**** BEGIN LOGGING AT Wed Dec 01 02:59:57 2010 Dec 01 03:14:09 hmm turns out busybox is built-in to ubuntu after all.. i'll save this problem for the mailing lists.. getting late over here.. thanks for ur help grg.. goodnight :) Dec 01 03:14:37 andyzammy, ok, then i suppose it should be reasonably tested. Dec 01 03:14:46 i'll submit a patch to fix... Dec 01 03:16:04 not sure if it's relevant but i used this setup script to instal OE: 'git clone git://gitorious.org/angstrom/angstrom-setup-scripts.git' Dec 01 03:17:57 andyzammy, the '--dereference' in openssl.inc just needs to be changed to use the short option. its a simple change Dec 01 03:19:08 oh right, where do i find the .inc file? should i just used find cmd? Dec 01 03:20:23 openembedded/recipes/openssl.inc Dec 01 03:25:31 it worked - thanks for the fix! :) never would have found that Dec 01 03:26:31 so is that how busybox works then? if a command is searched for in the path and busybox has it, it'll use busybox's version if it comes up first? Dec 01 03:27:06 andyzammy, not really, no Dec 01 03:27:22 its just about how things appear in your path Dec 01 03:28:02 e.g. if you have PATH=/bin and you have /bin/cp as a symlink to /bin/busybox-someversionblah, then invoking cp will call the busybox cp. Dec 01 03:29:18 most gnu systems use coreutils cp, not busybox. So you would have a full cp binary on those systems Dec 01 03:29:36 you may have both installed, but only one will live at /bin/cp Dec 01 03:32:14 i do have a cp binary, and as far as i can tell it's not a link Dec 01 03:41:44 okay gona let this run overnight, thanks again for ur help grg! Dec 01 03:41:49 np Dec 01 04:02:06 anyone around? Dec 01 04:02:21 ask directly the question Dec 01 04:03:38 I'm working on support for the iPhone3G in openembedded. Having some trouble with oe finding my defconfig. Here's the pastebin: http://pastebin.com/LeVZXch2 Dec 01 04:04:20 also gived more infos such as Dec 01 04:04:44 *the fact that you want gnu/linux on it and not ios Dec 01 04:05:02 *the name of the kernel recipe and its content Dec 01 04:05:12 *the machine config Dec 01 04:14:35 bye Dec 01 04:14:55 I'm working on support for the iPhone3G in openembedded for GNU/Linux. Having some trouble with oe finding my defconfig. Here's the pastebin from compiling: http://pastebin.com/LeVZXch2 ; Here's the machine config file: http://dl.dropbox.com/u/8083946/iphone3g.conf ; Here's the kernel recipe: http://dl.dropbox.com/u/8083946/linux-iphone_git.bb Dec 01 04:15:46 grg maybe you could take a look....I'm off Dec 01 04:15:59 ty GNUtoo Dec 01 04:18:46 Lopi, where have you put the defconfig? Dec 01 04:19:40 grg: openembedded/recipes/linux/linux-iphone3g/defconfig Dec 01 04:20:22 Lopi, you called your kernel recipe linux-iphone_git.bb - so the PN (package name) = linux-iphone. Dec 01 04:20:42 OE will look in openembedded/recipes/linux/${PN}/defconfig Dec 01 04:20:49 (among other places) Dec 01 04:21:21 grg: ahh I see. So it should be as simple as renaming linux-iphone3g to linux-iphone Dec 01 04:21:26 yep Dec 01 04:36:34 grg: Well, it found the defconfig file. I'm pretty sure it failed actually compiling the kernel though. http://pastebin.com/23bYSESi Dec 01 04:38:07 grg: I think I'll have to pull out the ethernet gadget for now Dec 01 04:47:25 Lopi, you probably just need to #include at the top of drivers/usb/gadget/ether.c Dec 01 04:47:43 That is one whacky mess of include madness going on in the ethernet gadget Dec 01 04:48:04 grg: ah yeah you're right one of the iDroid devs told me that awhile ago Dec 01 04:48:23 grg: I commented the ethernet gadget stuff in the defconfig for now as I can't really use it atm anyway Dec 01 04:48:29 ok Dec 01 04:48:46 did the iDroid dev tell you who to punch for writing that code? Dec 01 04:50:26 grg: hah no Dec 01 04:52:33 kergoth, automake seems to have the same problem as autoconf, with configure executing perl: http://tinderbox.openembedded.org/public/logs/task/12084516.txt Dec 01 04:52:38 grg: who do I need to