**** BEGIN LOGGING AT Sun Nov 12 02:59:57 2006 Nov 12 03:19:41 * raduga is cold :/ Nov 12 03:21:23 * raduga applied sashz' pxa270 frequency scaling patch, to OZ kernel Nov 12 03:21:26 it sortof works Nov 12 03:21:34 at least, i think it works :/ Nov 12 03:22:34 * raduga was contemplating sending this on to the bugtracker, but until i can resolve the 'sortof' probably not a useful idea Nov 12 08:45:01 03rwhitby 07org.oe.dev * r4e4a2d2a... 10/ (6 files in 4 dirs): Nov 12 08:45:01 slugos: Changed SLUGOS_FLASH_IMAGE to be multi-valued, to allow for multiple Nov 12 08:45:01 target upgrade image formats. Changed the image name, so that it overwrites the Nov 12 08:45:01 previously built image (and doesn't fill the autobuilder disks). Nov 12 08:59:30 anyone of you know an autofoo-wizard? Nov 12 08:59:47 e.g. something that asks me for source files and dependencies, then creates all autofoo infrastructure files Nov 12 09:00:56 don't some IDEs do that for you? Nov 12 09:02:10 probably, but I thought of something clean Nov 12 09:02:15 i.e. something that i can understand afterwards Nov 12 09:04:27 I heard something about gnome-common Nov 12 09:11:07 goodmorning Nov 12 09:11:29 hey Ifaistos Nov 12 09:11:33 kergoth: 2006-11-12 10:35:25 1Gj1Xb-0003Hn-4L kergoth.com [207.210.106.158] No route to host Nov 12 09:13:08 seem perl fail do configure if perl-native is 'assuped-provided' ... cause it try to use staging//bin/perl !!! Nov 12 09:44:10 a morning that start with a denial of service if a bad morning Nov 12 09:44:31 good morning all Nov 12 09:44:34 cyrilRomain: Or a blessing in disguise, if you can use it as an excuse for something else not working :) Nov 12 09:45:58 NAbyss: yes, but sadly not here :( Nov 12 10:04:35 morning all Nov 12 10:04:43 hi RP Nov 12 10:09:12 hey RP Nov 12 10:09:12 RP : Hi ! Nov 12 10:14:35 Hey RP Nov 12 10:15:16 RP: Poky would benifit from this: http://www.openembedded.org/viewmtn/diff.psp?id1=b1116e7c5f000ce7ebb2b39f584817d9243a5fb4&id2=e828cc8219ce3a989d3bbf7e05247cd303ff8cf2&fname=packages/sysvinit/sysvinit/angstrom/rc Nov 12 10:15:57 koen: It fixes the divide by zero on shutdown? Nov 12 10:16:28 RP: yes, but it sets random values '(10' and '1'), so you don't get a proper shutdown splash Nov 12 10:16:46 but at least the machine doesn't ask for the maintenance password :) Nov 12 10:17:03 koen: We should really fix it properly :) Nov 12 10:17:52 yeah Nov 12 10:18:04 * koen wonders about the dbus waitpid() problem Nov 12 10:18:15 i notice libxml-parser-perl-native-2.34-r2 fail to build cause of a space in the make parameters ... Nov 12 10:18:51 where are the riles to generate oe_runmake() ?? Nov 12 10:19:31 it generate >>make -e MAKEFLAGS= "$@" || die "oe_runmake failed"<< tha space after MAKEFLAGS= is the problem :( Nov 12 10:19:45 s/riles/rules/ Nov 12 10:20:41 gremlin[it]: patches accepted Nov 12 10:21:26 koen which patch ? Nov 12 10:21:40 for all the problems you are talking about Nov 12 10:22:26 gremlin[it]: Were you seeing it with something using qmake? Nov 12 10:23:11 the package is libxml-parser-perl-native-2.34-r2 ... part of perl ... Nov 12 10:23:33 gremlin[it]: ignore me then ;-) Nov 12 10:24:42 but you are true .. the only reference to MAKEFLAGS= and a space is in qmake ... but seem perl use something different ... Nov 12 10:25:15 koen : i just update oe ... but problem persist ... is the patch accepted and also applied ? Nov 12 10:25:51 gremlin[it]: I think koen was suggesting you were welcome to provide a patch :) Nov 12 10:26:13 ahh ... Nov 12 10:26:39 ok but i need some clues about where oe_runmake() is defined :) ... Nov 12 10:27:33 gremlin[it]: base.bbclass Nov 12 10:28:15 gremlin[it]: maybe it comes from the leading space in the EXTRA_CPANFLAGS of libxml-parser-perl-native_2.34.bb ? Nov 12 10:29:04 no cause in base.bbclass i found >>>${MAKE} ${EXTRA_OEMAKE} "$@"<<< Nov 12 10:29:57 probabbly that solve for perl .. but don't know if create problem for other packages ... let's see :) Nov 12 10:30:08 gremlin[it]: yep Nov 12 10:30:11 zecke? Nov 12 10:45:51 i have a completelly off-topic question ... a UK company asked me for a dettailed resume ... but i really don't know how to write and a bit what to write ... how in depth i have to list my previous jobs and my rules and what i produce ... any suggestion ? Nov 12 10:49:39 gremlin[it]: Google for "writing a CV" (Curriculum Vitae) Nov 12 10:50:12 i did ... but found what i already have ... nothing about 'dettailed CV' ... bha :) ... Nov 12 10:50:20 ok lunch time ... see u later Nov 12 10:57:12 when did the changes occur that dropped the libqte2 package? Nov 12 11:04:13 did anyone of you guys use autoproject before? Nov 12 11:04:36 (autotools skeleton creator) Nov 12 11:06:21 mickeyl: I've always just pinched conf code from other projects... Nov 12 11:06:48 RP: i see. that's what I'm afraid of... Nov 12 11:07:13 no one create autofoo from scratch Nov 12 11:07:19 I've not written that many projects from scratch ... Nov 12 11:07:28 it all stems from the same autofoo config file from the 70's ;) Nov 12 11:07:49 ohand autofoo stuff looks clean Nov 12 11:07:49 concise Nov 12 11:08:14 It was ohand code I pinched from :) Nov 12 11:08:37 *nod* i only need things like checking for pkgconfig libraries. nothing fancy Nov 12 11:09:18 then again, i could just continue using qmake... Nov 12 11:09:44 please, think of the developers Nov 12 11:09:50 :) Nov 12 11:10:10 you can see from maemo how hard people find it to install an sdk Nov 12 11:10:21 If you just need pkgconfig, you could probably write a skeleton genertor for your needs easily enough Nov 12 11:10:25 RP: linux-openzaurus-2.6.17-r30 patch fails for tosa (the 2.6.17-r28 worked though). Here is the patch: http://www.rafb.net/paste/results/QpbgMZ46.html (the two patches used are the same than in linux-openzaurus-2.6.18+git/) Nov 12 11:10:25 "but the phone runs gtk, why do I need qt?" Nov 12 11:11:13 cyrilRomain: ok, thanks - I'll look at that shortly Nov 12 11:17:29 well, sooner or later the phone should run both gtk and qt Nov 12 11:17:31 but i agree Nov 12 11:17:35 autoconf is a must Nov 12 11:17:52 it's just that I successfully went around without learning it Nov 12 11:17:56 for all those years now Nov 12 11:17:56 The bitbake fetcher codemake me feel ill :-( Nov 12 11:18:08 * mickeyl hands RP some pills against sickenss Nov 12 11:18:11 sickness, even Nov 12 11:18:28 mickeyl: I had the same problem as until I wrote zaurusd simple makefiles were fine :) Nov 12 11:18:58 The fetchers can get called at random places in the code and have no way to save state :-/ Nov 12 11:19:24 nasty :/ Nov 12 11:22:44 RP: is it OK to check in the current zaurusd code into the devmand repo? Nov 12 11:23:10 koen: Please wait for now :-/ Nov 12 11:23:37 RP: I don't have write access, so I would have delegated it anyway :) Nov 12 11:37:27 ~lart waitpid() Nov 12 11:37:28 * ibot executes killall -KILL waitpid() Nov 12 11:45:43 * koen spots activity on