**** BEGIN LOGGING AT Fri Jun 06 02:59:56 2008 Jun 06 04:39:50 latest pull of org.openembedded.dev seems to be borked Jun 06 04:59:27 * mwester waits for T0mW to complete his sentence Jun 06 05:02:03 mwester: in what way Jun 06 05:02:05 ? Jun 06 05:02:14 it is unusable. Jun 06 05:02:31 an older copy of OE.mtn works fine Jun 06 05:03:13 mwester: http://rafb.net/p/NozlYE68.html Jun 06 05:04:30 I can keep working, no worries, but someone else will prolly have "issues" with the latest copy. Jun 06 05:06:33 * mwester parses "prolly", rejects the spelling corrections offered, and settles for a strange colloquialism meaning "probably" Jun 06 05:08:01 What's your monotone version? (Did you perhaps just pull down some updates?) Jun 06 05:08:13 I tried 0.38 and 0.40 Jun 06 05:08:29 0.38 should work. Jun 06 05:09:05 You might post your bitbake version as well, then. And a bug should be filed if one doesn't exist already. Jun 06 05:09:09 won't though. I tried the h3900 machine + angstrom-2008.1 distro (in local.conf) Jun 06 05:09:18 bitbake 1.8.10 Jun 06 05:10:49 it was a totally clean pull of everything. New OE.mtn, bitbake, etc. Jun 06 05:11:16 just thought that someone should know, that's all. Jun 06 05:11:27 Ok, nothing "bleeding edge" about that bitbake version particularly. So, yes, it would seem that something in OE itself now finds itself in need of fixing. If a bug hasn't been filed, it should be. Jun 06 05:12:45 from past experience, I do not update the OE database when working in an project. never do that becuase the breakage potential is quite high. Jun 06 05:13:23 Just for new projects, I will tryout the latest stuff. Jun 06 06:21:16 libiconv not happy with new libtool Jun 06 07:48:58 03deller 07org.oe.stable * r2fd34b90... 10/ (1 packages/tasks/task-base.bb): Jun 06 07:48:58 task-base: include more crypto modules to get WPA working on 2.6.24 kernels Jun 06 07:48:58 Acked by xjqian Jun 06 07:48:58 Acked-by: Koen Kooi Jun 06 07:50:06 * * OE Bug 4361 has been created by  Jun 06 07:50:08 * * libtool-cross-2.2.4-autobuild Jun 06 07:50:10 * * http://bugs.openembedded.net/show_bug.cgi?id=4361 Jun 06 08:52:20 03koen 07org.oe.stable * r3d0f8ccc... 10/ (1 packages/tasks/task-base.bb): task-base: bump PR Jun 06 09:46:10 morning Jun 06 09:46:25 hi hrw Jun 06 09:46:27 hm ah Jun 06 09:47:56 >8bit utf-8 can be fun Jun 06 09:48:53 hm Jun 06 09:49:00 now its clear Jun 06 09:49:17 almost every package should inherit lib_package Jun 06 09:56:31 hi, all! do anybody knows which license ncurses uses? Jun 06 09:57:01 is it called somehow or just 'ncurses license'? I mean version 5.4 Jun 06 09:57:30 looks like bsd/mit variant Jun 06 10:45:57 hi thesing Jun 06 10:46:20 hi thesing Jun 06 10:46:37 good morning thesing Jun 06 10:46:57 I'd like to try and build an initramfs image Jun 06 10:47:06 I am still unsure as to how to do that Jun 06 10:48:02 morning Jun 06 10:49:03 Laibsch: All you need to do is to set INITRAMFS_IMAGE to your favorite initramfs image somewhere. Jun 06 10:51:42 hi florian Jun 06 10:51:44 huhu Jun 06 10:51:52 hi jineld Jun 06 10:51:58 hi woglinde Jun 06 11:04:29 03koen 07org.oe.dev * rcd8e0bde... 10/ (4 files in 3 dirs): linux-omap2 git: update to 2.6.26-rc5 and update smartreflex patches Jun 06 11:12:08 thesing: "MACHINE=spitz INTRAMFS_IMAGE=initramfs-kexec-image bitbake console-image" should work? What do I need to take care of when flashing? Jun 06 11:15:42 Laibsch: I not into the environment variable stuff. But if its the same as setting it in local.conf it should work. Jun 06 11:16:11 but the kernel might be to big to flash. Jun 06 11:17:55 the resulting zImage can be flashed normally. It will then mount the flash and kexec a kernel from it. (zImage in / of flash) Jun 06 11:20:48 hm Jun 06 11:21:07 what's is the preferred way to override a PREFERRED_VERSION_..? Jun 06 11:21:29 if I put it in local.conf, the conf/include/-file still 'wins' Jun 06 11:35:07 then conf/include should set it with ?= not with = Jun 06 11:35:57 hm, trouble is, that inc. is courtesy of Angstrom devs Jun 06 11:36:14 so I have to alter an 'upstream' file Jun 06 11:36:38 what I can't compile a working kernel with oe, but I already have one that works? can I use it with angstrom/familiar images? are there any restrictions? should I just copy the modules over rootfs? Jun 06 11:37:05 *What if Jun 06 11:37:36 oxo: what you need to change exactly? Jun 06 11:37:51 lantern_burns_: feel free to add it to your repo Jun 06 11:37:52 PREFERRED_VERSION_automake Jun 06 11:38:01 since 1.10 seem to break a lot of stuff here Jun 06 11:38:16 hrw: you mean... kinda package it? Jun 06 11:38:17 m4 recursion loops, failing Grub's, etc Jun 06 11:38:39 oxo: use older version of angstrom then Jun 06 11:39:00 oxo: it was announced on angstrom ML that it will be used Jun 06 11:39:10 lantern_burns_: s/repo/metadata Jun 06 11:39:26 * oxo has no troubles with automake 1.10, but some packages in the tree do Jun 06 11:39:30 that doesn't help :) Jun 06 11:40:39 oxo: fix them? Jun 06 11:40:41 so assuming I have a zImage / modules, how do I boot using them and a custom rootfs? What rootfs should I use? Jun 06 11:40:42 oxo: what things are breaking? Jun 06 11:40:54 well, grub and quagga Jun 06 11:41:32 oxo: we should fix them, are you compiling against .dev or .stable? Jun 06 11:41:37 dev Jun 06 11:41:40 !oebug 4276 Jun 06 11:41:41 * * Bug 4276, Status: UNCONFIRMED, Created: 2008-05-19 00:48 Jun 06 11:41:41 * * : upgrade automake to 1.10 Jun 06 11:41:43 * * http://bugs.openembedded.net/show_bug.cgi?id=4276 Jun 06 11:41:52 oxo: record your problems in that bug Jun 06 11:41:53 grub manifests this error: http://www.mail-archive.com/grub-devel@gnu.org/msg03609.html Jun 06 11:41:56 * oxo will Jun 06 11:42:05 create new bugs if necessary and block 4276 Jun 06 11:44:49 thesing: kexec-tools-static fails do_compile: http://tinderbox.openembedded.net/public/logs/188559.txt Jun 06 11:44:53 do you have an idea? Jun 06 11:48:29 not really. But it seems that you use an eabi compiler for collie. Jun 06 11:49:30 Does this work only for collie? Jun 06 11:49:55 I was trying to build an image for spitz (see the bitbake line I used above) Jun 06 11:50:57 thesing: "MACHINE=spitz INTRAMFS_IMAGE=initramfs-kexec-image bitbake console-image" Jun 06 11:52:54 I didn't try on spitz. But it should work too. Jun 06 11:53:18 I'll try here. Jun 06 12:04:44 Started a collie build now Jun 06 12:04:55 morning Jun 06 12:05:02 That fails in do_patch with linux-rp 2.6.24 Jun 06 12:05:35 Laibsch: have a look at oestats: we are the only two builders of kexec-tools-static Jun 06 12:05:50 and I built yesterday night Jun 06 12:07:34 well, not everybody uses oestats Jun 06 12:07:39 and not for every build Jun 06 12:07:51 So you cannot generalize from oestats Jun 06 12:08:05 well, nobody cares for initramfs on zaurus (apart us) Jun 06 12:08:57 INTRAMFS_IMAGE=initramfs-kexec-image was default in linux-rp.inc until RP removed it last week Jun 06 12:09:10 thesing: any changes to locomokeyb-2.patch lately? Jun 06 12:09:17 It doesn't apply cleanly here Jun 06 12:09:17 (personally I'd set ?