punch? it's really annoying having to put that every time I want to compile a kernel for my project ;p Dec 01 04:52:51 * kergoth rolls eyes, not too surprising Dec 01 04:52:52 kergoth, But when i fix it the same way i get a different error http://tinderbox.openembedded.org/public/logs/task/12087974.txt any clues? Dec 01 04:53:17 hmm, no, thats strange Dec 01 04:53:24 Lopi, i suppose you report it to whoever maintains the ethernet gadget Dec 01 04:56:22 Lopi, see if this gives you sane output (it may not) ./scripts/get_maintainer.pl -f drivers/usb/gadget/ether.c Dec 01 04:57:31 grg: I will check in a few mins. Going to hop in the shower to clear my head before the next error comes. Dec 01 06:06:54 03Martin Jansa 07master * rcd24f44394 10openembedded.git/docs/usermanual/chapters/common_use_cases.xml: Dec 01 06:06:54 usermanual: update info with SRCPV, in recipe SRCREV and remove sane-srcdates Dec 01 06:06:54 Signed-off-by: Martin Jansa Dec 01 06:06:55 03Martin Jansa 07master * r07bf6c4b71 10openembedded.git/recipes/linux/ (linux/spitz/defconfig linux_git.bb): Dec 01 06:06:55 linux_git: bump SRCREV for 2.6.37-rc4 Dec 01 06:31:01 grg: the compile finished with no errors, now all I have to do is test it :) Dec 01 06:31:14 cool Dec 01 07:32:15 gm Dec 01 07:49:52 good morning Dec 01 08:02:42 morning Dec 01 08:03:12 aww grg went to sleep Dec 01 08:03:23 I'm sooo close Dec 01 08:57:07 ka6sox: when you will be online next time please install Nuke Dec 01 09:31:37 eFfeM_work: are you around? Dec 01 09:32:25 about multiple providers, during build I found one of the strangest case... libsegfault Dec 01 09:32:49 03Martin Jansa 07master * r74d57180bc 10openembedded.git/ (conf/distro/include/sane-srcrevs.inc recipes/opkg/opkg.inc): Dec 01 09:32:49 opkg: move OPKG_SRCREV to opkg.inc directly Dec 01 09:32:49 * now OPKG_SRCREV was used only in this one .inc, so move it there Dec 01 09:32:49 Acked-by: Graham Gower Dec 01 09:32:49 Signed-off-by: Martin Jansa Dec 01 09:36:27 hm.. good idea to release manual with OE Dec 01 10:02:31 hi, is there a proven/recommended way of having only the smallest number of recipes in the /recipes dir? I tried deleting the unused dirs, but it all works well until I git pull new revisions. Symlinks perhaps? Dec 01 10:06:36 metiu: what about BBMASK? it should allow you to exclude groups of recipes (from being built, they may still be parsed) Dec 01 10:07:40 bluelightning: the problem is more with parsing 7k+ recipes, it takes some time (a couple minutes or so) even on a fast machine Dec 01 10:08:02 bluelightning: angstrom-2008.1 for qemux86/ben-nanonote was build fine from your branch Dec 01 10:08:19 metiu: slow parsing is only on first run - then it is cached Dec 01 10:08:19 collie/akita/tosa/efikamx was out of space.. Dec 01 10:08:23 retrying now Dec 01 10:08:28 metiu: ok... you could try the bitbake parallel parsing patch (experimental at the moment) Dec 01 10:08:44 it is in bitbake master already Dec 01 10:08:55 ah, I had missed that... cool Dec 01 10:08:58 metiu: consider using bb master from git Dec 01 10:09:23 Jay7: great, so I guess you didn't have any issues with the git fetcher like khem did... Dec 01 10:09:48 hrw: yes I know that, it's that I'm touching local.conf sometimes, and that invalidates the cache, as usual Dec 01 10:10:04 bluelightning: seems so.. but I've restarted builds multiple times.. so this is not clean case.. Dec 01 10:10:05 bluelightning: ok I'll give bb from git a try Dec 01 10:10:54 bluelightning: I'll try to look on git fetcher this evening and check opie-recipes fetching Dec 01 10:11:00 should go now Dec 01 10:11:06 Jay7: ok cool, thanks Dec 01 10:11:56 bluelightning: you can check results on tinderbox under Jay7-tb Dec 01 10:13:02 * Jay7 -> afk Dec 01 10:27:29 ant_work: I'm here, was afk for the last hour, but pidgin reminded me about your chat Dec 01 10:29:18 libsegfault seems very odd to have multiple providers, what recipes does it mention Dec 01 10:34:54 iirc was only building uclibc..I have the logs at