http://bugzilla.gnome.org/show_bug.cgi?id=350973 Nov 12 11:45:55 ~ugt Nov 12 11:46:02 ugt is probably Universial Greeting Time. Created in #mipslinux, it is a rule that states that whenever somebody enters an IRC channel it is always morning, and it is always late when the person leaves. The local time of any other people in the channel, including the greeter, is irrelevant. http://www.total-knowledge.com/~ilya/mips/ugt.html Nov 12 11:54:10 Hi koen and cyril, I fixed all problem on my bitbake package. There is time to send it do bugtrack Nov 12 12:05:20 acassis: :) Nov 12 12:05:30 Thanks for you! Nov 12 12:06:13 cyrilRomain: do you already submit some package to OE? Nov 12 12:06:37 acassis: no. I even never write a .bb file myself Nov 12 12:07:23 sure, koen said me I just need send it to bugtrack Nov 12 12:08:39 acassis: I tried to create one for ocaml, but it needs bootstraping (and thus a deep understanding about how .bb and how ocaml compile, which I do not have) Nov 12 12:09:05 hmm Nov 12 12:09:10 autoconf is not _that_ bad Nov 12 12:09:20 it kind of works Nov 12 12:09:21 * mickeyl surprised Nov 12 12:09:46 cyrilRomain: ok, I think ipatience was really simple to create the .bb Nov 12 12:09:56 cyrilRomain: there's a start at http://www.openembedded.org/filebrowser/org.openembedded.dev/packages/nonworking/ocamlc Nov 12 12:10:08 koen_: yep I saw Nov 12 12:10:51 koen_: maybe when I will have some more time :) Nov 12 12:12:55 RP: I was a bit enthousiastic: my patch is not enough to make it compile: asoc-v0.12.patch need to be updated too (IRQF_DISABLED seem to not exist anymore) Nov 12 12:13:08 koen: the bugtrack to submit the .bb is that: http://bugs.linuxtogo.org ? Nov 12 12:13:22 acassis: no Nov 12 12:13:35 acassis: the one in the topic Nov 12 12:13:37 cyrilRomain: asoc-v0.12 is probably diffed against much more recent kernels Nov 12 12:14:55 koen: I search on google: bugtrack openembedded but don't found it Nov 12 12:15:22 bugs.oe.org ? Nov 12 12:15:36 oe=openembedded Nov 12 12:15:37 RP: you're right, since the same patch is used for 2.6.18 and 2.6.18+git Nov 12 12:17:33 by the way what is the policy in OE for such case: when e.g. 2.6.18 compile for a machine but not 2.6.17 ? It is worth making 2.6.17 compiling ? Nov 12 12:18:00 cyrilRomain: While we still support it, yes Nov 12 12:18:17 RP: ok, I'll have a look then Nov 12 12:18:36 acassis: koen said the one in the channel topic Nov 12 12:19:30 cyrilRomain: what channel topic? sorry my english is no good Nov 12 12:19:48 acassis: each irc channel as a topic Nov 12 12:19:50 ~topic Nov 12 12:20:03 acassis: a sentence that present the channel Nov 12 12:20:05 http://bugs.openembedded.org/ Nov 12 12:20:20 type "/topic" into your irc client Nov 12 12:20:23 acassis: http://bugs.openembedded.org/ Nov 12 12:21:44 cyrilRomain: ok, then I need go to that channel? but I am using cgi:irc, is possible send my patch using cgi:irc? Nov 12 12:22:50 acassis: no, this is the topic of this channel (the #oe channel) Nov 12 12:23:02 ok, sorry my fault Nov 12 12:23:05 but maybe '/topic' does not work in your cgi:irc Nov 12 12:23:14 acassis: http://bugs.openembedded.org/ Nov 12 12:23:28 yes, I am there Nov 12 12:23:37 :) Nov 12 12:25:10 I am not IRC user, sorry :) Nov 12 12:28:55 cgi:irc sux Nov 12 12:29:10 RP: sorry to bother you, but ${RPSRC}/alsa/asoc-v0.12.patch is not in OE and 'RPSRC' let me guess you have the older version that patch successfully with 2.6.17 ;) Nov 12 12:29:43 RP: well, not a problem if you don't have