= and not =) Jun 06 12:09:43 I think RP removed it for good reason Jun 06 12:09:47 he Jun 06 12:09:53 It should not break stuff for others Jun 06 12:10:01 and we lost nothing Jun 06 12:10:02 BTW it breaks again on zaurus-updater Jun 06 12:10:30 this *really* needs a fix Jun 06 12:10:47 so, let's fix it Jun 06 12:11:00 I fear it's a multithread issue Jun 06 12:12:01 Laibsch: RP did some changes to it. Jun 06 12:12:45 I am sure it can be done in a way that satisfies everybody Jun 06 12:13:03 thesing: what about the do_patch step for collie in linux-rp Jun 06 12:13:09 ? Jun 06 12:15:12 it worked the last time I tried. Jun 06 12:15:31 which about 2 weeks ago. Jun 06 12:22:05 well, it fails now Jun 06 12:22:13 * Laibsch is inspecting changelog Jun 06 12:22:37 thesing: should we make 2.6.25+2.6.26-rc4 the default for collie? Would that make things easier for you? Jun 06 12:24:22 I don't think so. I should create a project at linuxtogo and publish my git tree there. Jun 06 12:25:22 http://gitweb.openembedded.net/?p=org.openembedded.dev.git;a=history;f=packages/linux has quite a few changes Jun 06 12:25:31 thesing: do I sense some frustration? Jun 06 12:26:28 http://gitweb.openembedded.net/?p=org.openembedded.dev.git;a=commitdiff;h=7450d856352098136734621687c63f67f24681a2 looks very interesting Jun 06 12:26:47 4 days old, update locomo keyboard patch Jun 06 12:27:59 no. I just don't have the time to to kernel work and maintain the kernel recipe in oe. Jun 06 12:28:44 thesing: publish your git tree and someone will make recipe for it? Jun 06 12:28:54 Laibsch: If I'm not wrong HvonTres pushed some changes for collie http://lists.linuxtogo.org/pipermail/openembedded-stablebranch/2008-May/000086.html Jun 06 12:29:21 (but I don't follow so much collie) Jun 06 12:29:34 hrw: yes I think this would be best. Jun 06 12:30:35 thesing: I guess the git server on amethyst would be a natural choice, branch off from org.oe.dev and maintain your patches in that branch Jun 06 12:30:39 How about that? Jun 06 12:31:33 I think its easier to create a new project at linux2go. I don't think that we want to mix kernel and oe. Jun 06 12:31:56 thesing: theere is linux-mobile on l2g Jun 06 12:32:18 Laibsch: let he maintain kernel git tree and someone other recipe for it Jun 06 12:34:37 hrw: yes but its probably better to maintain an own tree. Jun 06 12:35:28 either way is fine with me Jun 06 12:44:26 Hello Jun 06 12:46:42 Does someone knowns how I could tell BitBake cache about a package? I'd like to know since I'm having a trouble to add packages (locale ones) to my image due that. The code at package class splits the locale ones but doesn't tell the cache about them and I wouldn't like to put a PACKAGES_DYNAMIC on every .bb building locales. Jun 06 12:48:36 My current patch is http://pastebin.com/f1f467f78 Jun 06 12:49:07 otavio better write it to the devel-ml Jun 06 12:49:27 so more people read it Jun 06 12:49:38 and yes package.bbclass is a mess sometimes Jun 06 12:51:13 otavio: they must be something we're missing -- are we the only people trying to add locales to images? Jun 06 12:51:41 cbrake: at image building time, looks to be Jun 06 12:51:57 cbrake: and some packages can have the P_D on bb files Jun 06 12:52:02 cbrake: if you add it, it works Jun 06 12:52:15 cbrake: otherwise, opkg doesn't know it exists Jun 06 12:52:45 cbrake: I have solution probably for you Jun 06 12:53:04 hrw: so, it'll solve my problem too. It looks to be the same Jun 06 12:53:08 * cbrake eagerly awaits :-) Jun 06 12:53:20 * otavio crosses his fingers Jun 06 12:53:53 for Poky I wrote code to install all locale packages for installed packages Jun 06 12:54:10 hrw: :-D Jun 06 12:54:27 need to test it Jun 06 12:54:36 oe has more tester Jun 06 12:54:38 *g* Jun 06 12:55:03 hrw: does it do anything similar then my "patch"? Jun 06 12:55:06 and we dev and stable now Jun 06 12:55:07 woglinde: ugly solution which I first wrote is not proper quality Jun 06 12:55:14 otavio: I did not looked at your patc Jun 06 12:55:18 hrw: wow, so no one has ever installed locales before, or have things been changing Jun 06 12:55:44 cbrake: I think they moved the locales to the package class Jun 06 12:55:50 basically I do "for each_installed_package;do for eachlang do ipkg install package-locale-lang;done;done Jun 06 12:55:57 cbrake: this reduces the duplication but probable made this bug Jun 06 12:56:34 hrw: that sounds like the best solution Jun 06 12:56:36 hrw you didnt say something about the quality Jun 06 12:56:40 hrw: but how you tell opkg about the locale packages? That's where we fail Jun 06 12:57:19 guys - let me test it first ok? Jun 06 12:57:40 hrw: gladly :-) Jun 06 12:58:05 opkg is supposed to automatically install locales for packages installed if the base locale is installed Jun 06 12:58:36 XorA: base locale would be locale-base-* right? Jun 06 12:58:40 XorA: it does not Jun 06 12:59:02 otavio: it used to work when it was ipkg Jun 06 12:59:12 XorA: its broke now Jun 06 12:59:24 kill the maintainers :-D Jun 06 12:59:32 hehe Jun 06 12:59:43 omg ... I won't be to jail! hehehe Jun 06 13:00:06 XorA: was that code in ipkg or something in OE? Jun 06 13:00:40 in ipkg AFAIK, I just rember it working Jun 06 13:00:46 used to annoy me Jun 06 13:02:30 "find | xargs grep locale" returns nothing in both ipkg and opkg source Jun 06 13:02:38 hmm Jun 06 13:03:00 locale in OE suxx Jun 06 13:03:02 cbrake: take a look at rootfs_ipk Jun 06 13:03:09 hrw: we noticed Jun 06 13:03:23 cbrake: it has code to install glibc-localedata-i18n Jun 06 13:03:40 hrw: That is because we never lacked more basic problems that needed urgent fixing Jun 06 13:04:31 Laibsch: locale support is a serious issue from my POV. This makes it useless for people needing systems in their native language Jun 06 13:05:40 It surely is a serious issue Jun 06 13:06:17 But if you have your hands full, just making sure the stuff builds at all, it is most likely not the most serious one Jun 06 13:06:50 Are if you have grave bugs like your touchscreen moving left when you move up (90° twisted, etc.) Jun 06 13:07:00 OE had stuff like that all the time Jun 06 13:07:02 the problem is that my method is very slow Jun 06 13:07:11 cbrake: looks to be this code that is suppose to install locale-base for the LINGUAS_INSTALL Jun 06 13:07:27 otavio: that all works as far as I can tell Jun 06 13:07:36 otavio: but that is only the locale-base Jun 06 13:07:44 cbrake: yes, this works Jun 06 13:08:04 cbrake: but it does because glibc-package.bbclass has the P_D item Jun 06 13:08:07 cbrake: looks Jun 06 13:10:09 Downloading http://ewi546.ewi.utwente.nl/mirror/www.openzaurus.org/official/unstable/3.5.4.1/feed/upgrades/gpe-timesheet_0.31-r1_arm.ipk Jun 06 13:10:09 Adding satisfier for greedy dependence: gpe-timesheet-locale-fr Jun 06 13:10:15 thats what ipkg used to do Jun 06 13:10:37 What is greedy dependence? Jun 06 13:10:51 otavio: no idea, but thats how it used to summon in locales Jun 06 13:11:09 XorA: was that on the device or doing do_rootfs? Jun 06 13:11:16 s/doing/during/ Jun 06 13:11:22 Laibsch: on device, but it should work for both Jun 06 13:11:37 XorA: that you installed to get it? Jun 06 13:11:54 hrw: I had to google for old logs to find that message Jun 06 13:12:19 XorA: I and cbrake wants it at rootfs building time Jun 06 13:12:31 otavio: its the same ipkg/opkg used Jun 06 13:12:35 otavio: and the same feed Jun 06 13:12:45 otavio: so fix it for one, it will fix it for all Jun 06 13:13:01 we need to find out what changed, and what greedy means Jun 06 13:13:08 XorA: From what I researched ... it looks that ipkg doesn't "knows" about some locale packages Jun 06 13:13:20 XorA: opkg, sorry Jun 06 13:13:33 otavio: something obviously used to though, go back in time Jun 06 13:14:00 * otavio isn't that good on OE to fully understand it Jun 06 13:14:09 XorA: I still have an ipkg based system running for an older project -- I'll try it there ... Jun 06 13:14:15 * slapin agrees wit cbrake and want that feature too Jun 06 13:14:57 grep -rni "greedy" * Jun 06 13:14:57 libopkg/pkg_depends.c:108: if (compound_depend->type == GREEDY_DEPEND) { Jun 06 13:14:57 libopkg/pkg_depends.c:153: opkg_message(conf, OPKG_NOTICE, "Adding satisfier for greedy dependence: %s\n", pkg_scout->name); Jun 06 13:15:08 so opkg still has greediness Jun 06 13:15:20 I suspect ipkg-utils optimisations removed Greedy Jun 06 13:17:27 XorA: ok, I found it Jun 06 13:17:32 XorA: opkg is working Jun 06 13:17:39 XorA: I've just tested it Jun 06 13:17:51 The real bug is the package.bbclass spliting Jun 06 13:18:00 dunno how to solve it Jun 06 13:18:06 