home Dec 01 10:35:05 not on tinderbox? Dec 01 10:35:19 ah, no, ..too many failures so I disabled it Dec 01 10:35:27 (ashamed ;) Dec 01 10:35:51 btw those logs won't appear in tinderbox Dec 01 10:36:01 ah ok Dec 01 10:36:07 I tee the output on screen to a file, thougg Dec 01 10:36:19 I should have all the logs Dec 01 10:36:55 fwiw minimal has a bluez-libs conflicts Dec 01 10:37:51 I'll put all those notes together this evening (angstrom + minimal + various libc) Dec 01 10:37:57 khem, (for when you're back) wrt php: it seems I made a mistake in the last patch, it should be --without-iconv not --disable-iconv Dec 01 10:38:44 khem, net effect is that the current recipe tries to pick up iconv, but configure fails; tinderbox log: http://tinderbox.openembedded.net/public/logs/task/12098699.txt Dec 01 10:39:17 khem, also worried about the large # of warnings from autoconf Dec 01 10:40:23 no idea how to fix this (ofc I can fix the disabling) Dec 01 10:41:00 ant_work: minimal has conflicts with bluez (or maybe bluezlibs), mysqlclient and depmod Dec 01 10:54:47 03Martin Jansa 07master * r7cbc1459f2 10openembedded.git/recipes/shr/ (24 files): (log message trimmed) Dec 01 10:54:47 shr: add more PACKAGE_ARCH = all for theme and e-wm recipes Dec 01 10:54:47 * shr-splash-theme: convert to INC_PR Dec 01 10:54:47 * other recipes just bump PR Dec 01 10:54:47 * because changes of PACKAGE_ARCH doesn't work very well you have to run: Dec 01 10:54:57 03Martin Jansa 07master * ra8e0c9bb43 10openembedded.git/recipes/ (15 files in 2 dirs): (log message trimmed) Dec 01 10:54:57 shr: remove EFL_SRCREV from PV Dec 01 10:54:57 * with comming EFL release it doesn't make much sense, and would be Dec 01 10:54:57 misleading if later built against efl release Dec 01 10:54:57 * in case of edje-cc backward incompatible change we can bump PR like with other recipes Dec 01 10:56:26 hi, I've that for fennec_hg.bb: | make[5]: *** No rule to make target `-lgconf-2', needed by `libmozgnome.so'. Stop. any idea? Dec 01 10:56:40 I grepped the recipe for lgconf-2 Dec 01 10:56:42 no matches Dec 01 10:56:54 hi all, i'm trying to build base-image for beagleboard (angstrom kernel), and getting an error: http://pastebin.com/dYxyM6Fy could someone help me troubleshoot it for me please? Dec 01 10:56:59 and mozilla stuff isn't easy to cross-compile Dec 01 10:57:58 andyzammy, what is ulinux? Dec 01 10:58:48 tbh i'm not sure Dec 01 10:59:25 bluelightning: I installed bb from origin/master, but it just exits without a sign on any command Dec 01 10:59:39 I switched back to 1.8.18 and it works Dec 01 11:03:44 how can i find out what ulinux is? Dec 01 11:07:33 andyzammy, grep? Dec 01 11:07:54 in the workdir of the package Dec 01 11:20:46 metiu: hmm, not sure sorry... I haven't tried it Dec 01 11:21:06 bluelightning: np, 1.8.19 works perfectly :) Dec 01 11:21:18 you may find 1.10 is a little faster at parsing and is fairly well tested Dec 01 11:21:47 I'll try it, now I'm making sure I have all the prereqs Dec 01 11:22:00 thank you Dec 01 11:24:09 np Dec 01 11:28:08 GNUtoo|laptop: grep is still chugging away, but here is a snippet of it: http://pastebin.com/DJxwY2JZ Dec 01 11:31:08 andyzammy, no idea, maybe | grep -v gnulinux to filter out all theses gnulinux results Dec 01 11:43:22 GNUtoo|laptop: this is what it's bringing up so far: http://pastebin.com/T5HqTrVi Dec 01 11:44:08 (CMAKE_SYSTEM "ulinux") Dec 01 11:44:25 maybe try to see why it's setup that way, and change it Dec 01 11:46:11 i've no idea how to do that, i'm a noob at this Dec 01 11:47:52 i might be starting a holy war here but wouldn't the makefile way be a lot simpler than this lol? OE has brought me nothing but confusion so far and the recipie thing is not paying off at all Dec 01 12:22:15 XorA: hi Dec 01 12:23:17 XorA: you don't want to change toolchains for already released distro, but you're breaking it anyway http://cgit.openembedded.org/cgit.cgi/openembedded/commit/?id=041fe58428d00d336a4264e0f91ae58a3730f9a7 Dec 01 12:23:25 XorA: I'm