.in ] Mar 09 14:28:54 that can't work Mar 09 14:36:46 it's crazy how fast this stuff can eat away some days like it's few hours :) Mar 09 14:37:46 if [ "$(echo *)" = '*' ]; then echo empty; fi seem to produce empty on empty dir Mar 09 14:40:05 I'll ask when there are more people Mar 09 14:40:10 and I'll go back to work Mar 09 14:40:15 because it's kind of 15:38 Mar 09 14:40:18 here Mar 09 14:40:56 what you working on ? Mar 09 14:41:30 maths Mar 09 14:42:10 and I must also solve that slow internet connection problem Mar 09 14:43:13 yesterday it was : we solve that backlight problem and it's over, then today you need fennec that is really hard to fix Mar 09 14:43:38 why can't we keep eve? Mar 09 14:43:46 s/we/you Mar 09 14:43:59 yeah let's keep eve for now Mar 09 14:44:16 I documented that errors I came across that's all I'm capable of Mar 09 14:44:18 I'll see what they tell in the mailing list Mar 09 14:44:22 but it can be very long Mar 09 14:44:26 ok Mar 09 14:44:39 ok next time I will also report such things on the ML- problem is I asked here before if I should and nobody replied.- Mar 09 14:44:40 note that bugtracker is often ignored Mar 09 14:44:43 so I put it on BZ Mar 09 14:44:47 the best way to report is the mailing list Mar 09 14:44:52 ok Mar 09 14:45:00 because nobody was there Mar 09 14:45:05 wrong hour Mar 09 14:45:28 maths... I have to learn inorganic chemistry actually :) Mar 09 14:45:46 ok Mar 09 14:45:47 but I give it some time. I can work better when there is a certain degree of time preassure (lazy?) Mar 09 14:46:17 "I have a motivation problem until I have a time problem" Mar 09 14:47:23 ah ok Mar 09 14:47:34 then tell yourself that cotulla won't wait forever Mar 09 14:47:48 so maybe fix that backlight issue Mar 09 14:47:53 and ship the image to cotulla Mar 09 14:48:11 everything can't work out of the box on a new device Mar 09 14:48:24 s/new device/a not yet supported device/ Mar 09 14:53:06 good morning all Mar 09 14:53:43 Good Morning Mar 09 15:02:22 hi guys, are there any checklists for isolating bitbake erors? Mar 09 15:12:14 Soopaman, depend on the error but usually look at the log you have Mar 09 15:12:17 then isolate the error Mar 09 15:12:27 and then pastebin it here Mar 09 15:18:29 k Mar 09 15:19:58 http://pastebin.com/8Dfa9Fen Mar 09 15:20:28 bitbake is "properly" installed and referenced in the appropriate paths as per http://wiki.openembedded.net/index.php/Getting_started#Setup_the_environment Mar 09 15:29:17 ka6sox-away: ping Mar 09 15:29:52 Jay7, pong...GM Mar 09 15:31:23 ka6sox-away: I was away these days Mar 09 15:31:33 anything interesting? :) Mar 09 15:31:57 cascading failures on MySQL but other than that nothing. Mar 09 15:32:08 things are moving in a number of fronts. Mar 09 15:32:28 pb__, you should be setup and an email on the way to you. Mar 09 15:32:31 Soopaman: a variable is undefined that shouldn't be. most likely, distro or machine, which indicates a setup problem Mar 09 15:33:26 ka6sox-away: ok Mar 09 15:38:17 kergoth: oh ok. I'm kind of at a loss as to where to look after double checking my environment against the getting started page, and googling for oe issues on ubuntu 10.10 Mar 09 15:38:49 well, there's always bitbake -DDD, crank up the debugging to try to see more. Mar 09 15:39:24 k Mar 09 15:39:27 lemme try that Mar 09 15:39:45 Jay7, out all day today visiting clients but back tonight..I have more to add to the Yocto page on stats. Mar 09 15:40:15 also I should find out about setting up meeting with pidge after release. Mar 09 15:43:53 btw, I've thought today about QA runtime testing Mar 09 15:44:02 seems this is another entity as well.. Mar 09 15:44:47 kergoth: it seems to be failing on this one line in every package Mar 09 15:44:48 http://pastebin.com/8VbFB6cQ Mar 09 15:45:04 I'm not sure it is good idea to have it as bitbake plugin, like yocto does now Mar 09 15:45:06 gonna google any ubuntu 10.10+python issues Mar 09 15:50:13 Jay7, we need something to do both command as well as recording results/errors Mar 09 15:52:47 Soopaman1: then you are missing an item in your local.conf, I cannot remember which line. (distro or machine is INVALID) Mar 09 15:55:24 k Mar 09 15:55:27 lemme look Mar 09 16:00:10 likewise: is the most recent stable angstrom distro "angstrom-2010.x" ? Mar 09 16:02:56 funk, i think that may have done it, thanks likewise && kergoth Mar 09 16:03:09 Soopaman1: what was the fault? Mar 09 16:03:26 the distro had an unclosed quote :/ Mar 09 16:03:34 Soopaman1: cool. tnx Mar 09 16:09:42 fray, ping Mar 09 16:10:24 Tartarus: hey Mar 09 16:10:48 ant_work, hey Mar 09 16:12:04 I've seen you