otavio: woo woo Jun 06 13:18:06 Let me explain Jun 06 13:18:45 package.bbclass does the locale spliting for us now Jun 06 13:19:09 but adding PACKAGES_DYNAMIC over there doesn't look to have efect ... Jun 06 13:19:15 * otavio goes to test it Jun 06 13:19:17 again Jun 06 13:21:26 XorA: otavio: still get about the same things with ipkg: http://pastebin.ca/1040368 Jun 06 13:21:57 cbrake: the bug isn't at opkg. It works Jun 06 13:22:03 cbrake: here it grabed the locales Jun 06 13:22:14 cbrake: if I add the PACKAGES_DYNAMIC on the .bb Jun 06 13:24:43 florian: IIRC you fixed the gpe-login English/C Locale thing...well, last time I tried a GPE Image it did'n save that info and asked for it on every login. Is it supposed to do so? Jun 06 13:28:07 * * OE Bug 1088 has been RESOLVED (FIXED) by heinold(AT)inf.fu-berlin.de Jun 06 13:28:09 * * Problem to build autoconf with ksh as /bin/sh Jun 06 13:28:10 * * http://bugs.openembedded.net/show_bug.cgi?id=1088 Jun 06 13:29:03 otavio: how does PACKGES_DYNAMIC change the Provides/Depends in the package control data? Jun 06 13:30:02 cbrake: it, in someway, tells bitbake about those dynamic packages. Looking at bb code it is used at cache Jun 06 13:30:07 * * OE Bug 804 has been RESOLVED (WORKSFORME) by Jun 06 13:30:09 * * glibc-package.bbclass contains wrong shell expression Jun 06 13:30:11 * * http://bugs.openembedded.net/show_bug.cgi?id=804 Jun 06 13:30:28 cbrake: however, I would like to avoid writing it on every package building locales Jun 06 13:30:36 otavio: but in the Packages file in your feed, you see something different Jun 06 13:30:53 otavio: because it is opkg, not bitbake that is building the image Jun 06 13:31:02 cbrake: that is the problem Jun 06 13:31:15 cbrake: if it's not tagged, it doesn't ends in Packages.filelist Jun 06 13:31:28 cbrake: and then it doesn't end on Packages index Jun 06 13:31:36 cbrake: and then opkg doesn't know about them Jun 06 13:32:42 otavio: what exatly does not end up in the Packages index? I see Package: glibc-locale-de in my Package index Jun 06 13:33:09 cbrake: let me do another test Jun 06 13:33:43 otavio: that may give us some clues -- diff Package index before and after your PACKAGES_DYNAMIC changes Jun 06 13:34:07 * * OE Bug 4362 has been created by udovdh(AT)xs4all.nl Jun 06 13:34:09 * * flash_eraseall gives segfault Jun 06 13:34:11 * * http://bugs.openembedded.net/show_bug.cgi?id=4362 Jun 06 13:34:56 is a mtd-utils 1.1.0 I built myself. Jun 06 13:36:42 ant|work: No it doesn't save this info iirc. But you are right... that would be a useful enhancement. Jun 06 13:39:54 udovdh: Attaching long text such as logs keeps a bug more readable Jun 06 13:40:20 wrapping text not too soon helps as well. Jun 06 13:40:27 but I cna make an attachment if you like! Jun 06 13:42:06 Laibsch, done! Jun 06 13:42:19 cbrake: it's indexed, indeed Jun 06 13:42:41 Laibsch, see http://bugs.openembedded.net/attachment.cgi?id=9763 Jun 06 13:42:56 cbrake: the strange thing is that I didn't found where it installs the packages (others then the base) Jun 06 13:43:03 udovdh: thanks, that won't help too much now Jun 06 13:43:06 cbrake: it does, but I don't know where Jun 06 13:43:17 what could I do to help Laibsch ? Jun 06 13:43:27 udovdh: nothing Jun 06 13:43:42 I was just pointing out that long text in comments is harder to read Jun 06 13:43:58 otavio: another odd thing -- I can install glibc-locale-de using opkg on the device, but not at rootfs time -- sending mail to list ... Jun 06 13:44:12 udovdh: an attachment can be quickly skipped over when appropriate Jun 06 13:44:18 gdb perhaps Laibsch ? Jun 06 13:44:33 cbrake: yes, I said ... Jun 06 13:44:41 cbrake: it looks to be a bb cache issue Jun 06 13:45:19 otavio: but how can it be bb cache, if opkg is doing the rootfs operation? Jun 06 13:45:57 cbrake: it doesn't appears on the rootfs log Jun 06 13:46:17 cbrake: my rootfs has xfdesktop-locale-pt-br installed but it doesn't appear on rootfs log Jun 06 13:46:57 cbrake: no, it doesn't... Jun 06 13:47:17 cbrake: grr base-pt-br is installed Jun 06 13:47:30 cbrake: xfdesktop-locale-pt-br doesn't Jun 06 13:47:50 otavio: it seems that this must be resolved at the opkg level so it works the same at rootfs time and installing packages at run time on the device. Jun 06 13:48:47 XorA: the log you found about greedy Jun 06 13:48:53 otavio: yes Jun 06 13:48:55 XorA: is it at log.do_rootfs? Jun 06 13:49:05 otavio: no, package install on device Jun 06 13:49:23 XorA: on device .. hum Jun 06 13:49:25 otavio: but if opkg doesnt work the same during image creation it has a problem Jun 06 13:49:34 otavio: maybe it looks at the host by mistake? Jun 06 13:49:44 otavio: you would have to trace the code Jun 06 13:50:31 XorA: you mean the host arch? Jun 06 13:51:10 otavio: opkg is using -o (offlineroot) during image creation, but there were some bugs where it looked outside the offline root maybe this is one of them Jun 06 13:54:20 otavio: ahah, you need gpe-contacts-locale* to make greedy work Jun 06 13:54:31 otavio: in Depends: line Jun 06 13:54:46 XorA: *? Jun 06 13:55:10 otavio: I think that means depend on all installable ipk's that match the pattern Jun 06 13:55:21 /me tries that ... Jun 06 13:55:35 otavio: so it will install all locales where the base-locale is installed making them installable Jun 06 13:55:50 * otavio tries at image Jun 06 13:57:14 RDEPENDS/RRECOMMENDS or otherwise requires the runtime entity 'glibc-locale*' but it wasn't found in any PACKAGE or RPROVIDES variables Jun 06 13:57:17 NOTE: Runtime target 'glibc-locale*' is unbuildable, removing... Jun 06 13:57:38 same error here Jun 06 13:57:44 * cbrake takes another hard look at opkg source ... Jun 06 13:58:51 cbrake: this error comes from bb.taskdata Jun 06 13:59:09 otavio: ahh, good point Jun 06 13:59:40 otavio: so bitbake is bailing before opkg has a chance to do anything -- excellent Jun 06 14:00:02 cbrake: probably Jun 06 14:00:05 * cbrake tries on the target device .... Jun 06 14:01:07 # Disabled by RP 18/06/07 Jun 06 14:01:07 # Wildcards aren't supported in debian Jun 06 14:01:07 # They break with ipkg since glibc-locale* will mean that Jun 06 14:01:07 # glibc-localedata-translit* won't install as a dependency Jun 06 14:01:07 # for some other package which breaks meta-toolchain Jun 06 14:01:16 in packages.bbclass Jun 06 14:02:01 otavio: not much better: http://pastebin.com/m58075263 Jun 06 14:02:05 otavio: that comment probably explains it Jun 06 14:02:27 otavio: line 283 downwards in package.bbcllass Jun 06 14:02:42 otavio: so basically its been turned off by RP Jun 06 14:02:49 XorA: it doesn't; it adds a rdepends Jun 06 14:02:56 XorA: same as if we add on the image Jun 06 14:03:09 XorA: but the problem is that bb doesn't know they does exist Jun 06 14:03:15 otavio: it stops an ipk havind Depends: blah-locale* Jun 06 14:03:26 otavio: the problem isnt at .bb level its opkg level Jun 06 14:03:44 XorA: sure. but you adding it byhand would solve and it doesn't Jun 06 14:03:51 XorA: look the cbrake log above Jun 06 14:03:57 otavio: we dont make any greedy Depends: so opkg cant know to install them Jun 06 14:04:04 is there some autobuilds for -dev branch? Jun 06 14:04:07 otavio: you cant add it to the .bb file Jun 06 14:04:19 otavio: you need to add it to the .ipk files Jun 06 14:05:01 otavio: I think this probably needs something in package_ipk.bbclass to add them as a special case for .ipk files Jun 06 14:05:21 * otavio builds to test Jun 06 14:06:51 XorA: what does a greedy depends look like in the ipk control data? Jun 06 14:07:08 cbrake: Depends: blah-locale* Jun 06 14:07:30 cbrake: old OZ feeds http://www.openzaurus.org/feed-browser/?action=details&pnm=gpe-contacts Jun 06 14:08:17 * otavio builds an image for test Jun 06 14:09:59 XorA: doesn't work Jun 06 14:10:16 XorA: bb doesn't know about any *-locale-* packages