time Nov 12 12:29:56 cyrilRomain: ${RPSRC} is defined in linux-openzaurus.inc Nov 12 12:34:03 cyrilRomain: I create this package using org.openembedded.dev to familiar, but I think it will run in others distros, can I place version unspecified instead Familiar Nov 12 12:37:45 acassis: sorry I'm not the one to ask for such thing. (you could indeed choose 'unspecified' if it is not a familiar specific package) Nov 12 12:40:14 sure, thank you Nov 12 12:54:34 hi guys, I submitted the ipatience package, please verify it at http://bugs.openembedded.org/show_bug.cgi?id=1574 Nov 12 12:56:56 OK, let me try to rephrase my earlier question and explain a bit of the background. I have successfully compiled jards with OE. Now I want to test the package on my collie. Apparently there was some major switch in libs about 1 or 2 weeks ago. Nov 12 12:57:04 Previously there was a package called libqte2 which apparently got replaced by something else. I would like to know when that change occured so I can compile just jards against the old tree instead of recompiling everything since the latest changes are not in the feeds for collie. Nov 12 12:58:06 Would compiling against oz354x instead of dev work? Have the changes made it into oz354x already? Nov 12 12:58:31 Laibsch: as answer to your last 2 questions: no and no Nov 12 12:58:49 koen: thanks for the answer. Nov 12 12:58:54 So what can I do now? Nov 12 13:01:47 hi all Nov 12 13:02:03 hi florian Nov 12 13:05:48 03acassis 07org.oe.dev * rd3403321... 10/ (6 files in 5 dirs): ipatience: add 0.1, closes #1574 Nov 12 13:10:03 hi everybody Nov 12 13:10:17 hey O_Neil Nov 12 13:10:45 hi O_Neil Nov 12 13:10:49 I thinks there is a bug with bitbake, Nov 12 13:11:08 it can't handle task properbly Nov 12 13:11:34 NOTE: package glib-2.0-native-2.6.5-r4: task do_unpack: completed Nov 12 13:11:44 NOTE: package glib-2.0-native-2.6.5: completed Nov 12 13:11:53 NOTE: package gtk-doc-native-1.0: started Nov 12 13:12:10 it did not compile glib Nov 12 13:12:14 O_Neil: looks ok to me Nov 12 13:12:24 multithreaded bitbake behaves that way Nov 12 13:12:46 I thinks it's only occured in bitbake 1.7.3, and 1.6.3 Nov 12 13:14:45 and bitbake sometimes build a package of 2 different versions twice Nov 12 13:19:12 O_Neil: It would only happen in 1.7.3, not 1.6.3 Nov 12 13:19:36 The 2 packages of different versions is a problem. Which package was it? Nov 12 13:20:13 last one occurs in 1.6.3 Nov 12 13:20:13 e2fsprogs Nov 12 13:20:16 I selected a preferred version of 32fsprogs Nov 12 13:20:40 and 1.6.3 does it too? :-/ Nov 12 13:20:44 yes Nov 12 13:21:00 maybe .. Nov 12 13:21:20 I'll check it, it was found by others Nov 12 13:21:28 That is rather worrying given the dependency code was totally rewritten between those versions... Nov 12 13:26:24 hi Nov 12 13:26:29 where is the tslib cvs now? Nov 12 13:26:43 it seems not to be on armlinux.couk anymore Nov 12 13:26:58 Marex: see the tslib .bb in org.openembedded.dev Nov 12 13:27:11 IOW, stop using outdated OE trees Nov 12 13:27:23 me ze neho asi jebne Nov 12 13:27:26 errr Nov 12 13:27:29 Marex: the host was down yesterday iirc Nov 12 13:27:35 koen, I'M NOT USING OE Nov 12 13:27:45 Marex: then stop complaining here Nov 12 13:27:47 koen, I've moved to something normal - buildroot Nov 12 13:27:52 Marex: THIS IS AN OE CHANNEL Nov 12 13:28:01 tslib moved to berlios Nov 12 13:28:01 koen, I WASNT COMPLAINING Nov 12 13:28:16 koen, IT WAS A QUESTION Nov 12 13:29:14 cyrilRomain, the host is ok ... tslib