touched mtd-utils as last. Is the recipe for oe-core? Needs some care, e.g. BBCLASSEXTEND for native Mar 09 16:12:35 I've just been fixing up odds and ends Mar 09 16:13:08 I tried 1.4.2 and compiles with a couple of patches less Mar 09 16:13:31 ok Mar 09 16:13:33 still I have a runtime problem...maybe old-kernel related Mar 09 16:33:01 * Tartarus prepares some gettext in DEPENDS -> inherit gettext changes Mar 09 16:33:22 * Tartarus is leaving rpm alone since fray has fixed it in poky/oe-core Mar 09 16:33:54 * kergoth thinks its pretty sad any of those are left, kgiven how long gettext.bbclass has been around Mar 09 16:42:40 kergoth: many of the developers, including me, have no overview of the whole project meta data or infrastructure. Wish I had. Mar 09 16:43:03 yeah, but in this case, people should have been commenting on it as reviewing the code Mar 09 16:43:20 Tartarus: yes but the amount of people aware of it is too limited Mar 09 16:43:40 the amount of know-it-all reviewers (and their time) is too limited. Mar 09 16:43:56 time doesn't scale, the amount of people do, but it's a chicken egg problem Mar 09 16:44:40 That's why I RFC'd the idea to put more answers to "why?" in the commit msgs, so that developers can understand and learn from the made improvements. Mar 09 16:45:01 the current commit msg is "X has changed." Mar 09 16:45:16 and I always want to learn why X has changed. Mar 09 16:45:54 Something else; are we not filling oe-core with a selected set of recipes? Mar 09 16:45:58 not=now Mar 09 16:46:08 oe-core and meta-oe, yes Mar 09 16:46:53 where do the machine definitions go? we need to make seperate repo's for them? p2020 and friends for example Mar 09 16:47:19 would love to bring our powerpc set over to oe-core to test against Mar 09 16:47:29 oe-core will have qemu machines only Mar 09 16:47:35 And there will be layers for other hardware Mar 09 16:47:36 yes, I know Mar 09 16:47:54 but where will those layers live? Mar 09 16:48:20 Well, Koen has asked the board about hosting them on git.oe.org, for vendor involved ones Mar 09 16:48:31 And for others, somewhere on git.oe.org Mar 09 16:48:52 * Tartarus hopes to have time to setup something for Freescale hardware, once the board comes to an answer about Koen's question Mar 09 16:49:40 * likewise added the Freescale bits to OE once, same here :-) Mar 09 16:49:57 Tartarus, did you see my reply to denix with a timeline? Mar 09 16:50:16 Saw but not read Mar 09 16:50:19 * Crofton needs to run out the door and go skiing rsn :) Mar 09 16:50:20 * Tartarus is a tad swamped atm Mar 09 16:50:20 k Mar 09 16:50:28 * Crofton is supposed to be on vacation Mar 09 16:53:16 the two things I see we need to work out is access control and community bsp's Mar 09 16:53:28 I can see people wanting both Mar 09 16:54:29 Crofton: ping Mar 09 16:54:38 for the sake of simplicity I think it would be sane to remove the stale repositories in cgit (oetest, eclipsetools ?, meta-openembedded ??) and the stale branches in openembedded Mar 09 16:54:43 pong Mar 09 16:54:49 Crofton: I'm afraid what Koen makes people believe is not correct! Mar 09 16:55:27 heh Mar 09 16:55:38 I use a koen filter :) Mar 09 16:55:54 :) Mar 09 16:56:33 like I said, the end result is deliver a good user experience Mar 09 16:56:58 I think koen wants to make sure he can count on having a solution Mar 09 16:57:11 is worries about our ability to make decisions in a timely fashion Mar 09 16:57:20 that's all good Mar 09 16:57:27 Crofton: I think what we first need is a good overview of the new repo's and how they interact. Even I lost track of the oe-core, meta-oe, etc. layout. Mar 09 16:57:27 and the solution is really needed Mar 09 16:57:36 exactly Mar 09 16:57:46 it is critical to making sure people can get work done Mar 09 16:57:52 likewise: that's true :) Mar 09 16:57:57 without fishing for bits and pieces all over the interweb Mar 09 16:58:12 likewise, add that to the discussion :) Mar 09 16:58:16 ok, must go afk Mar 09 16:58:17 Crofton: git is distributed Mar 09 16:58:22 Crofton: all documentation floating around refers to the old situation. Mar 09 16:58:32 the new, is too new :p Mar 09 16:58:37 hence, you can host the same repo on many servers Mar 09 16:58:45 denix: since when? git.kernel.org was overloaded 4 hours ago. then what? :-) Mar 09 16:59:02 what my mgmt is worried, is to have control over own BSP layer Mar 09 16:59:06 denix: that's not distributed, that's cloned. Mar 09 16:59:34 likewise: https://wiki.yoctoproject.org/wiki/OpenEmbedded-Core (that's all about that for now I think) Mar 09 16:59:50 likewise: so what, don't you know 100 other