Jun 06 14:10:34 XorA: in my case xfdesktop-locale-pt-br Jun 06 14:10:36 otavio: and it shouldnt Jun 06 14:10:43 XorA: it should Jun 06 14:10:47 otavio: shouldnt Jun 06 14:10:50 XorA: the taskdata checks for it Jun 06 14:10:53 XorA: and fail Jun 06 14:10:57 works. Jun 06 14:11:08 ERROR: '['/home/otavio/hacking/ossystems/oe/packages/images/osthindesktop-image.bb']' RDEPENDS/RRECOMMENDS or otherwise requires the runtime entity 'xfdesktop-locale-pt-br' but it wasn't found in any PACKAGE or RPROVIDES variables Jun 06 14:11:30 otavio: you are mixing up .bb and .ipk Jun 06 14:11:45 XorA: i built without that and it didn't install it Jun 06 14:12:17 otavio: I think because we no longer give .ipk files greedy dependencies on their locales so opkg ignores locales Jun 06 14:12:31 XorA: I've added the RP disabled code Jun 06 14:12:45 otavio: you would need to regenerate the whole feed Jun 06 14:13:01 otavio: rm -rf tmp/ time Jun 06 14:13:13 XorA: that package only would be enough I guess Jun 06 14:13:21 otavio: yeah Jun 06 14:13:29 XorA: rebuilding it Jun 06 14:13:46 then check the control info Jun 06 14:13:53 XorA: where? Jun 06 14:13:57 I am pushing some changes to poky now Jun 06 14:14:13 dpkg-deb -I tmp/deploy/*/*/blah.ipk Jun 06 14:14:27 hrw: you might write the patch somewhere Jun 06 14:14:40 why do I get feeling hrw is coding while we whitter on :-) Jun 06 14:15:56 Depends: xfdesktop-locale*, ... Jun 06 14:16:51 check poky revisions: 4602-4604 4588-4589 Jun 06 14:17:21 where is poky source? Jun 06 14:17:31 hrw: can't you do a diff for us? Jun 06 14:17:35 it adds new variable: IMAGE_LOCALES which is used by ROOTFS_POSTINSTALL_COMMAND install_all_locales Jun 06 14:17:39 hrw: we have to wait until the public svn catches up Jun 06 14:17:48 XorA: it is in public svn Jun 06 14:17:50 hrw: 4582 is the latest rev at moment Jun 06 14:17:59 XorA: I just pushed changes Jun 06 14:18:17 hrw: web interface shows 4582 as last checkin Jun 06 14:18:19 hrw: why not use IMAGE_LINGUAS? Jun 06 14:18:50 otavio: because they keep list of languages in en-us en-gb style and locale pacakges are en, fr, de Jun 06 14:19:11 hrw: pt-br is one locale package Jun 06 14:19:30 and we cant use IMAGE_LINGUAS (as let use all before '-') because zh_cn != zh_tw, pt != pt_br etc Jun 06 14:19:51 hrw: ah I see getting confused Jun 06 14:20:10 it is overcomplicated Jun 06 14:21:49 example: IMAGE_LINGUAS = "pl_PL en_GB de_DE zh_CN" should install Polish, British, German and Chinese language and translations. but it does not install all translations... so we set IMAGE_LOCALES = "pl en en-gb de zh-cn" to get all *-locale-{entries from IMAGE_LOCALES} for installed packages into rootfs Jun 06 14:22:29 omg ... this is really confusing Jun 06 14:22:35 and we have to use IMAGE_LOCALES = "en en-gb" as some packages have multiple English translations (American/British/Australian) Jun 06 14:22:47 but this looks like a workaround instead a solution from how I see the problem Jun 06 14:23:09 hrw: and it'll add the base translation properly Jun 06 14:23:30 hrw: with only image_linguas Jun 06 14:23:51 otavio: the effect is that if IMAGE_LINGUAS = "en_GB" and IMAGE_LOCALES = "en pl" you will also get Polish stuff Jun 06 14:24:19 wouldnt it be easier for the whole world to learn chinese :-) Jun 06 14:24:19 hrw: but this is non-sense ... at least for me Jun 06 14:24:22 otavio: and yes, this is workaround. Jun 06 14:24:34 hrw: hehe Jun 06 14:24:45 03thebohemian 07org.oe.dev * rb099f51e... 10/ (11 files in 5 dirs): (log message trimmed) Jun 06 14:24:45 classpath-native: Add patch allowing PhoneME builds with Classpath tools. Jun 06 14:24:45 classpath-minimal: Move to bugfix release 0.97.2. Jun 06 14:24:45 classpath: Move to bugfix release 0.97.2. Jun 06 14:24:46 chinook-compat.conf: Set classpath preference to 0.97.2. Jun 06 14:24:48 preferred-om-2008-versions.inc: Set classpath preference to 0.97.2. Jun 06 14:24:50 classpath 0.97.2: Jun 06 14:24:51 XorA: or Polish Jun 06 14:25:02 hrw: need oil for my throat for polish Jun 06 14:25:07 03florian 07org.oe.dev * r7cdf64b4... 10/ (1 packages/gpe-login/gpe-login_0.93.bb): gpe-login: add 0.93 Jun 06 14:25:13 rschuster1: can you next time split each commit into separate one? Jun 06 14:25:18 03woglinde2 07org.oe.dev * r002eda28... 10/ (1 packages/zlib/zlib_1.2.3.bb): Jun 06 14:25:22 zlib: add inherit pkgconfig Jun 06 14:25:28 * after applying the .pc-file-patch zlib use the pkgconfig-class now Jun 06 14:25:32 * bump PR Jun 06 14:25:36 03woglinde2 07org.oe.dev * r3c3876e3... 10/ (3 files in 3 dirs): Jun 06 14:25:37 XorA: try with this: "Król Karol kupił królowej Karolinie korale koloru koralowego" Jun 06 14:25:44 autoconf-2.59: fix building when sh=ksh Jun 06 14:25:48 * add autoconf-sh.patch Jun 06 14:25:52 * closes long outstanding bug 1088 Jun 06 14:25:56 * bump PR Jun 06 14:26:02 hrw: speak that in english and it sounds almost scottish Jun 06 14:26:08 cbrake: did you have found anything at bb code? Jun 06 14:27:09 otavio: no, I have not. It seems that once you set a Provides, BB does not let you use the package file name as a RDEPENDS any more Jun 06 14:27:15 otavio: see last email I sent to the list Jun 06 14:27:26 hrw: hm. I thought it makes sense to put all related stuff into one patch (nothing workable would have been produced without the other changes) Jun 06 14:27:38 cbrake: want to do a test? Jun 06 14:27:46 cbrake: you're using glibc right? Jun 06 14:27:50 otavio: right Jun 06 14:27:55 cbrake: hold. Jun 06 14:28:24 hrw: but sure I can do multiple 'commits' and only one 'push' next time. I am just still not that used to mtn Jun 06 14:28:36 rschuster1: can you give first line a summary then? Jun 06 14:28:42 cbrake: good. Open glibc-package.bbclass Jun 06 14:28:45 hrw: ok Jun 06 14:28:54 cbrake: go to PACKAGES_DYNAMIC line Jun 06 14:29:03 cbrake: add 'glibc-locale-*' Jun 06 14:29:17 cbrake: you don't need to rebuild the package Jun 06 14:29:24 rschuster1: I use that way with git(-svn) Jun 06 14:29:29 cbrake: now, add the language on the image file Jun 06 14:29:34 cbrake: it will work Jun 06 14:29:42 hrw: you poky changes look workable. They only apply at rootfs time, but at run time, you can just manually install the -locale-. Jun 06 14:30:07 its a pity we lose the automatic handling Jun 06 14:30:08 cbrake: on device you can do that. but not every device has network for it Jun 06 14:30:33 I think the commented out code should move to packge_ipk.bbclass somehow Jun 06 14:30:42 hrw: what cbrake means is that it neesd to be done package by package Jun 06 14:30:53 XorA: yes, it should Jun 06 14:30:56 for those who are interested: Jun 06 14:30:56 root@om-gta02:/usr/lib/jvm/phoneme-advanced-foundation/bin# ./cvm -cp ~/bsh.jar bsh.Interpreter Jun 06 14:30:56 BeanShell 2.0b4 - by Pat Niemeyer (pat@pat.net) Jun 06 14:30:56 bsh % print(System.getProperty("java.vm.name")); Jun 06 14:30:56 CVM Jun 06 14:31:00 XorA: but it doesn't work anyway Jun 06 14:31:09 XorA: not at rootfs time Jun 06 14:31:19 thats beanshell working on phoneme advanced (interpreter only atm) Jun 06 14:31:21 otavio: it gets one step closer Jun 06 14:31:41 otavio: then you just have to get the base locales to install and opkg should work same on rootfs creation as on device Jun 06 14:31:54 XorA: it doesn't Jun 06 14:31:59 XorA: base locale installs Jun 06 14:32:16 otavio: then find the bug in opkg that prevents it working Jun 06 14:32:27 XorA: i'm trying to do so Jun 06 14:33:04 question to ARM experts: is AAPCS == GNUEABI ? Jun 06 14:33:05 * otavio tries another thing Jun 06 14:33:12 other idea was "opkg --with-locales 'de en-gb en pl zh-cn' install ${IMAGE_INSTALL}" but this require teaching opkg what locale are Jun 06 14:33:52 XorA: ok ... Jun 06 14:33:53 APCS is oabi Jun 06 14:33:58 http://www.chiark.greenend.org.uk/~theom/riscos/docs/CodeStds/APCS.txt Jun 06 14:33:59 XorA: it works