seems to have moved Nov 12 13:29:28 RP, oh, thanks :) Nov 12 13:30:20 people, someone here is using gpe-nmf? Nov 12 13:30:45 it getting a lot of problems with gpe-nmf Nov 12 13:31:03 it is not moving to next track Nov 12 13:31:52 I receive disk_source error every time the current track finish Nov 12 13:33:03 koen: Any hint on what I can do to test the jards package? Nov 12 13:33:16 Laibsch: no idea Nov 12 13:33:19 RP, got it - svn, thanx :) Nov 12 13:33:25 shit Nov 12 13:34:29 koen: But I guess I can just compile against an older tree of OE. If I just knew when the switch was made that dropped libqte2. "mtn log|grep libqte" did not tell me. Can you tell me (roughly)? Nov 12 13:35:02 Laibsch: afaik nothing was dropped Nov 12 13:35:38 Well, maybe not dropped but the dependencies and what is compiled against what seems to have changed. Nov 12 13:35:42 Just an example Nov 12 13:36:24 the default is multithreaded nowadays Nov 12 13:36:30 since months iirc Nov 12 13:36:37 possible that some bbs need to be adjusted Nov 12 13:37:31 previously there was a package called libjpeg62 (which also depended on libqte2). Now that apparently got replaced with the jpeg package. Jards depends on jpeg if I compile it against the latest tree. But libjpeg62 and jpeg cannot be installed at the same time. Nov 12 13:38:00 I was able to compile and use jards about two weeks ago (maybe three) Nov 12 13:38:16 libjpeg62 looks like a bug Nov 12 13:38:23 the libjpeg vs jpeg dilemma is the root cause of my trouble. Nov 12 13:38:49 koen: I think that was the default package in opie just recently. Let me check. Nov 12 13:40:57 hmmm Nov 12 13:41:00 let me see Nov 12 13:43:16 hi Nov 12 13:43:24 bye guys Nov 12 13:43:40 hey CoreDump|home Nov 12 13:43:58 CoreDump|home: your fsck code works great Nov 12 13:44:25 koen: That is good to know Nov 12 13:45:17 CoreDump|home: there's only one slight bug: you can't enter the root passwd if fsck corrected errors and halts the boot sequence Nov 12 13:45:24 (on a pda) Nov 12 13:45:30 a reset solves that Nov 12 13:45:51 koen: ohhh, the "missing keyboard" problem strikes again Nov 12 13:46:02 ok, that _is_ annoying Nov 12 13:46:20 I had to attach serial to see that Nov 12 13:46:34 the only indicator is the psplash 'hanging' Nov 12 13:47:02 right, have the same on OZ actually. A chvt 1 should fix it? Nov 12 13:47:18 no idea Nov 12 13:48:14 something is not right with the return codes, since it will also stop if it returns '1' (errors found and corrected) Nov 12 13:48:25 well, it fixes the problem on OZ at least since the bootsplash runs on VT2 Nov 12 13:48:29 but as I said, a reset solves it Nov 12 13:48:52 psplash always runs on a new VT (i.e. not 1) Nov 12 13:49:35 RP: OE shouldn't try to debianify RREPLACES Nov 12 13:49:51 Package: libjpeg62, Replaces: libjpeg62 Nov 12 13:50:03 the .bb has RREPLACES_${PN} = "jpeg" Nov 12 13:51:30 koen: if test "$?" -gt 1 Nov 12 13:51:30 then Nov 12 13:51:43 CoreDump|home: right Nov 12 13:51:46 it should not stop in returncode 1 Nov 12 13:52:19 it does stop Nov 12 13:52:27 which makes me wonder what the return code is Nov 12 13:52:41 koen: Hmm. That doesn't work unless you enforce that every possible combination of previous names is put into RREPLACES (i.e. with and without debian.bbclass for example) Nov 12 13:52:41 koen: can I specify a DEPENDS that require package with particular version? like DEPENDS = "glibc-2.4"? Nov 12 13:53:16 * CoreDump|home scratches his head Nov 12 13:54:27 03koen 07org.oe.dev * rd47d579d... 10/ (1 packages/jpeg/jpeg_6b.bb): Nov 12 13:54:27 jpeg: fix packaging Nov 12 13:54:27 * due to some bugs in shlib renaming we can't add RREPLACES, so packagers beware Nov 12 13:54:53 koen: could you prepare a test case to show this shlib renaming bug? Nov 12 13:54:57 * koen says 'hrm' and rebuilds packages to get the correct jpeg depends Nov 12 13:55:24 zecke: put RREPLACES_${PN} = "jpeg" in jpeg_6b.bb Nov 12 13:56:00 Laibsch: the jpeg problem should be gone now Nov 12 13:56:02 zecke: debian.bbclass alters the RREPLACES values as it does with the rest of the R* variables. Its both a feature and a bug whether it does it or not... Nov 12 13:56:36 koen: thanks a bunch. Nov 12 13:56:41 * Laibsch will recompile. Nov 12 13:57:26 koen: just for understandings sake: your fsck checked the rootfs and corrected errors, and after that it dropped you to sulogin? Nov 12 13:57:58 CoreDump|home: yes Nov 12 13:58:19 koen: that is not the case here Nov 12 13:58:35 CoreDump|home: I'll look into it later Nov 12 13:58:45 are you using busybox fsck by chance (if there even is such a beat) Nov 12 13:59:21 ehm Nov 12 13:59:49 no idea Nov 12 13:59:55 but that would explain a lot Nov 12 14:00:20 * CoreDump|home is using the real deal FWIW Nov 12 14:02:14 this machine already has bash and coreutils :) Nov 12 14:02:22 so a real fsck wouldn't hurt Nov 12 14:03:29 busybox is starting to annoy me Nov 12 14:03:40 no kidding =) Nov 12 14:04:27 03rpurdie * r646 10bitbake/lib/bb/fetch/ (8 files): fetchers: Create a urldata object which is passed between the different url methods. Start to use this to remove duplicate code and duplicate function execution, simplifying the fetchers Nov 12 14:05:12 koen: the real problem I see is when fsck _really_ can not fix your rootfs. You'd be totally screwed on a keyboard-less machine Nov 12 14:05:19 yep Nov 12 14:05:49 on most pdas that means 'eject sd card and fsck it on a real machine' Nov 12 14:06:08 right, that would be less than ideal tho Nov 12 14:06:36 I'd have to ask mreimer if the LAB bootldr already understands ext3 Nov 12 14:06:44 that would solve most issues Nov 12 14:06:51 heh indeed Nov 12 14:09:12 03rpurdie * r647 10bitbake/lib/bb/fetch/ (__init__.py cvs.py svk.py svn.py): fetchers: Remove check_for_tarball() in favour of try_mirror directly Nov 12 14:39:30 mickeyl: I need some python help and design pattern help Nov 12 14:39:57 mickeyl: I have my AST and I have certain operations on it Nov 12 14:40:33 mickeyl: e.g. generating a DOT graph of the AST, optimizing it, and evaluating it to a bb.data instance Nov 12 14:41:16 03rpurdie * r648 10bitbake/lib/bb/fetch/ (__init__.py cvs.py git.py ssh.py svk.py svn.py wget.py): fetchers: Cleanup more variables Nov 12 14:42:15 ibot: test Nov 12 14:42:21 Test Passed! Nov 12 14:43:43 RP: ERROR: Exception:exceptions.NameError Message:global name 'parm' is not defined Nov 12 14:43:57 RP: in _svn.bbs Nov 12 14:44:03 03coredump 07org.oe.oz354x * r0b010b19... 10/ (3 files in 3 dirs): initscripts: checkroot.sh: chvt 1 if fsck returns with an errorcode and waits for user input. Fixes the "hanging" bootsplash on some distros Nov 12 14:44:13 03coredump 07org.oe.dev * r4edccdfa... 