places to d/l kernel from? :) Mar 09 17:00:24 denix: not the freescale powerpc tree at git.kernel.org :-) Mar 09 17:00:28 denix: right, so they should set up their own git repo (if they don't have one already) and host the BSP there. seems like the easiest answer for everyone really. Mar 09 17:00:42 pb__: correct Mar 09 17:00:56 if you/they also want to host a clone of oe-core on your own servers so that you have all the bits in one place, that's also fine of course. Mar 09 17:01:06 that's what we've been doing before Mar 09 17:01:43 also, couple years ago there was discussion to establish mirroring of out-of-tree layers on oe.org, unfortunately that never materialized... Mar 09 17:01:52 what is this: http://cgit.openembedded.org/cgit.cgi/meta-openembedded/ Mar 09 17:02:11 layer Mar 09 17:02:24 likewise: that's koen's attempt at layering oe on top of yocto... Mar 09 17:02:52 yes, but is this the "official" meta layer for the next OE release, or ... ? Mar 09 17:03:27 I mean, where to start now, as we agreed to build on top of oe-core. Mar 09 17:06:30 I'm not aware about any URL one could point you at Mar 09 17:06:39 denix: do you know where this pressure to host BSPs on oe.org is coming from? is it just something that koen wants to do for his own amusement, or is there actual demand for it from users and/or vendors? Mar 09 17:07:11 Anyone care about the apr recipe? It's using libtool from the host, not sysroot... Mar 09 17:07:43 It looks like an unusual recipe, so I could use a little guidance on how to make it work correctly :) Mar 09 17:07:46 denix: I see toolchain bits in meta-openembedded, why? Mar 09 17:08:11 pb__: why not something like community-bsp? is everyone supposed to have it's own git repo, there's hundred of machines in oe Mar 09 17:09:04 can you guys make sure you get your points on the members list? Mar 09 17:09:10 one machine should open the way and give the example Mar 09 17:09:13 It is hard for me to track irc his week :) Mar 09 17:09:21 pb__: in general for any BSPs - I'm not sure. for TI BSP, koen wants to own it and host it on oe.org, which is questionable... Mar 09 17:10:02 Crofton: your skiing can wait... :) Mar 09 17:10:09 tell my wife Mar 09 17:12:01 likewise: better ask on mailing list, many would benefit from this answers(if any :)) as it's going to be faq here soon :p Mar 09 17:12:20 ynezz: sorry, I don't think I understand what you're saying Mar 09 17:13:13 pb__: ok, I try to mantain the ts72xx machine in OE, what should I do now? Mar 09 17:13:35 create github repository for just this single machine? Mar 09 17:14:09 I've learned and has been influenced a lot from the work done by others Mar 09 17:14:27 this will be gone, since the bsp fragmentation Mar 09 17:14:33 or am I mistaken? Mar 09 17:14:37 ynezz, you're mistaken Mar 09 17:14:45 The next person will fork your repo probably :) Mar 09 17:14:58 hi Mar 09 17:15:02 :) Mar 09 17:15:34 Tartarus: ok, there will be some kind of upstream as it's today? Mar 09 17:15:51 guys, when ipkg says "Multiple replacers for xxx available, using first one (yyy)" - is it somehow possible to tell opkg to prefer a particular package? Mar 09 17:16:16 ynezz, that's partially why there's a desire to host many layers on oe.org Mar 09 17:16:42 And I suppose it's also possible that meta-oe will just allow machines in the layer Mar 09 17:16:49 but oe-core is going to just be qemu machines Mar 09 17:16:55 Jin^eLD: PREFERRED_PROVIDER Mar 09 17:17:03 and vendors (and their sw partners) may want to try and support things more directly Mar 09 17:17:09 ok, good to know Mar 09 17:17:22 mwester-laptop: isn't PREFERRED_PROVIDER only influencing the stuff at build time? Mar 09 17:17:27 but hard to understand from the information available up to this days Mar 09 17:17:45 Jin^eLD: ah, yes that Mar 09 17:17:48 is build only Mar 09 17:18:13 mwester-laptop: so.. is there anything that would help during runtime? Mar 09 17:18:22 my problem is that for some reason opkg thinks it should pick a -dbg package Mar 09 17:18:29 which of course wrecks havoc Mar 09 17:18:32 That's probably a bug. Mar 09 17:18:52 I think with normal packages, the user has to decide which of the alternates to pick. Mar 09 17:19:45 sounds like two bugs, in fact. Mar 09 17:20:00 firstly, if you've got a -dbg package declaring Replaces, that's almost certainly a rogue RREPLACES in the metadata. Mar 09 17:20:01 well, its happening during an automatic update... problem is that the library was not doing .so.1 links properly, and now that it does do so the mangling/package name has changed Mar 09 17:20:16 secondly, it does seem wrong that opkg just