partially Jun 06 14:34:14 So we have two issues here Jun 06 14:34:33 The commented out code, needs to be in to use opkg greedy dependencies Jun 06 14:35:01 and P_D line needs to be on .bb files since it's the only read thing while cache building Jun 06 14:35:12 If those two are done, it works Jun 06 14:35:21 * otavio just tested it Jun 06 14:35:51 woo woo Jun 06 14:36:00 so just need to wriggle the code in cleanly Jun 06 14:36:04 Now, the second part of bug is the hard one Jun 06 14:36:30 XorA: it works but you need to write, byhand, PACKAGES_DYNAMIC at every .bb that builds locales Jun 06 14:36:38 XorA: with 'foo-locale-*' Jun 06 14:36:53 XorA: and this is nonsense Jun 06 14:36:55 otavio: that makes no sense, bitbake doesnt care about locales Jun 06 14:37:03 XorA: exactly Jun 06 14:37:31 otavio: maybe time to speak to RP Jun 06 14:37:47 * otavio tries another thing again Jun 06 14:37:50 XorA: somehow, bitbake tracks both RDEPENDS and DEPENDS, so it does care about stuff at the package level Jun 06 14:38:20 cbrake: it will try and make sure all RDEPENDS are available for debian renaming purposes Jun 06 14:38:29 cbrake: I guess we trip that up Jun 06 14:38:55 cbrake: if that code was moved to package_ipk.bbclass I guess it could inject into control file without adding to RDEPENDS Jun 06 14:39:08 XorA: no no Jun 06 14:39:21 cbrake: XorA is right Jun 06 14:39:23 XorA: is RDEPENDS also required so the stuff in IMAGE_INSTALL can automatically add DEPENDS? Jun 06 14:39:34 cbrake: we don't need to add the package on IMAGE_INSTALL Jun 06 14:40:58 XorA: will put it in ipk, somewhere Jun 06 14:40:58 cbrake: when making images I think all RDEPENDS get built, this is so we know what debian renaming did to the package name Jun 06 14:41:09 cbrake: XorA is right, it works Jun 06 14:41:19 cbrake: so I can why adding RDEPEND += "blah*" would upset it slightly Jun 06 14:41:24 cbrake: I'll add it to ipk and do a full build Jun 06 14:41:33 cbrake: hold on Jun 06 14:42:13 but we are now deep into bitbake voodoo and Im getting lost Jun 06 14:45:02 lrg, broonie: ping Jun 06 14:45:06 * * OE Bug 430 has been RESOLVED (WONTFIX) by heinold(AT)inf.fu-berlin.de Jun 06 14:45:08 * * patch to run fltklib with diet-x11 Jun 06 14:45:10 hi Sup3rkiddo Jun 06 14:45:10 * * http://bugs.openembedded.net/show_bug.cgi?id=430 Jun 06 14:45:43 florian, howdy Jun 06 14:45:46 03woglinde2 07org.oe.dev * re3912964... 10/ (5 files in 3 dirs): Jun 06 14:45:46 fltk: update to 1.1.9 and make it build again Jun 06 14:45:46 * make fltk build again Jun 06 14:45:46 * disable test-dir for now and until we fixed that fluid is build natvie Jun 06 14:45:46 * commit makes bug 430 invalid Jun 06 14:49:57 thesing: Why is it that a simple "MACHINE=collie bitbake linux-rp" now builds initramfs-kexec-image (which fails for me)? Jun 06 14:50:16 building a test package Jun 06 14:51:56 cbrake: let me test and then I post the patch Jun 06 14:52:06 Laibsch: is the INITRAMFS_IMAGE still in linux-rp ? Jun 06 14:52:22 Don't think so Jun 06 14:52:26 Let me see Jun 06 14:53:39 can I ask a little question? just to ensure I'm going in the right direction. Jun 06 14:55:06 I'm trying to boot my device with a self-built kernel. I shows me a nice logo and a lot of kernel messages. It mounts a partition on a SD card okay, then it just says 'Freeing init memory: 96K' Jun 06 14:55:49 Does that mean that init is broken and that I should fix this or is it smething else? Jun 06 14:55:53 lantern_burns__: do you have a jtag debugger attached? Jun 06 14:56:07 no, I even don't know what's that Jun 06 14:56:37 is it some hardware? Jun 06 14:56:48 lantern_burns__: http://www.abatron.ch/products/bdi-family.html Jun 06 14:56:57 lantern_burns__: can sometimes cause problems like that if not configured correctly Jun 06 14:57:10 lantern_burns__: what processor? Jun 06 14:57:21 thesing: It does not look like it: http://rafb.net/p/k4qRm511.html Jun 06 14:57:26 oh, I don't think that I can even buy it, I'm from Russia Jun 06 14:58:24 device: HTC Beetles, processor... Intel XScale PXA272 Jun 06 14:58:36 lantern_burns__: if ARM, sometimes this can happen if you don't have EABI support enabled in the kernel Jun 06 14:59:07 then I should enable-recompile-try again?.. Jun 06 14:59:39 'Use the ARM EABI to compile the kernel' - that one?.. Jun 06 14:59:47 lantern_burns__: do you have CONFIG_AEABI=y in your .config? Jun 06 15:00:05 no Jun 06 15:00:30 lantern_burns__: that is likely the problem if you are using a EABI toolchain Jun 06 15:00:35 lantern_burns__: to build the rootfs Jun 06 15:00:47 ok, I'll try with EABI enabled Jun 06 15:01:13 thanks a lot Jun 06 15:01:59 Laibsch: no idea then Jun 06 15:02:15 cbrake: works! Jun 06 15:02:37 cbrake, it just fails to compile Jun 06 15:03:08 cc1: error: invalid option `abi=aapcs-linux' Jun 06 15:03:43 http://pastebin.com/f7c89a4f8 Jun 06 15:03:47 cbrake: ^ Jun 06 15:04:36 * otavio builds two packages to test if image ends with proper locale packages Jun 06 15:05:06 lantern_burns__: try angstrom distro in OE then :-) Jun 06 15:05:12 otavio: trying here ... Jun 06 15:05:21 cbrake: you need to rebuild the package Jun 06 15:05:29 otavio: ok Jun 06 15:05:35 cbrake, angstrom kerbel just doesn't boot on my machine Jun 06 15:05:54 can I compile angstrom rootfs with eabi disabled? how? Jun 06 15:06:05 XorA: the patch should do what you suggested. It moved it to package_ipk Jun 06 15:06:13 lantern_burns__: are you using the OE toolchain? Jun 06 15:06:47 I use OE toolchain to build angstrom helloworld rootfs Jun 06 15:06:59 and I use handhelds.org utils to build the kernel Jun 06 15:07:12 including hh.org toolchain Jun 06 15:07:39 o fsck.. Jun 06 15:07:42 lantern_burns__: hh.org toolchain??? stick with OE built toolchain Jun 06 15:08:28 umm... my English ain't so good Jun 06 15:08:53 have you just told me to compile everything with OE toolchain? Jun 06 15:09:09 XorA: this addes the foo-locale* on the control Jun 06 15:09:15 XorA: so this part of bug is solved Jun 06 15:09:36 then how can I use it... 'outside' OE?.. Jun 06 15:09:42 otavio: well Im not brave enough to just apply patches like that to classes/* they tend to backfire in corner cases Jun 06 15:09:56 XorA: is a single line patch Jun 06 15:10:02 XorA: but ok Jun 06 15:10:15 otavio: sinble liners are sometimes the worst :-) Jun 06 15:10:25 RP: http://pastebin.com/f7c89a4f8 Jun 06 15:10:31 otavio: infact wont that make all locales depend on themselves? Jun 06 15:11:11 otavio: can you pastbin your Packages file from your feed? Jun 06 15:11:26 cbrake: which one? Jun 06 15:11:29 cbrake: i586? Jun 06 15:11:33 cbrake: the target? Jun 06 15:11:38 otavio: right, the target Jun 06 15:11:42 cbrake: sure Jun 06 15:13:19 http://pastebin.com Jun 06 15:14:01 otavio: complete URI? Jun 06 15:14:02 ops Jun 06 15:18:40 http://paste.debian.net/5512/ Jun 06 15:19:55 XorA: it does, I'll fix this Jun 06 15:23:57 XorA: i guess i fixed it to not recursively add it Jun 06 15:30:49 cbrake: found something? Jun 06 15:44:28 humm ... confusing. I've just found that -dev, -dbg packages has - virtual-locale-foo Jun 06 15:44:36 why pkg doesn't has it Jun 06 15:49:10 otavio: you might want to post your ideas to the maillist and see RP has to say about adding package-locale* back into Depends Jun 06 15:59:42 have a nice weekend guys Jun 06 16:20:47 anyone tried building from scratch after libtool update on angstrom ? Jun 06 16:23:51 03koen 07org.oe.dev * r48c629fa... 10/ (1 packages/i2c-tools/read-edid_1.4.1.bb): read-edid: add util that build parse-edid Jun 06 16:25:36 khem hm yes I Jun 06 16:26:00 woglinde: does libiconv build for you Jun 06 16:26:14 ah Jun 06 16:26:16 or you can try to build if it is not in there Jun 06 16:26:18 le me see Jun 06 16:26:48 reminds me Jun 06 16:26:56 to update it to 12 Jun 06 16:27:06 but I am on uclibc Jun 06 16:27:14 uclibc is better Jun 