quite lost here :) Dec 01 12:49:36 hi mickey|office, zecke Dec 01 12:50:29 he pb_______ Dec 01 12:50:36 hi woglinde__________ Dec 01 12:50:43 *g* Dec 01 12:53:37 hi hrw Dec 01 12:54:08 hello woglinde Dec 01 12:54:42 hi ant Dec 01 13:45:56 eFfeM_work: you were talking about this? NOTE: multiple providers are available for runtime module-init-tools-depmod (module-init-tools, module-init-tools-cross) Dec 01 13:46:10 yes that is one of them Dec 01 13:46:22 eFfeM_work: please notice that those are runtime providers not build-time providers (which are easy to fix) Dec 01 13:46:40 eFfeM_work: and this runtime are reported only with bitbake master not with 1.10 and older Dec 01 13:46:46 hi effem hi jama Dec 01 13:46:50 hi woglinde Dec 01 13:46:55 there is also one for bluez-libs and for mysclient Dec 01 13:46:58 eFfeM_work: http://git.kernel.org/?p=fs/ext2/e2fsprogs.git;a=tags ... it means nobody has tested builds last two years !?! Dec 01 13:46:59 woglinde: hi Dec 01 13:47:30 JaMa: I feel we should be able to compile using bb 1.10.1 too Dec 01 13:47:47 eFfeM_work: I know I have 15 of those even after blacklisting many of those in shr.conf Dec 01 13:48:14 JaMa: and if they are runtime providers then I'd say module-init-cross should not provide it, I do not expect a package named -cross to deliver target stuff Dec 01 13:48:29 eFfeM_work: master says 1.11.0 and those errors are not fatal (or are they for you? Dec 01 13:50:00 JaMa: I don't think they are fatal but they are at least confusing and it is also not too clear which one is chosen (at least not until you actually get to building it) Dec 01 13:51:30 ant_work: not sure what you mean with tested builds, oe has only one recipe for e2fsprogs which is for 1.41.9 Dec 01 13:51:43 not sure if I have actually build that one recently, check tinderbox Dec 01 13:56:23 eFfeM_work: here is example how to fix those http://www.mail-archive.com/openembedded-devel@lists.openembedded.org/msg07433.html Dec 01 14:01:49 eFfeM_work: another patch for module-init sent to ML Dec 01 14:03:03 JaMa: still reading your fix, got interrupt, patch to ML is good; love to see also a patch for bluez-libs and mysqlclient (actually still not sure why we want to keep that old stuff anyway) Dec 01 14:04:39 JaMa: saw your matchbox-keyboard-inputmethod, seems a good solution Dec 01 14:04:52 but for bluez the problem is that we have conflicts between two versions of the libs Dec 01 14:05:37 ie gst-plugins will be much harder to fix imho and as this warning has really low priority for me (similar to QA Note) I wouln't care if it's in release tag or no - not many people will use release tag + bleeding edge bitbake master Dec 01 14:05:39 oops, sorry, no Dec 01 14:06:28 it is bluez4 conflicting with bluez-utils_3 (but we do not seem to have bluez recipes any more (which was bluez3) Dec 01 14:06:33 but you're right "NOTE: consider defining a PREFERRED_PROVIDER entry to match gpsd" is really confusing as this won't help at all Dec 01 14:06:51 maybe an RCONFLICTS could solve the bluez thing Dec 01 14:07:06 JaMa: agree Dec 01 14:07:12 no it will say the same even with RCONFLICTS set Dec 01 14:07:26 ie runtime e-wm-sysactions (e-wm, e-wm-sysactions-shr) has it Dec 01 14:07:48 even with RCONFLICTS and pinning? Dec 01 14:08:00 but e-wm needs to be built for shr-image and only as subpackage is e-wm-sysactions created too (but not used later in image) Dec 01 14:08:58 so multiple runtime providers are pulled to image but only e-wm-sysactions-shr used in the end Dec 01 14:09:20 eFfeM_work: bitbake itself doesn't do anything with rconflicts, only opkg cares about that field. Dec 01 14:09:46 pb__ ah ok Dec 01 14:09:49 and, even opkg only cares about it to the extent of refusing to install two packages which conflict. Dec 01 14:10:11 of course (coudl have figured that myself as they are R vars) Dec 01 14:10:47 heh. well, of course, bitbake _does_ care about RDEPENDS/RPROVIDES, so you could have been forgiven for expecting it to look at RCONFLICTS too Dec 01 