makes a unilateral decision which package to install Mar 09 17:21:06 but, luckily, the first bug is probably easy to fix and I suspect that will make the second one a non-issue. Mar 09 17:21:45 well I am indeed messing with RREPLACES right now Mar 09 17:22:24 ah, well then Mar 09 17:22:59 ^_^ Mar 09 17:23:02 problem is that same .bb produces different lib*.ipk files Mar 09 17:23:09 are you getting any qa warnings, and if so are you paying attention to them? Mar 09 17:23:36 I am picking attention but indeed let me look closer if I maybe missed something Mar 09 17:23:41 aem, paying attention that is :) Mar 09 17:27:32 hmm nope, no QA issues reported Mar 09 17:27:41 can you put your recipe in a pastebin? Mar 09 17:28:00 specifically the RREPLACES bits, whatever you're doing with those Mar 09 17:28:26 sure one moment Mar 09 17:32:11 ok this is the old package: http://gitorious.digitalstrom.org/dss-oe/dss-oe/blobs/master/targets/common/recipes/ds485-core/ds485-core_0.9.0.bb the package did not generate .so.1 symlinks, so ipk library name became ds485-core Mar 09 17:32:18 now new version does generate those symlinks Mar 09 17:32:40 so ipk becomes libds485-core1 Mar 09 17:32:52 and this is what I tried with RREPLACES http://pastebin.mozilla.org/1140213 Mar 09 17:33:39 what's that python at the end about? Mar 09 17:34:36 that is quite an ingenious way of evading recipe_sanity :-} Mar 09 17:34:57 pb__: problem is that RREPLACES gets automatically changed to libds485... Mar 09 17:35:08 and thus is not replacing ds485-core Mar 09 17:35:49 basically if I omit the python part the RREPLACES thing loses effect completely, probably because I am trying to replace "an older versin of myself" Mar 09 17:36:00 oh, crumbs. I guess that is a bug in package.bbclass. Mar 09 17:36:25 I'm not convinced that package_mapping_rename_hook() ought to be touching RREPLACES at all really. Mar 09 17:36:44 I have to add that I am 6 months behind of oe.dev :P I'll compare the .bbclass Mar 09 17:37:05 try overwriting that function (not appending to it) with a version that doesn't map RREPLACES and see if the situation improves. Mar 09 17:37:05 can't upgrade now for various reasons Mar 09 17:37:13 that does sound funky Mar 09 17:37:23 you probably also want to stop it mapping RCONFLICTS and RPROVIDES Mar 09 17:37:59 hi kergoth Mar 09 17:38:06 * pb__ go home now Mar 09 17:38:07 later all Mar 09 17:38:09 hey Mar 09 17:38:30 ok thanks for the hints, I'll try that Mar 09 18:22:06 goodness bitbake, why must you always be so difficult Mar 09 19:01:07 re Mar 09 19:04:17 ok I tried to replace the mapping rename hook instead of appending to it Mar 09 19:04:18 http://pastebin.mozilla.org/1140516 Mar 09 19:04:30 I am not sure why it does not work Mar 09 19:05:14 in the opkg list the package does list the repaces and provides correctly Mar 09 19:05:15 i.e. Mar 09 19:05:15 Replaces: ds485-core Mar 09 19:05:16 Conflicts: ds485-core Mar 09 19:05:39 I wonder what else I am missing Mar 09 19:12:36 hi all, is there any command to know what all packages will be built for a given recipe , something which gives all the dependencies of the package Mar 09 19:13:29 sreddy: I think you can use the -g option which will produce a dependency graph in .dot format Mar 09 19:13:58 it's not too convenient to read but usually you can figure out what pulls in what if that's what you want to know Mar 09 19:14:58 Jin^eLD : thanks Mar 09 19:21:36 03Tom Rini  07master * r1adf798619 10openembedded.git/recipes/classpath/ (classpath-initial_0.93.bb classpath-native.inc): Mar 09 19:21:36 classpath: Update to use 'inherit gettext' Mar 09 19:21:36 Signed-off-by: Tom Rini Mar 09 19:21:36 03Tom Rini  07master * rbaab204d07 10openembedded.git/recipes/gnome/gnome-doc-utils.inc: Mar 09 19:21:36 gnome-doc-utils: Update to use 'inherit gettext' Mar 09 19:21:37 This means we don't need a separate DEPENDS for virtclass-native Mar 09 19:21:38 (these are updated correctly). Mar 09 19:21:38 Signed-off-by: Tom Rini Mar 09 19:21:39 03Tom Rini  07master * r0f61c69281 10openembedded.git/recipes/glib-2.0/ (6 files): Mar 09 19:21:39 glib-2.0: Update to use 'inherit gettext' Mar 09 19:21:40 This means never spelling out gettext-native, virtual/libiconv and Mar 09 19:21:40 virtual/libintl. We only need to add pkgconfig-native to DEPENDS for Mar 09 19:21:42 virtclass-native. Finally, move EXTRA_OECONF_virtclass-native to be with Mar 09 19:21:42 the rest of the EXTRA_OECONF's. Mar 09 19:21:43 Signed-off-by: Tom Rini Mar 09 19:21:43 03Tom Rini  07master * r7f4be279d8 10openembedded.git/recipes/e2fsprogs/e2fsprogs.inc: Mar 09 19:21:44 e2fsprogs: Update to