06 16:27:27 because console-image should pull it in automatically Jun 06 16:27:42 where it fails four you? Jun 06 16:27:47 config or build? Jun 06 16:27:50 in link stage Jun 06 16:27:54 hm oh Jun 06 16:28:07 * * OE Bug 4363 has been created by yan(AT)seiner.com Jun 06 16:28:09 * * matchbox-applet-card broken with 2.6 kernel Jun 06 16:28:10 because libtool thinks that -shared should not be passed to link tool Jun 06 16:28:10 * * http://bugs.openembedded.net/show_bug.cgi?id=4363 Jun 06 16:28:25 while creating .so file Jun 06 16:28:32 http://pastebin.com/f6c889607 Jun 06 16:28:35 cbrake: ^ Jun 06 16:28:39 XorA: ^ Jun 06 16:29:01 This does the right thing regarting the package Jun 06 16:29:07 okay fails here too Jun 06 16:29:21 woglinde: ok so I am not alone in the world Jun 06 16:29:23 otavio: can post pastebin the Package file again? Jun 06 16:29:24 :) Jun 06 16:29:42 khem thanks for the hint Jun 06 16:29:45 I will fix this Jun 06 16:29:46 I hate libtool and I dont know to fix it Jun 06 16:29:48 cbrake: it has the packages built properly Jun 06 16:29:55 woglinde: cool thanks Jun 06 16:29:57 cbrake: however the package aren't add to the image Jun 06 16:29:57 but the gettext/libiconv perverts autotools Jun 06 16:30:02 woglinde: today ? Jun 06 16:30:08 cbrake: we're dealing with two issues I guess Jun 06 16:30:12 khem no always Jun 06 16:30:26 he hides is little libs under tons of m4 files Jun 06 16:30:29 his Jun 06 16:30:38 woglinde: ok Jun 06 16:30:41 brb Jun 06 16:31:37 otavio: we probably don't want to install locale for dev packages and such, do we? Jun 06 16:31:54 otavio: bahh, this is complex Jun 06 16:31:59 cbrake: we do, if we're adding them on the image Jun 06 16:32:24 otavio: ok Jun 06 16:32:30 cbrake: someone not wanting would be the corner case i guess Jun 06 16:32:45 otavio: OK, post to the list and we'll see what happens Jun 06 16:32:45 cbrake: and could add a post rootfs_ hook to remove them Jun 06 16:32:56 cbrake: I'm not subscribe to the list yet Jun 06 16:33:01 cbrake: will need to do that Jun 06 16:33:12 otavio: oe-dev maillist? Jun 06 16:33:16 cbrake: http://paste.debian.net/5524/ Jun 06 16:34:45 otavio: still seeing stuff like: Jun 06 16:34:46 # Jun 06 16:35:12 otavio: xfdesktop-locale-be Depends on xfdesktop-local-be-locale* Jun 06 16:35:20 otavio: is that just because you did not rebuild that package? Jun 06 16:35:47 cbrake: yes Jun 06 16:36:02 cbrake: i've rebuild xfce4-panel Jun 06 16:37:00 otavio: I'm suggesting you post your patch to the oe maillist so the others can comment on issues before we consider applying. Jun 06 16:37:33 cbrake: sure. I'm just wanting to understand a missing thing before doing that Jun 06 16:38:07 otavio: I don't see any issues with it at this point, but I also don't understand RP's comments about "They break with ipkg since ..." Jun 06 16:39:50 cbrake: so do I Jun 06 16:40:10 cbrake: I'm not testing another theory ... Jun 06 16:45:07 * * OE Bug 3938 has been RESOLVED (FIXED) by yan(AT)seiner.com Jun 06 16:45:09 * * Marvel 8385 libertas driver will not wake from suspend 2.6.24 Jun 06 16:45:11 * * http://bugs.openembedded.net/show_bug.cgi?id=3938 Jun 06 16:48:40 tmp/work/oppitz-lx800-ossystems-linux/osthindesktop-image-1.0-r0/temp/log.do_rootfs.16517:Adding satisfier for greedy dependence: xfdesktop-locale-pt-br Jun 06 16:48:43 tmp/work/oppitz-lx800-ossystems-linux/osthindesktop-image-1.0-r0/temp/log.do_rootfs.16517:Adding satisfier for greedy dependence: xfce4-panel-locale-pt-br Jun 06 16:49:10 Hmm - I like that machine name.... :) Jun 06 16:56:24 thank you, guys - now I have a working kernel that runs uclibc built images from OE! Jun 06 16:56:34 thanks a lot! X) Jun 06 17:40:05 * otavio compiling from scratch for testing Jun 06 17:40:23 otavio which arch? Jun 06 17:40:34 woglinde: x86 Jun 06 17:46:19 03jeremy_laine 07org.oe.dev * r1abf4506... 10/ (1 packages/linux/linux.inc): Jun 06 17:46:23 linux.inc: optionally generate a kernel-devicetree package Jun 06 17:46:31 * if KERNEL_DEVICETREE is set, the devicetree will get shipped in a package Jun 06 17:46:35 called "kernel-devicetree" as /boot/devicetree-${KERNEL_VERSION} Jun 06 17:51:24 woglinde: Do you know why new libtool frets on libiconv Jun 06 18:01:45 khem not yet Jun 06 18:01:47 as I said Jun 06 18:02:00 the author use autotools in an isane way Jun 06 18:38:34 re flo Jun 06 18:40:09 re Jun 06 18:41:47 erm... don't you know, can I just 'debootstrap --arch arm sid ./target/' onto my sd card, install modules and then succesfully use debian on my PDA?.. Jun 06 18:51:01 Hello Jun 06 18:51:13 Does anyone else get !oebug 4343 Jun 06 18:51:15 ? Jun 06 20:21:57 hvontres|work: Sadly only one person acked you backport patch (http://lists.linuxtogo.org/pipermail/openembedded-stablebranch/2008-May/000086.html) up to now... (in http://lists.linuxtogo.org/pipermail/openembedded-stablebranch/2008-June/000092.html) Jun 06 20:22:57 helgeD_: yeah, I think I'll bump it tonight to see if I can't get the second vote :) Jun 06 20:23:11 helgeD_: btw, do you have rw acess to OE? Jun 06 20:23:31 hvontres|work: no (sadly). Jun 06 20:25:22 helgeD_: np. hrw asked me before, but I had beed holding out to see what the mtn replacement was going to be, but I think I should take him up on it Jun 06 20:26:22 hvontres|work: agreed. mtn is ugly :-) Jun 06 20:28:51 helgeD_: sorry, lost my connection Jun 06 20:29:23 helgeD_: actually, I don't really mind mtn so much, I just didn't want to re-learn everything right away Jun 06 20:36:13 damnit Jun 06 20:36:20 when libtool creates a .lai file Jun 06 20:38:37 hm Jun 06 20:50:34 cbrake: are you there? Jun 06 20:52:05 otavio: yes Jun 06 20:52:23 cbrake: ok. I've tested my theory and it looks correct ... Jun 06 20:52:36 cbrake: the problem is two connected issues Jun 06 20:52:52 cbrake: first, is solved by adding the change on package_ipk ... Jun 06 20:53:03 cbrake: so it adds foo-locale* Jun 06 20:53:51 cbrake: however it's not enough to get packages installed at rootfs creating time. It only happens when .bb file has PACKAGES_DYNAMIC = "foo-locale-*" Jun 06 20:54:02 cbrake: this second part I really don't understand Jun 06 20:56:06 * * OE Bug 4364 has been created by gwossum(AT)acm.org Jun 06 20:56:08 * * alsa-lib doesn't work properly with uClibc Jun 06 20:56:10 * * http://bugs.openembedded.net/show_bug.cgi?id=4364 Jun 06 20:58:33 otavio: noted Jun 06 20:58:47 otavio: I don't have any explanation for the P_D issue. Looking a bit more ... Jun 06 21:00:06 cbrake: I don't understand how cache might affect the root creation Jun 06 21:00:22 otavio: what kind of error do you get? Jun 06 21:00:42 cbrake: nothing, it just doesn't add the package on the image automaticaly Jun 06 21:01:07 cbrake: if you add it there, and set IMAGE_LINGUAS all packages that has P_D set, works Jun 06 21:01:33 otavio: have you diffed the Packages file before and after adding PACKAGES_DYNAMIC? Jun 06 21:02:11 cbrake: No but it does't change the package but the cache of bb Jun 06 21:02:44 otavio: it escapes me how that can change anything unless bb is running a second or third opkg-cl step Jun 06 21:03:35 cbrake: I don't understand how it works so I can't guess where the fault is Jun 06 21:08:15 Crofton|work: 2.6.25 works well on omap its not the kernel Jun 06 21:08:42 Crofton|work: its the compiler 4.3 which produced a unbootable kernel same kernel works ok with gcc 4.2 Jun 06 21:11:06 * * OE Bug 4365 has been created by gwossum(AT)acm.org Jun 06 21:11:08 * * Modify atime of sources when used Jun 06 21:11:10 * * http://bugs.openembedded.net/show_bug.cgi?id=4365 Jun 06 21:23:43 dropbear currently asks for a carriage return when password is empty how can I make it not to ask even that Jun 06 21:24:03 I can do it in openssh Jun 06 21:26:43 khem hm I found it Jun 06 21:26:55 woglinde: libtool thing ? Jun 06 21:27:02 or I found it in the libiconv