14:10:58 but, as it happens, it doesn't. Dec 01 14:12:09 ok, thanks for the explanation! Dec 01 14:12:23 pb__: but even if with bitbake lookin at RCONFLICTS then shouldn't refuse to build e-wm when there is PACKAGES += "e-wm-sysactions" and image wants e-wm-sysactions-shr (which RCONFLICTS with it) Dec 01 14:12:59 JaMa: yeah, indeed. I'm certainly not saying that I think bitbake ought to look at RCONFLICTS: as you mention, it's hard to imagine what it could usefully do with that var. Dec 01 14:14:27 the only point I was trying to make was that, since bitbake doesn't use that variable at all, setting it (to any value) is certainly not going to help matters. Dec 01 14:15:06 I don't think there is any good way of eliminating the diagnostic that you're getting at the moment. This is one of those corner cases where having two runtime providers for the same thing really is correct behaviour. Dec 01 14:15:33 ok, imho the warning should be written only if resulting .ipk will have same filename (overwritten in feed) Dec 01 14:16:06 but no idea how could bitbake check that while building run queue (for DYNAMIC_PACKAGES and debian naming) Dec 01 14:16:31 yeah, maybe. Dec 01 14:17:02 that would make all my multiple provider cases go away :) Dec 01 14:17:03 is e-wm-sysactions a semivirtual? that has historically been a sticky area with bitbake and opkg, though I guess it is perhaps not directly related to the situation here. Dec 01 14:17:30 I don't know what you mean by semivirtual Dec 01 14:17:41 wrt module-init-tools-cross_3.2.2.bb should't that one actually be module-init-tools-native ???? Dec 01 14:17:53 jama: a package that both exists concretely and as an RPROVIDES. Dec 01 14:18:17 pure virtual is a package that only exists as RPROVIDES, concrete is a package that exists as an actual .ipk, and semivirtual is a package that exists both ways. Dec 01 14:18:43 the difficulty with semivirtuals is that there is no good way to tell opkg that you want to install the concrete one, because the Provides: tends to always win. Dec 01 14:18:47 pb__ not sure whether it is good to have two runtime providers providing the same thing, e.g. for bluez I highly doubt taht bluez4 and bluez-libs_3* provide a same lib with the same api etc Dec 01 14:19:07 pb__: then yes Dec 01 14:19:10 eFfeM_work: the library autonamer ought to take care of that particular case. Dec 01 14:19:35 pb__ somehow it does not, not for bluez, not for libmysqlclient Dec 01 14:19:53 ah. that is probably a bug then. Dec 01 14:20:01 pb__: e-wm_svn.bb has it in PACKAGES as ${PN}-sysactions and e-wm-sysactions-shr_git.bb has RPROVIDES_${PN} = "e-wm-sysactions" RCONFLICTS_${PN} = "e-wm-sysactions" Dec 01 14:20:31 pb__: JaMa had some more of these, not sure if these are libs too Dec 01 14:20:41 JaMa: right, yeah, that's the situation. so the general problem is that, if you have e-wm-sysactions-shr in your package feed, there is no way to tell opkg that you want to install the version from e-wm. Dec 01 14:20:51 of course, that might be fine in your situation, but it isn't a good arrangement in general. Dec 01 14:21:14 * pb__ lunchtime now Dec 01 14:21:31 mickey|office, zecke, RP: are we due to have a meeting at 1600 today? Dec 01 14:22:26 eFfeM_work: here is my current list http://paste.pocoo.org/show/298472/ Dec 01 14:23:31 JaMa: hm somehow surprised by a few of these Dec 01 14:23:34 pb can the tsc still discuss a topic? Dec 01 14:24:31 pb__: yes in e-wm-config which is choosen by shr.conf:ILLUME_CONFIG = "e-wm-config-illume2-shr" there is explicit RRECOMMENDS for e-wm-sysactions-shr and the version from e-wm is "overriden" by that Dec 01 14:25:34 eFfeM_work: back to e2fsprogs, probably my last angstrom-uclibc build was done in 2008... Dec 01 14:26:00 I know I should use some now variable like E_WM_SYSCONFIG in e-wm RRECOMMENDS and e-wm-config-illume2-shr to RRECOMMEND e-wm-sysactions-shr now without RPROVIDES directly Dec 01 14:26:10 woglinde: sure Dec 01 14:26:11 s/now/new/ Dec 01 14:26:24 lunchtime