use 'inherit gettext' Mar 09 19:21:44 While in here, we do not need separate entries for virtclass-native. Mar 09 19:22:36 Signed-off-by: Tom Rini Mar 09 19:31:30 is oe@osx a nice experience now? been some months since I tried. Mar 09 19:33:04 building OE on OSX? No, still not functional iirc Mar 09 19:34:53 I'll have to install Ubuntu on my Mac soon then :) Mar 09 19:52:51 hmmm interesting http://www.thedailytech.co/2011/03/09/all-hp-computers-shipped-in-2012-to-include-webos/ Mar 09 19:53:18 so HP is all wedded to WebOS then they should probably fuel OE a bit shouldn't they ? Mar 09 19:53:47 oh may be they even dont know that WebOS is using OpenEmbedded Technology Mar 09 19:54:16 its like android hiding linux kernel beneath Mar 09 19:54:57 :) Mar 09 19:57:16 khem: very possible that top-management just doesn't know about OE behind WebOS Mar 09 19:57:35 someone should contact them :) Mar 09 19:58:03 that said I was encouraged to see OE logo on webos-internals booth at SCaLE Mar 09 19:58:21 I can but I dont have contacts Mar 09 19:58:35 ah, yes, I've read your report :) Mar 09 19:58:38 may be Rich Pixley Mar 09 20:04:19 ah now I see why my machine froze every hour for 2 mins Mar 09 20:04:37 there is this stupid cron which cd into / and does a find Mar 09 20:04:47 and I have autofs Mar 09 20:04:56 which mounts servers from all over the world Mar 09 20:05:05 still 2 mins was impressive :) Mar 09 20:05:14 * khem kills the hourly cron job forever Mar 09 20:05:58 Mr. khem -- if a recipe needs libtool, should it get it from the host, or from sysroots? Mar 09 20:06:23 mwester-laptop: it should get it from host-sysroot Mar 09 20:07:44 Ok. So then apr is broken. But the problem is that I have no idea how to fix it. Building libtool-native doesn't put libtool in sysroots; it puts -libtool in sysroots. Mar 09 20:08:21 yes Mar 09 20:08:33 thats the one you need Mar 09 20:08:52 actually if you use inherit autotools it should do everything right Mar 09 20:09:04 * mwester-laptop checks Mar 09 20:09:10 but if recipe mucks with do_configure then all bets are off Mar 09 20:09:48 it does the latter. do_configure_prepend Mar 09 20:09:53 And it all breaks. Mar 09 20:10:26 I guess nobody has caught it because every builder except the new SlugOS autobuilder has libtool installed. Mar 09 20:14:54 Well Mar 09 20:15:10 you can often pass LIBTOOL=${HOST_SYS}-libtool or so in EXTRA_OEMAKE Mar 09 20:15:11 welcome to myriad of build hosts Mar 09 20:15:48 we will bring this myriad to pacific Yocto place ;) Mar 09 20:15:55 mwester-laptop: btw your state abolished death penalty :) new world heh Mar 09 20:16:08 so now there are 16 states Mar 09 20:16:30 This is incredibly tricky :) How to keep host dependencies from leaking into recipes... Mar 09 20:16:52 mwester-laptop: hard problem Mar 09 20:16:58 yocto's swobber or whatever should be of use for that sort of thing Mar 09 20:17:06 either we trust or we do not trust host provided packages Mar 09 20:17:13 LD_PRELOAD monitoring of host file access, afaik Mar 09 20:17:44 we need something like a file monitor which track all open/read/write requests Mar 09 20:17:56 that's what i'm talking about. yocto has it Mar 09 20:17:56 kehm: Yeah, well, my state had to abolish the death penalty. It's the most corrupt state in the union, according to the FBI, and has multiple governors in prison (!), so they did it out of a sense of self-preservation. Only a matter of time until some state politician is found guilty of a capital offense. Mar 09 20:17:57 that would be incredibly slow but accurate Mar 09 20:18:52 ah, right, it uses strace, not preload Mar 09 20:18:53 http://git.yoctoproject.org/cgit/cgit.cgi/swabber/tree/README Mar 09 20:18:58 same idea though Mar 09 20:19:04 cool Mar 09 20:19:13 we can even make use of that to generate dependencies Mar 09 20:19:20 they will be accurate Mar 09 20:19:37 and we can find all kind of crap during compilation Mar 09 20:19:49 Or just use IBM/Rational's ClearMake. Mar 09 20:19:58 $$ Mar 09 20:20:06 x10 Mar 09 20:20:16 clearmake is actually pretty interesting, conceptually Mar 09 20:20:25 even if storing outputted artifacts in an scm makes me cringe Mar 09 20:20:28 it is Mar 09 20:20:35 * khem evaluated it last year Mar 09 20:20:53 The whole idea of an auditing fs is kind of cool, but the issue of where/how to manage what is discovered becomes the problem then. Mar 09 20:21:11 * kergoth nods Mar 09 20:22:49 generate a makefile dependencies file Mar 09 20:23:24 actually we have developed an internal tool here which generate a .meta file in a given syntax Mar 09 20:23:29 in xml Mar 09 20:23:43 and it is can be consumed