cvs Jun 06 21:27:04 I hope Jun 06 21:27:12 what is that Jun 06 21:27:19 hm Makefile.in Jun 06 21:27:29 but let me see if it finishes Jun 06 21:27:37 otavio: thanks for all your work on this -- we'll get to the bottom of it eventually. I suspect hrw has good reasons for going the route he did ... Jun 06 21:27:37 btw. I have added a 1.12 recipe here which I will commit has same problem currently Jun 06 21:27:49 ah the gentoo fix ? Jun 06 21:27:53 otavio: sending a note to the list asking about RP's comments in the source ... Jun 06 21:27:59 I have it applied locally did not help me Jun 06 21:28:02 khem he wait Jun 06 21:28:06 I am working on 1.12 Jun 06 21:28:23 * RP notes people are talking about a change he made Jun 06 21:28:37 hey, its RP :-) Jun 06 21:28:39 RP: Hey! Jun 06 21:28:42 rp did you see my mail about FILES_? Jun 06 21:28:46 * RP hides Jun 06 21:29:01 * otavio runs and catches RP from arm.... Jun 06 21:29:06 khem args Jun 06 21:29:11 I missed a header Jun 06 21:29:17 RP: libtool upgrade has caused some hiccups it may be hidden bugs in packages though Jun 06 21:29:22 woglinde: I did. Why not just enhance bitbake.conf? Jun 06 21:29:54 khem: That doesn't surprise me in the least. I didn't push that upgrade so please don't blame me ;-) Jun 06 21:29:57 RP: cbrake and I are having trouble with locale packages. Jun 06 21:30:17 RP: I've found the problem and dunno how to fix the second part of it. Jun 06 21:30:32 otavio: Whats the second part? Jun 06 21:30:59 http://pastebin.com/f4cc6b764 Jun 06 21:31:07 RP: ^ this is related about your change Jun 06 21:31:08 otavio: I remember there being two problems with that code. One was it didn't work for dpkg. The second was it seemed to backfire for ipkg too :( Jun 06 21:31:08 rp I didnt know about bitbake.conf Jun 06 21:31:20 woglinde: Its where the defaults for FILES comes from Jun 06 21:31:44 RP: this is looking to work fine with opkg. Jun 06 21:31:51 rp ah Jun 06 21:31:58 why I didnt found it Jun 06 21:32:07 otavio: Do you really want to install all locales all the time? Jun 06 21:32:18 RP: for the specified languages Jun 06 21:32:33 otavio: Where do you specify the languages? Jun 06 21:32:38 args Jun 06 21:32:41 RP: IMAGE_LINGUAS Jun 06 21:32:46 I only searched under classes Jun 06 21:33:20 otavio: and how does -locale* get influenced by IMAGE_LINGUAS? Jun 06 21:33:35 rp hm whats the befinit putting it in bitbake.conf? Jun 06 21:33:42 benefit Jun 06 21:33:49 woglinde: then it just works? Jun 06 21:33:49 RP: Depends: libc6, virtual-locale-de Jun 06 21:34:31 cbrake: this is some magic in opkg/ipkg? Jun 06 21:34:43 RP: greedy depends Jun 06 21:34:43 RP: What worries me is that for rootfs creating to work (with the linguas and locales added) each .bb that builds locale-* needs to have PACKAGES_DYNAMIC set Jun 06 21:35:03 RP: yes, opkg handles it Jun 06 21:35:17 What exactly does opkg/ipkg do? Jun 06 21:35:31 RP: so, if the base package depends on glibc-locale*, then all glibc-locale-xx packages will be installed if possible Jun 06 21:35:32 rp hm I am think in modules or tools and thats why I think its better when it is seperated Jun 06 21:35:34 RP: Does matchbox-desktop-2 from svn need matchbox-common from svn too? Jun 06 21:35:41 RP: at least that is my understanding of how it used to work Jun 06 21:35:48 RP: opkg on seeing a * looks at what packages match blah* and already have their dependencies installed AFAIK Jun 06 21:36:06 * eekee can understand why RP wanted to hide : Jun 06 21:36:07 :) Jun 06 21:36:19 RP: and locale-base-LINGUA is installed Jun 06 21:36:28 flo_lap: We removed mb-common in poky iirc Jun 06 21:36:45 eekee: lol Jun 06 21:36:51 eekee: indeed Jun 06 21:37:33 I remember trying to understand how this worked at the time but it all just seemed broken :/ Jun 06 21:37:49 I guess the logic I was missing is this "already have their dependencies installed" Jun 06 21:37:54 RP: do you understand the second part of the problem? Jun 06 21:38:03 RP: the P_D related one Jun 06 21:38:05 otavio: The PACKAGES_DYANMIC issue? Jun 06 21:38:09 yes, I understand it Jun 06 21:38:12 RP: yep Jun 06 21:38:14 RP: I don't Jun 06 21:38:16 RP: hum ok... what contains the vfolders? Jun 06 21:38:45 RP: yeah, I had a quick look at the code and it looks quite insane, I hope I got the right reading of it Jun 06 21:38:46 RP: It looks that rootfs step depends somehow of bb cache Jun 06 21:39:02 flo_lap: I think the categories are created from the desktop files? Jun 06 21:39:08 RP: since bb cache is the only place where I found a reference for P_D usage Jun 06 21:39:43 otavio: PACKAGES_DYNAMIC is used where we can't tell the exact name out of the generated packages in advance Jun 06 21:39:44 RP: and since locale package are being dynamic built, from package class, it doesn't has P_D set anymore on .bb files Jun 06 21:39:50 RP: hence fails Jun 06 21:40:11 RP: at rootfs step? Jun 06 21:40:12 RP: ah maybe, I gues I'm too old ;) I just wondered why it does not really work on my Universal. Jun 06 21:40:38 otavio: It should be relatively safe to add PACKAGES_DYNAMIC = "${PN}-locale-*" to bitbake.conf Jun 06 21:40:46 woglinde: you said you had uclibc rfs Jun 06 21:40:58 The only issue will be packages that use PACAKGES_DYNAMIC = rather than =+ Jun 06 21:41:21 flo_lap: I know it works fine in Poky and that mb-common is no more Jun 06 21:41:27 RP: but then those would be bugs, that could be solve Jun 06 21:41:31 RP: let me try that Jun 06 21:41:44 flo_lap: You could always email Ross, he wrote it so should know ;-) Jun 06 21:41:46 khem? Jun 06 21:41:58 woglinde: dropbear crashed for me when I try to setup passwordless entry Jun 06 21:42:21 woglinde: I copy my id_dsa.pub to target's .ssh/authorized_keys Jun 06 21:42:26 otavio: I've no objection to reenabling that stuff if its make ipk specific and moved out the generic code Jun 06 21:42:42 RP: saw my patch? Jun 06 21:42:46 RP: worst case I can bug him with my experiments :) I just built it for that device and found out only the "media" desk has contents Jun 06 21:43:19 flo_lap: I know we did a big .desktop file cleanup in poky Jun 06 21:43:26 otavio: yes Jun 06 21:43:37 RP: does it looks fine for you? Jun 06 21:43:55 khem hm Jun 06 21:44:02 I only build so far, sorry Jun 06 21:44:07 otavio: thinking ;-) Jun 06 21:44:26 RP: I would just use Poky, but you don't have GPE ;) Jun 06 21:44:40 flo_lap: Create a meta-gpe overlay ;-) Jun 06 21:45:18 RP: I just posted this to the list to summarize our earlier discussion: http://projects.linuxtogo.org/pipermail/openembedded-users/2008-June/000608.html Jun 06 21:45:31 * XorA|gone still fails to see why we need to add PACKAGES_DYNAMIC for locales to everything Jun 06 21:46:00 RP: Maybe I start with a matchbox*2* overlay for Angstrom :) Jun 06 21:46:30 RP: http://paste.debian.net/5561/ (with P_D on bb.conf) Jun 06 21:46:37 otavio: I'm going to test your patch now ... Jun 06 21:47:49 flo_lap: sure ;-) Jun 06 21:48:21 otavio: That looks odd :/ Jun 06 21:49:55 RP: any clue? Jun 06 21:49:58 cbrake: great Jun 06 21:50:02 cbrake: the classes one? Jun 06 21:50:47 otavio: yes Jun 06 21:50:58 otavio: probably be tomorrow before I finish, but I'll let you know what happens Jun 06 21:51:48 otavio: What patch did you try exactly? Jun 06 21:51:53 otavio: it worked here :/ Jun 06 21:52:10 * otavio redoing it Jun 06 21:52:22 * otavio removed the cache redone it Jun 06 21:55:25 http://pastebin.com/f2ba039e Jun 06 21:55:29 RP: ^ Jun 06 21:55:40 RP: same traceback Jun 06 21:56:28 RP: probably it doesn't expands ${PN} Jun 06 21:56:40 otavio: which bitbake? Jun 06 21:57:11 1.8.10-3 (sid) Jun 06 21:57:33 otavio: Can you try with bb from the 1.8 svn branch please? Jun 06 21:58:04 RP: where I get it? Jun 06 21:59:10 otavio: svn co http://svn.berlios.de/svnroot/repos/bitbake/branches/bitbake-1.8 Jun 06 21:59:43 RP: checking it out Jun 06 