now, really Dec 01 14:26:30 good apetite Dec 01 14:26:31 haven't done uclibc builds recently myself, I seem to recall somehone here mentioning that cairo or pango or so does not get along well with uclibc and that hence we moved to eglibc Dec 01 14:26:37 pb__: enjoy your lunch! Dec 01 14:27:41 http://lists.linuxtogo.org/pipermail/openembedded-devel/2008-December/007379.html Dec 01 14:27:55 pb__ new developer accounts Dec 01 14:28:01 seems there is a stuck Dec 01 14:28:12 eFfeM_work: december 2008, after "ANGSTROM_MODE renamed to ANGSTROMLIBC " Dec 01 14:28:19 pb__ maybee its not tsc topic Dec 01 14:28:44 effem building or running? Dec 01 14:28:58 Anyone around that can unlock http://wiki.openembedded.org/index.php/User_contributions ? Dec 01 14:29:20 I'd really like to make it redirect to http://wiki.openembedded.org/index.php/How_to_submit_a_patch_to_OpenEmbedded Dec 01 14:29:39 Hi. I have successfully built webkit-gtk for panda board. and it resulted in lot of ipkg files.. How to flash these packages onto to the board? Dec 01 14:29:45 ant_work: ah ok Dec 01 14:30:14 i might need to peek into uclibc again as I mgiht need to reduce an image but will probably try a few other things first Dec 01 14:30:34 eh..reduce..how much do youexpect? Dec 01 14:30:45 effem use git version Dec 01 14:30:46 I've seen 5-7% diffs, no more Dec 01 14:30:52 git version has nptl for arm Dec 01 14:30:58 xc0ffee: just copy them to some storage you can mount on the panda or use scp. you can install them with opkg on the board Dec 01 14:31:15 hi florian Dec 01 14:31:26 hi woglinde Dec 01 14:31:30 florian: Thank.. I will try it out.. Dec 01 14:31:45 woglinde: we stumbled in an old patch of you...disabling tls for oabi back then Dec 01 14:31:57 hm Dec 01 14:31:57 wog, will check Dec 01 14:32:06 ant_work: 5-7 % compared to eglibc Dec 01 14:32:16 btw that was woglinde, will check Dec 01 14:32:18 well, eglibc size > glibc :/ Dec 01 14:32:25 hopefully next year I will have more time again for oe Dec 01 14:32:32 ant_work: is it ? Dec 01 14:32:49 hm, I expected eglibc to be smaller Dec 01 14:32:52 second.. Dec 01 14:33:31 florian: There are some shared libraries which webkit.ipkg depends on... These shared libraries need to be copied and installed separately I guess.. Right? Is there a way in which I can copy and install all the webkit and its dependent ipkg in a single go? Dec 01 14:33:49 xc0ffee yes Dec 01 14:34:09 eFfeM_work: http://pastebin.ca/2006352 Dec 01 14:34:16 eFfeM_work eglibc is building faster Dec 01 14:34:28 woglinde: Thanks.. Dec 01 14:34:35 xc0ffee provide a local opkg archive Dec 01 14:34:40 via http Dec 01 14:34:52 and add it to /etc/opkg stuff Dec 01 14:35:10 woglinde: the issue is angstrom has an old uclibc without nptl/tls Dec 01 14:35:10 you can use the filesystem structure oe is building Dec 01 14:35:26 ant change it Dec 01 14:35:34 and recent e2fsprogs have the deps Dec 01 14:35:45 woglinde: true that it is faster Dec 01 14:36:13 woglinde: cool.. I am doing it for the first.. need to explore it more.. thanks for the tip.. pls let me know if this procedure is documented somewhere Dec 01 14:36:24 eFfeM_work: calculate there is a 1,2-1,3 Mb kernel in those rootfs's Dec 01 14:36:33 and modules Dec 01 14:36:57 xc0ffee hm I think its not really documented Dec 01 14:37:25 anybody working on glib-2.26 ? Dec 01 14:37:37 we (shr) would very much want gdbus :-) Dec 01 14:37:39 ant_work: odd, why is for some images eglibc smaller and for some bigger ? I would expect always to have the same one win, as e.g. eglibc.so < glibc.so (or the other way around). unless it has to do with static build executables Dec 01 14:37:40 woglinde: No probs.. I will find a way.. thanks.. Dec 01 14:38:11 ant_work: have you compared the rfs of a case where glibc < eglibc? Dec 01 14:38:20 xc0ffee explore the deploy dir Dec 01 14:39:26 eFfeM_work: eglibc < glibc happens with minimal distro Dec 01 14:39:30 only Dec 01 14:39:36 woglinde: Ok.. Dec 01 14:39:39 but why? Dec 01 14:40:18 why is for