by many tools Mar 09 20:24:08 people are using it to find all kind of crap that is pulled in and managers like GUI Mar 09 20:24:12 so its selling :) Mar 09 20:24:23 SIGFOOD Mar 09 20:24:28 bbl Mar 09 20:56:57 hiho Mar 09 20:57:48 hi woglinde Mar 09 21:45:37 hi woglinde Mar 09 21:46:06 he pb Mar 09 21:46:13 will go to sleep soone Mar 09 21:46:29 hi, any1 here caring to help me with something ? Mar 09 21:47:32 marenas: just a general thing, or one particular thing? Mar 09 21:47:54 a particular thing about linking programs and libs Mar 09 21:48:04 just ask Mar 09 21:48:10 hi likewise Mar 09 21:48:47 hi woglinde, you are always fast in welcoming me :-) Mar 09 21:49:00 i'm trying to build the program mcron for an openembedded image im making and the image is created ok, but when i execute the program i have an error Mar 09 21:49:01 always an eye at the screen, right? :-) Mar 09 21:49:21 it says 'ERROR: In procedure dynamic-link:' and can't find a lib that is supposed to be there Mar 09 21:49:44 likewise sometimes Mar 09 21:50:34 the program is trying to find '/usr/lib/libguile-srfi-srfi-1-v-3.so' but the file isn't there, only '/usr/lib/libguile-srfi-srfi-1-v-3.so.3' Mar 09 21:51:49 marenas dev version is missing Mar 09 21:52:32 so i should include de dev version of the lib when building the image ? Mar 09 21:53:26 if you want to compile somethink on the target? Mar 09 21:53:28 that'd be ok as a workaround but clearly it sucks as a solution Mar 09 21:53:41 but this looks like linker commando error Mar 09 21:53:50 in the longer term I think you need to fix guile to do the right thing Mar 09 21:54:27 it is a bit silly for its extension modules to have sonames if it isn't going to use them. Mar 09 21:55:11 pb__: something's not working out at all, I even tried renaming the new package but still RREPLACES/RCONFLICTS is ignored, which is strange because I have the same stuff working with a number of other packages Mar 09 21:55:42 the program works if I create a symlink for the so.3 file to the so file in the device, but i need it to be in the image from the start Mar 09 21:57:25 so the link must be there at config and compile time... but i doesn't get copied or linked correctly Mar 09 21:57:36 *it Mar 09 22:00:45 i try to creat a symlink in the recipe file but it doesn't appear in the final open embedded image :( Mar 09 22:23:47 docbook-sgml-dtd-native fails to build; the sed step fails... Mar 09 22:24:26 which one? Mar 09 22:28:20 4.1? Mar 09 22:28:31 docbook-sgml-dtd-3.1-native fails. sed: can't read docbook.cat: No such file or directory Mar 09 22:28:51 my 2nd guess, heh Mar 09 22:28:57 my fault, fixing Mar 09 22:29:49 Hmm. no idea why that old one is being chosen, though. Not pinning any docbook versions... Mar 09 22:29:57 Well, it's a funny recipe Mar 09 22:30:11 it's docbook-sgml-dtd-VER-native_version Mar 09 22:30:19 so that recipes can say they need dtd VER Mar 09 22:30:29 Ah - so the VER (capital) is part of teh package name. Mar 09 22:30:36 yeap Mar 09 22:31:29 03Tom Rini  07master * r61e9a8f059 10openembedded.git/recipes/docbook-sgml-dtd/ (2 files): Mar 09 22:31:29 docbook-sgml-dtd-{3,4}.1-native: Fixup SRC_URI Mar 09 22:31:29 Missed adding subdir param to these versions previously. Mar 09 22:31:29 Signed-off-by: Tom Rini Mar 09 22:33:23 BTW, is the ML rather quiet this week (or is my mail server being problematic)? I'm not seeing the usual number of OE emails... Mar 09 22:34:14 btw, do we still needed docbook utils on host machine? Mar 09 22:34:27 oh my yest Mar 09 22:34:29 *yes Mar 09 22:34:31 imho we need more Mar 09 22:34:38 evince doesn't build for me due to missing a ton of stuff Mar 09 22:34:42 but it's not trivial to get just right Mar 09 22:34:46 :\ Mar 09 22:34:47 * Tartarus started on it, got side tracked Mar 09 22:35:00 docbook have heavy dependencies Mar 09 22:35:18 iirc, some kind of TeX Mar 09 22:35:38 Maybe I was just doing it wrong but it seemed like we had most things, I just couldn't convince the tools to not use the system stuff Mar 09 22:39:39 I wonder if anyone tried to build X-less image via OE :\ Mar 09 22:40:07 bluez pulls gstreamer, gstreamer pulls gtk, pygtk, glade, all X-stuff, etc Mar 09 22:40:14 gnutls pulls gtk-doc for some reason Mar 09 22:40:27 dbus pulls libx11 Mar 09 22:40:28 anarsoul: but image itself contains no X :) Mar 09 22:40:43 oh, really? Mar 09 22:40:51 anarsoul: SlugOS is x-less. Mar 09 22:40:55 but WHY it pulls all this shit? Mar 09 22:41:02 heh! Mar 09 22:41:08 anarsoul: to build that shit :) Mar 09 22:41:23 dbus DOES NOT NEED libx11 to build! Mar 09 22:41:56 If you feed it the right flags, it doesn't. Mar 