21:59:56 bitbake poky-image-minimal-mtdutils -e | grep ^PACKAGES Jun 06 21:59:57 PACKAGES_DYNAMIC="poky-image-minimal-mtdutils-locale-*" Jun 06 22:00:21 RP: any easy way to run it without installing it? Jun 06 22:00:34 otavio: You don't need to install bitbake to run it... Jun 06 22:00:53 otavio: Although it may conflict with your installed one :/ Jun 06 22:00:59 * RP never installs bitbake Jun 06 22:01:08 RP: I purged it already Jun 06 22:01:19 otavio: ok, just run it then Jun 06 22:02:37 XorA|gone: ping Jun 06 22:02:46 ant_: gnop Jun 06 22:02:56 hi Jun 06 22:03:02 ant_: wazzup? Jun 06 22:03:15 I have a question for you: http://gitweb.openembedded.net/?p=org.openembedded.dev.git;a=commitdiff;h=c8bd4fea4efb229747f9edddb38d79ac8d3dcfe6 Jun 06 22:03:42 could be that it killed keyboard modifiers on c7x0 (xserver-kdrive-imageon)? Jun 06 22:03:48 RP: same traceback Jun 06 22:04:06 RP: now, bitbake 1.8.11 Jun 06 22:04:41 ant_: no, as thats xserver-kdrive-1.4.0.90 Jun 06 22:04:51 otavio: :( Jun 06 22:05:06 XorA|gone: let's move to #angstrom Jun 06 22:06:31 RP: any clue? Jun 06 22:06:50 otavio: I'm trying it in an OE checkout Jun 06 22:08:37 otavio: It crashed. Its some recipe in OE thats not in poky Jun 06 22:09:20 RP: great! Jun 06 22:10:36 otavio: Its "uclibc++" that does it Jun 06 22:10:43 the libsigc++ Jun 06 22:10:46 then Jun 06 22:11:33 RP: any idea _why_? Jun 06 22:11:41 re Jun 06 22:12:07 otavio: ++ in a regular expression? Jun 06 22:12:46 RP: so bitbake expading needs to space them, no? Jun 06 22:12:54 RP: \+? Jun 06 22:13:11 RP: scape Jun 06 22:14:16 otavio: http://pastebin.com/m12cf9408 works Jun 06 22:14:34 but then the bugs about fo+ not matching foo will start :( Jun 06 22:16:55 RP: but it'll match uclibc++ Jun 06 22:17:04 RP: since + is escaped Jun 06 22:17:58 otavio: It means we can't use + characters in PACKAGES_DYANMIC regexps though Jun 06 22:20:09 * XorA|gone still doesnt see why we need to do this Jun 06 22:20:31 why should a dodgy opkg hack care about PACKAGES_DYNAMIC Jun 06 22:21:06 XorA|gone: I also doesn't see why too Jun 06 22:21:17 XorA|gone: why a cache is related to rootfs building Jun 06 22:21:44 Well, the lack of a PACKAGES_DYNAMIC for the locale packages is a bug Jun 06 22:21:55 As soon as someone does RDEPENDS = "contacts-locale-en-gb" we're screwed Jun 06 22:22:14 I'm presuming there is some issue like that? Jun 06 22:23:45 RP: well with the opkg hack we should need to do that RDEPEND Jun 06 22:23:54 I also don't like the opkg solution particularly since we can't apply it to dpkg and this means the two different package classes are going to give very different results Jun 06 22:24:10 yeah well that was my second complaint :-) Jun 06 22:24:19 I guess debian doesnt fine grain locales Jun 06 22:24:44 XorA|gone: The RDEPEND is made at a point long after bitbake has read and cares about the value of the RDEPENDS Jun 06 22:25:14 XorA|gone: It doesn't. Poky's solution Marcin has just added works for both ipkg and dpkg Jun 06 22:25:32 RP: marcins only works for image creation though? Jun 06 22:25:53 yes it works at build time, an "update-locales" script for images should be relateivly easy to write though Jun 06 22:26:12 RP: opkg solution is nice in that if locale-base-blah is install then all packages that have a locale in that language also install their locale Jun 06 22:26:14 then it just becomes a matter of running it Jun 06 22:26:31 update-locales would sort it once and for all Jun 06 22:26:31 XorA|gone: yes Jun 06 22:26:38 debian does have a purge-locales :-) Jun 06 22:27:17 I would like to see opkg and dpkg both working Jun 06 22:27:20 03jeremy_laine 07org.oe.dev * r943f8d7f... 10/ (4 files in 3 dirs): cairo-directfb_1.6.4.bb: resync cairo-directfb version with plain cairo Jun 06 22:27:38 RP: opkg is a nasty hack, I think an update-locales makes more sense Jun 06 22:27:44 RP: and works for both Jun 06 22:28:02 XorA|gone: Agreed IMO but this needs to go to the mailing list Jun 06 22:28:08 and now I need to sleep ;-) Jun 06 22:28:27 RP: sounds like a plan, I need to get drunker :-) Jun 06 22:34:06 RP: opkg doesn't works recursively Jun 06 22:34:20 RP: I guess update-locales would be the best solution indeed Jun 06 22:36:19 RP: I want to override dropbear with openssh Jun 06 22:36:23 how to do it Jun 06 22:36:43 I tried to add it in ANGSTROM_EXTRA_INSTALL Jun 06 22:36:55 dropbear still is default sshd Jun 06 22:38:28 khem: If you install both update-alternatives comes into play Jun 06 22:38:48 khem: You need to remove one from the list of packages to install as well as add the other Jun 06 22:39:05 XorA|gone: I never said I was sober ;-) Jun 06 22:39:08 khem: there is something like DISTRO_SSH Jun 06 22:39:26 anyhow, /me is really gone :) Jun 06 22:39:43 hurray Jun 06 22:39:50 libiconv compiles again Jun 06 22:39:56 khem: DISTRO_SSH_DAEMON ?= "dropbear" Jun 06 22:40:09 khem: in task-base so override that Jun 06 22:45:14 XorA|gone: if I set it in local.conf will that be ok too ? Jun 06 22:45:30 khem: yes, then just rebuild task-base Jun 06 22:45:39 XorA|gone: ok Jun 06 22:46:33 XorA|gone: England is playing New Zealand (cricket) Jun 06 22:46:55 khem: there is no point mentioning cricket, a loss is a sure thing :-) Jun 06 22:47:39 XorA|gone: heh Jun 06 22:48:38 someone from australia told me nice joke last year when englang lost badly to aussies in ashes Jun 06 22:50:00 * XorA|gone is even more lost in despair as he is an Englishman in Scotland Jun 06 22:50:09 and the only thing scotland ever won was the elephant polo Jun 06 22:50:42 and there are no elephants in scotland Jun 06 22:54:49 :)) Jun 06 22:55:33 * XorA|gone invents haggis polo Jun 06 22:55:46 everyone can come ride wild haggis Jun 06 22:56:06 * * OE Bug 4366 has been created by andrea.adami(AT)gmail.com Jun 06 22:56:08 * * broken xmodmap for c7x0 Jun 06 22:56:10 * * http://bugs.openembedded.net/show_bug.cgi?id=4366 Jun 06 23:28:35 hurray Jun 06 23:28:46 khem I can check in now libiconv Jun 06 23:29:56 khem: what was the joke? Jun 06 23:36:00 03woglinde2 07org.oe.dev * r2bf25e1d... 10/ (6 files in 3 dirs): Jun 06 23:36:02 libiconv: update to version 1.12 and fixing libtool-2 build Jun 06 23:36:04 * backport preload.patch from libiconv-cvs Jun 06 23:48:03 good nite Jun 07 00:27:13 rwhitby: here it goes Jun 07 00:29:21 rwhitby: once there was a small boy and his dad would beat him so his mom sent him to his grand mother. At grandmothers his cousin brother will beat him. So his mom sent him to his uncle. There uncle's son would would beat him. She was very frustrated and then she met her friend from australia who suggested Jun 07 00:29:41 why dont you sent him to english cricket team. They cant beat anybody Jun 07 00:34:23 Ouch. :D Jun 07 00:51:08 mwester: do u use git Jun 07 01:06:06 khem: nice one :-) Jun 07 01:12:42 03jeremy_laine 07org.oe.dev * r53d0d221... 10/ (3 files in 3 dirs): libsvg_0.1.4.bb: add patch to solve autoreconf errors Jun 07 01:12:57 rwhitby: r u in uk Jun 07 01:13:17 khem: right now, in the US. usually, in Australia. Jun 07 01:13:26 rwhitby: this time around they are beating the kiwis Jun 07 01:13:37 indeed :-) Jun 07 01:13:41 rwhitby: where in us Jun 07 01:14:32 Urbana/Champaign (near uiuc.edu) today, LAX tomorrow. Jun 07 01:14:41 i see Jun 07 01:14:48 i m in san jose Jun 07 01:15:08 ah, I was there two days ago. Jun 07 01:15:30 hmm we could have met Jun 07 01:15:43 are u on a tour Jun 07 01:15:56 like rockstars Jun 07 01:15:59 :) Jun 07 01:21:14 ERROR: Cannot satisfy the following dependencies for task-base-extended: dropbear Jun 07 01:21:59 even though i select openssh to provide ssh in my local.conf how come task-base is asking for dropbear Jun 07 02:35:03 03jeremy_laine 07org.oe.dev * r73964f8e... 10/ (5 files in 3 dirs): directfb: don't use linux/config.h in powerpc assembly **** ENDING LOGGING AT Sat Jun 07 02:59:56 2008