angstrom glibc < eglibc ? Dec 01 14:40:25 and for minimal teh other way around Dec 01 14:40:48 (btw as I am using minimal/eglibc it seems it is the good choice) Dec 01 14:40:52 I'll decompress two images and do side-by-side check Dec 01 14:43:10 ant_work: would be great Dec 01 14:57:48 isn't testlab suited for that? Dec 01 14:58:15 I mean size comparisons Dec 01 14:58:28 (but never used it :)) Dec 01 14:59:08 pb__: i think so. no idea whether we have something to talk about, but we should at least check Dec 01 15:04:42 ynezz: to some extend Dec 01 15:05:04 I have a script somewhere Dec 01 15:06:25 I hacked this one http://www.mail-archive.com/angstrom-distro-devel@linuxtogo.org/msg02595.html Dec 01 15:10:42 wow, eFfeM_work, take a look ..Angstrom did inflate a bit Dec 01 15:11:47 console-image.rootfs.tar.gz was 8975689 now is 11292582 Dec 01 15:14:11 hi Dec 01 15:14:29 7am in ka6sox timezone. too early to bother him to install Extension:Nuke Dec 01 15:14:38 oh..the spammers... Dec 01 15:18:20 ~lart meego for only providing live-images Dec 01 15:18:20 * ibot eats meego and falls over dead for only providing live-images Dec 01 15:18:37 When I try to build a webkit-gtk package, I get "NOTE: Not creating empty archive for webkit-gtklauncher", hence no ipkg files get created for this package.. How to solve this? Dec 01 15:43:42 * mckoan is unpacking his new Archos 5 IT Dec 01 15:44:42 hehe Dec 01 15:44:51 mckoan: enjoy :) Dec 01 15:44:52 install android 2.2 hacked image Dec 01 15:45:11 woglinde: any link? Dec 01 15:46:59 no sorry Dec 01 15:49:24 woglinde: np Dec 01 15:51:18 jo zecke Dec 01 16:54:39 what is the method for getting patches committed that have not been ack/nak'd? I've posted several new recipes in the past couple of weeks and a fix for avrdude parallel make (that one at least should get into release)? Dec 01 16:58:36 tharvey: stable branch have a strict ack policy Dec 01 16:59:10 I see lots of patches that never get ack'd that are good patches Dec 01 16:59:20 tharvey: critical recipes modifications must be ack-ed by someone Dec 01 16:59:48 tharvey: not critical may be autonomously pushed once you have RW access Dec 01 17:00:22 tharvey: there are a few people who decide which one are good patches ;-) Dec 01 17:00:31 why should RW access be required for a casual user who has some patches to submit? Dec 01 17:00:49 tharvey: solecite ack on ML Dec 01 17:01:26 can anyone look at http://patchwork.openembedded.org/patch/3558/ Dec 01 17:01:47 also http://thread.gmane.org/gmane.comp.handhelds.openembedded/40127 Dec 01 17:01:47 tharvey: I meant urge ack on ML Dec 01 17:02:21 shouldn't posting the patch be enough? if not, whats the rule of thumb for responding to your own e-mail patch urging for response? 1 week? Dec 01 17:02:38 tharvey: TI stuff are a hard story, please ask on ML Dec 01 17:03:25 tharvey: disable parallel make for avrdude recipes Dec 01 17:03:28 why Dec 01 17:04:17 because it fails occasionally on parallel builds Dec 01 17:04:24 have a log right now I can post Dec 01 17:05:06 in fact a simple bitbake -cclean -f avrdude && bitbake avrdude in a loop about 5 times reproduces a build failure for me without PARALLEL_MAKE="" Dec 01 17:05:32 heh, avrdude doesn't even fetch for me Dec 01 17:06:27 have a nice rest of the day Dec 01 17:06:47 http://pastebin.com/mxJhmYsq - avrdude make -j8 failure Dec 01 17:08:03 does it help if you add an explicit dependency of config.c on config_gram.h? Dec 01 17:08:48 pb__, fetched fine for me, just tested after wiping my source Dec 01 17:08:59 if so then that might be slightly better than just defeating PARALLEL_MAKE. Dec 01 17:09:03 tharvey: hm. what url did it fetch from? Dec 01 17:09:13 I get 404 for http://savannah.nongnu.org/download/avrdude/avrdude-5.10.tar.gz which appears to be the primary Dec 01 17:09:26 pb__, don't know... didn't really dig that deep into the issue. I figure there are lots of packages that have makefile issues that have been solved with PARALLEL_MAKE="" - not clear if its OE's job