09 22:42:17 mwester-laptop: btw, how is your attempt to get rid of all that shit? :) Mar 09 22:42:21 gnutls does not need gtk-doc to build Mar 09 22:42:46 and bluez does not need gstreamer to build Mar 09 22:43:06 anyway, why bluez is mandatory in console-image? Mar 09 22:43:07 I think something snuck in recently that made it a lot worse, anarsoul -- it used to be that dbus didn't build x11, but if your machine had bluetooth support, it pulled it in. Now something is pulling in gtk, which sucks in X11, and the universe. Mar 09 22:43:38 mwester-laptop: situation was same 1.5year ago Mar 09 22:44:02 Jay7: my attempt is going well, with older snapshots of OE. :) I have to figure out what has happened that is pulling in all sorts of gtk stuff. Mar 09 22:44:27 is there some automated way of generating a list of package dependencies for a recipe with license info and source version info? seems like you could get bitbake to create a log if it doesn't already? Mar 09 22:44:51 bitbake -g package Mar 09 22:45:09 thats deps... but doesn't show license and source fetch info right? Mar 09 22:46:07 it doesn't Mar 09 22:46:28 * anarsoul 's disappointed by OE once again :( Mar 09 22:46:44 seems like it would be a common thing - management asking to generate a list of GPL software that is used in product xyz ;P Mar 09 22:47:12 tharvey: we've had a class for outputting license info about sources for like 8 yeras Mar 09 22:47:24 i think poky has more detailed manifest stuff though. Mar 09 22:47:35 which means we will when we move to oe-core Mar 09 22:47:37 anarsoul: there was looong discussion about that deps Mar 09 22:47:39 Does it output licence info for everything in DEPENDS or only what makes it into the image? Mar 09 22:47:43 poky also does license checksumming Mar 09 22:47:55 what class and how to use it? Mar 09 22:47:56 anarsoul: mostly DISTRO vs MACHINE features Mar 09 22:48:05 assume inherit something and log is built? Mar 09 22:48:45 tharvey: src_distribute_local, it's default in most distros. it's not a log, it's a directory of the sources organized by license. but it's not separated by 'this recipe and its dependencies only' Mar 09 22:52:44 ah... thats nice Mar 09 22:53:06 perhaps I could add to that class a single log file as well Mar 09 22:53:09 kergoth, thanks! Mar 09 22:57:34 Evening. Mar 09 23:00:01 What's the status of nativedk? I see sdk is deprecated, but e.g. meta-toolchain (vis task-sdk-host) uses sdk Mar 09 23:00:24 should this one day end up with nativesdk, or do I misunderstand sdk vs nativesdk? Mar 09 23:05:30 kinda sorta Mar 09 23:05:35 It's a bit better in oe-core Mar 09 23:05:40 but I'm not sure if it's 100% there yet Mar 09 23:05:55 (and I need to get some time and get to know that stuff better up in oe-core) Mar 09 23:07:55 I have not been following OE recently... I only heard oe-core mentioned Mar 09 23:08:13 is it some "cool" subset of OE recipes and/or some poky hybrid? Mar 09 23:11:45 both, heh Mar 09 23:12:14 wunderbar :) could you give me some pointer? I guess I could google harder, but... Mar 09 23:12:25 glib-2.0-native failing on current tip: http://paste.debian.net/110206/ Mar 09 23:13:02 otavio, git bisect time Mar 09 23:13:22 * Tartarus did test his stuff w/ both minimal-uclibc and angstrom-2010.x Mar 09 23:14:17 Tartarus: didn't follow you Mar 09 23:14:49 i changed glib-2.0 stuff a tiny bit recently, but tested it Mar 09 23:15:02 So... time to pop open git bisect and see where the failure came from Mar 09 23:16:14 Tartarus: I guess it is due 2.28 new version. Mar 09 23:16:20 Tartarus: it is r1 so quite new Mar 09 23:19:03 Tartarus: 2.26.1 also fails with same error Mar 09 23:19:09 | tests/Makefile.am:133: variable `date_test_LDADD' is defined but no progr Mar 09 23:19:15 yeah, so it's probably something else that changed Mar 09 23:19:18 when did you last build fine? Mar 09 23:19:18 * otavio is out now Mar 09 23:19:32 wow just checking the logs Mar 09 23:19:35 hold Mar 09 23:19:59 Tartarus: 0b3127a0d308547bebf84ed44678d0a1c75e6d23 was fine Mar 09 23:20:07 great Mar 09 23:20:07 Tartarus: it was my last rebuild from scratch Mar 09 23:20:11 so do a git bisect :) Mar 09 23:27:30 Tartarus: about compressors, we have an insane -c -9 in IMAGE_CMD_cpio.lzma (bitbake.conf) Mar 09 23:27:40 this needs much more ram than 32Mb Mar 09 23:27:56 I thought it only impacted the compressor not decompressor Mar 09 23:28:03 thus, no boot on devices with 32M Mar 09 23:28:13 http://paste.debian.net/110203/ Mar 09 23:28:15 ^^ Mar 09 23:28:51 I'm testing -7 and seems ok Mar 09 23:46:19 Tartarus: imho we should introduce something like LZMA_COMPRESSION_LEVEL Mar 09 23:46:44 or use internal tables to find better level from targets memory amount Mar 09 23:50:21 Jay7, perhaps, yeah Mar 09 23:50:29 03Tom Rini  07master * r5d121d8828 10openembedded.git/recipes/ (28 files in 13 dirs): Mar 09 23:50:29 Many places: Refer to ${HOST_SYS}-libtool not ${TARGET_PREFIX}libtool Mar 09 23:50:29 Given that the libtool recipes install as ${HOST_SYS}-libtool Mar 09 23:50:29 we need to reference it as such and not ${TARGET_PREFIX}libtool Mar 09 23:50:29 as we will otherwise break when those evaluate differently. Mar 09 23:50:29 Signed-off-by: Tom Rini Mar 09 23:50:32 maybe it's xz where it's just host side that it impacts? Mar 09 23:50:40 Or maybe I'm crazy Mar 09 23:50:43 it has been one of those days :) Mar 09 23:51:18 * Tartarus doesn't want to think how much time he's wasted before realizing the above pushed change was his problem today Mar 09 23:53:07 it always affect decompressor Mar 09 23:53:35 but I'm wondering why decompressing about 1Mb will eat all possible kernel RAM Mar 09 23:53:35 yeah, ok Mar 09 23:53:41 i was thinking about xz's -e Mar 09 23:53:46 which claims to not Mar 09 23:54:28 -e is a way to make compression better at same level Mar 09 23:54:40 yeah Mar 09 23:54:47 So anyhow, yes Mar 09 23:54:50 we shouldn't use -9 Mar 09 23:54:58 we should use -7 for lzma and -6 for xz Mar 09 23:54:59 easy way is lowering to -7 Mar 09 23:55:00 as those are the defaults Mar 09 23:55:04 yep Mar 09 23:55:14 i'll ack if someone else does it :) Mar 09 23:55:22 * Tartarus has already pushed far too many things sans posting today, heh Mar 09 23:58:17 ant__: push fix with removed -9 then :) Mar 09 23:59:05 I'll write on ML, it's touching bitbake.conf Mar 09 23:59:05 or with LZMA_COMPRESSION_LEVEL Mar 09 23:59:42 * Jay7 is still experimenting with Open-E DSS.. Mar 10 00:38:01 * Jay7 -> sleep Mar 10 00:48:07 nn Jay7 Mar 10 01:21:09 Tartarus: quite easy. It has been break by your commit about gettext class Mar 10 01:25:50 otavio, heh, but I tested that Mar 10 01:25:53 What are you running on? Mar 10 01:28:20 Tartarus: probably not in a clean tmp Mar 10 01:28:27 shouldn't matter Mar 10 01:28:29 Tartarus: so you probably had the depends built Mar 10 01:28:33 Tartarus: it does Mar 10 01:29:08 it shouldn't tho Mar 10 01:29:25 Tartarus: well. I reverted it and it works Mar 10 01:29:28 What is the distro/machine combo Mar 10 01:29:38 Tartarus: i586-generic/minimal Mar 10 01:30:02 otavio: what errors do u get Mar 10 01:30:25 khem: I pasted it earlier. Mar 10 01:30:29 khem: let me find it out Mar 10 01:30:32 http://paste.debian.net/110206/ Mar 10 01:30:47 * Tartarus waits for cache parse Mar 10 01:30:54 khem: http://paste.debian.net/110206/ Mar 10 01:32:21 hummm Mar 10 01:34:06 that conveys nothing Mar 10 01:34:09 paste the full log Mar 10 01:34:19 glib-2.0 is special Mar 10 01:34:22 though Mar 10 01:35:00 i'm missing something Mar 10 01:35:44 Tartarus: glibc-2.0 always needs gettext Mar 10 01:36:50 Er Mar 10 01:36:57 What's ripping it out here Mar 10 01:37:14 glib-2.0 + USE_NLS=no fails, I know that Mar 10 01:37:19 But that's not what's tripping it up Mar 10 01:37:24 ORIG_DEPENDS has it Mar 10 01:37:27 then it's dropped out Mar 10 01:38:50 I am still waiting to have a look at complete log Mar 10 01:38:55 Don't need it Mar 10 01:39:01 gettext-native is missing from DEPENDS Mar 10 01:40:45 Tartarus: ok is it inheriting gettext ? Mar 10 01:40:49 yes Mar 10 01:41:02 does it have virtual/libintl Mar 10 01:41:04 DEPENDS_GETTEXT="gettext-native" Mar 10 01:41:05 in DEPENDS Mar 10 01:42:00 it's a native recipe Mar 10 01:42:38 hmm Mar 10 01:42:42 let me look at src Mar 10 01:44:11 hmm Mar 10 01:44:33 yeah, ok, thinko on my part Mar 10 01:44:35 checking for similar Mar 10 01:44:43 inherit autotools gettext Mar 10 01:48:23 DEPENDS_virtclass-native is way too fragile Mar 10 01:52:25 hum Mar 10 01:53:11 * mwester-laptop begins humming, as instructed. Mar 10 02:13:32 kergoth_, does virtclass-native += "foo" ever work like someone might expect or do we always get caught in evaluation order un-fun and really want to have used prepend/append Mar 10 02:13:33 ? Mar 10 02:26:28 03Tom Rini  07master * r4db2d44858 10openembedded.git/recipes/glib-2.0/ (6 files): Mar 10 02:26:28 glib-2.0: Fix DEPENDS_virtclass-native Mar 10 02:26:28 Oops, meant to use _append here not +=. Mar 10 02:26:28 Signed-off-by: Tom Rini Mar 10 02:36:19 is there a place that documents what the branches in git are for? Mar 10 02:36:34 i'm guessing i should go with origin/org.openembedded.stable Mar 10 02:36:40 to get started with oe **** ENDING LOGGING AT Thu Mar 10 02:59:57 2011