to fix that or work around it Dec 01 17:10:25 your right... I'm picking it up from a mirror Dec 01 17:10:35 yeah, it's true there are plenty of packages which have PARALLEL_MAKE="", and that's a reasonable workaround if the actual bug is too hard to fix. but, obviously, if you end up disabling parallel make for too many packages then you might as well just turn it off globally. Dec 01 17:10:54 so, if the fix is straightforward, generally better to patch the makefile and enjoy faster builds. Dec 01 17:11:00 http://www.angstrom-distribution.org/unstable/sources/avrdude-5.10.tar.gz Dec 01 17:11:04 ah Dec 01 17:12:04 http://download.savannah.gnu.org/releases/avrdude/avrdude-5.10.tar.gz Dec 01 17:12:08 seems to be the new url Dec 01 17:12:18 I don't consider myself an expert on avrdude or even gmake - I figure if PARALLEL_MAKE="" is a solution then thats good enough. Dec 01 17:13:08 there are 251 PARALLEL_MAKE="" 's in my oe tree Dec 01 17:25:02 pb__, avrdude uses autotools and config_gram.h comes from lexing config_gram.y - not really sure how to fix the dep issue Dec 01 17:25:35 tharvey: try adding a line to Makefile.am which just reads Dec 01 17:25:40 config.c: config_gram.h Dec 01 17:32:16 pb__, testing - I can post a revised patch that adds a make.patch to fix Makefile.am as well as fix the SRC_URI - hopefully you can review/ack/commit? Dec 01 17:32:52 yup Dec 01 17:34:36 who can review ti recipes? looks like the major contributors don't hang out on this channel Dec 01 17:35:26 pb__, is it ok to put the patch in avrdude.inc considering it patches Makefile.am and there is only one versioned recipe? (or should it be in avrdude_5.10.bb Dec 01 17:35:38 tharvey: I guess putting it in the .inc is fine Dec 01 17:35:46 if/when avrdude fixes that upstream we can always revisit it Dec 01 17:36:04 if they fix it the patch can be removed Dec 01 17:37:52 well, clearly they can't fix it in historical versions. if 5.11 fixes it then we can move the patch to the 5.10 recipe, and then delete it if/when 5.10 is removed from the oe tree. Dec 01 17:38:06 avrdude doesn't seem to be the most active package, though, so I would not hold my breath for a new release Dec 01 17:39:37 ah, I see your point Dec 01 17:50:41 pb__, ok patch sent - thx for assistance Dec 01 17:51:56 ok, cool Dec 01 17:52:17 that should be cherry-picked to release-2010.12 too Dec 01 18:15:12 ant, pong Dec 01 18:16:02 hi Dec 01 18:26:38 he blindvt Dec 01 18:26:46 woglinde, hi Dec 01 18:27:01 blindvt do you have some sparetime? Dec 01 18:27:16 for oe Dec 01 18:27:36 woglinde, hmz. I was about to update the busybox docs and then saw http://validator.w3.org/check?uri=http%3A%2F%2Fbusybox.net%2Fdownloads%2FBusyBox.html&charset=%28detect+automatically%29&doctype=Inline&group=0 Dec 01 18:28:12 woglinde, wonder who ruined pod2html _that_ badly :( Dec 01 18:28:19 dont know Dec 01 18:28:31 hm Dec 01 18:28:41 ah will take took much time now Dec 01 18:29:00 I am thinking about how to simplify the uclibc config in oe Dec 01 18:29:02 woglinde, but yea, i ment to fix 2 uClibc bugs and then roll an 0.9.32-rc1, so would have to build current oe. shoot Dec 01 18:29:16 yeah .32 Dec 01 18:29:18 yeah yeah Dec 01 18:29:24 christmas presents Dec 01 18:29:43 okay have to leave now Dec 01 18:29:56 bye till maybee later Dec 01 18:30:09 woglinde, in what way simplify it? I was under the impression that the uclibc-config stuff is pretty easy and obvious to use, no? Dec 01 18:30:31 bummer Dec 01 18:30:38 * blindvt waves Dec 01 18:32:21 kk. so what was /me doing? ah, prettify ext4 image with -pn_e2fsprogs like i was told instead of looking at packaged-staging directly. So be it Dec 01 18:48:10 gm Dec 01 18:56:13 JaMa: ping Dec 01 18:56:38 I saw your patch. Why not include the changes in python bindings to use autotools? Dec 01 19:09:01 JaMa: nevermind Dec 01 19:15:42 crofton you here? Dec 01 19:16:34 yeah Dec 01 19:16:49 bad memory in laptop so it may fall over Dec 01 19:30:09 is some patchwork deity around? Dec 01 19:31:26