10/ (3 files in 3 dirs): initscripts: checkroot.sh: chvt 1 if fsck returns with an errorcode and waits for user input. Fixes the "hanging" bootsplash on some distros Nov 12 14:44:27 morning Nov 12 14:44:27 RP: I'm at r648 with bitbake Nov 12 14:44:32 hey chouimat Nov 12 14:46:45 CoreDump|home: did you see http://lists.linuxtogo.org/pipermail/openembedded-devel/2006-November/000749.html ? Nov 12 14:48:03 hmm no Nov 12 14:48:35 03rpurdie * r649 10bitbake/lib/bb/fetch/svn.py: svn.py: Fix a reference to parm Nov 12 14:48:39 koen: I suspect there might be a few of these kind of bugs given I've renamed just about all the variables :-/ Nov 12 14:49:11 RP: that's why you have testers :) Nov 12 14:49:12 The fetchers themselves are getting smaller and smaller though :) Nov 12 14:49:35 koen: true, thanks :) Nov 12 14:49:44 koen: it complains about "PACKAGE_ARCH_${PN} = all" Nov 12 14:49:56 looks legal to me =) Nov 12 14:50:18 RP: another one: ERROR: Exception:exceptions.NameError Message:global name 'date' is not defined Nov 12 14:50:30 in NOTE: package gnu-config-native-0.1+cvs20050701-r4: task do_fetch: started Nov 12 14:51:00 CoreDump|home: it isn't legal for the parser :) Nov 12 14:51:10 ~lart the parser Nov 12 14:51:11 * ibot rm -rf's the parser Nov 12 14:51:14 ;) Nov 12 14:51:57 it should be PACKAGE_ARCH_${PN} = "all" Nov 12 14:52:11 that's nitpicking Nov 12 14:52:36 but I'll change it if it makes the parser happy Nov 12 14:53:04 03rpurdie * r650 10bitbake/lib/bb/fetch/ (cvs.py svk.py svn.py): svn.py: Fix a references to date Nov 12 14:53:12 if all those errors get fixed, zecke can publish the new parser Nov 12 14:53:20 and that parses everything in 2 seconds Nov 12 14:55:28 zecke: sound good... where's the question? ;) Nov 12 14:57:47 mickeyl: I do not want to have two methods in the classes that make the AST Nov 12 14:58:02 mickeyl: currently I have eval() I would need evalDot, evalData Nov 12 14:58:23 mickeyl: I look for a clever way to either operate on the AST (knowing the types and meaning) Nov 12 14:58:32 sounds like a task for a mixin. create the eval data in a seperate class and then derive a new class from both the AST and the Eval Nov 12 14:58:32 mickeyl: or somehow injecting the strategy/method into the AST :) Nov 12 14:58:58 i'm always a fan of multiple inheritance if methods are disjunct and base classes as well Nov 12 14:59:00 mickeyl: can I dynamically do this inherit? Nov 12 14:59:04 sure Nov 12 14:59:13 mickeyl: e.g. having a factory to generate the 'base' Nov 12 14:59:21 yep Nov 12 14:59:27 bar = Object Nov 12 14:59:30 class foo (bar) Nov 12 14:59:33 should work fine Nov 12 14:59:41 or even using new() Nov 12 14:59:44 but i never used that Nov 12 15:00:09 remember: classes are first class object, you can do almost everything you like with 'em Nov 12 15:02:46 ok, 7 hours work Nov 12 15:02:48 need a break now Nov 12 15:02:49 bbl Nov 12 15:06:03 03coredump 07org.oe.oz354x * r68f42c8e... 10/ (3 files in 2 dirs): altboot: Make Mr. Parser happy again Nov 12 15:06:35 03coredump 07org.oe.dev * rd2780d1e... 10/ (4 files in 2 dirs): altboot: Make Mr. Parser happy again Nov 12 15:08:07 03rpurdie * r651 10bitbake/lib/bb/fetch/ (__init__.py cvs.py svk.py svn.py wget.py): fetchers: Only call createCopy and update_data after checking if the download already exists. Refactor parameters to wget's fetch_uri and make md5 a global urldata option. Nov 12 15:10:21 RP: once you're done with the fetchers, could you bump it up to 1.7.4? Nov 12 15:11:00 koen: That sounds sensible :) Nov 12 15:27:22 BTW: anyone knows when the RX bytes of ifconfig/kernel wrap around? Nov 12 15:27:46 zecke: 2^32 on this box (x86-32) Nov 12 15:28:03 Kernel 2.6.18.1 Nov 12 15:28:23 NAbyss: and does it count the overruns? Nov 12 15:28:25 ~4GB iirc Nov 12 15:28:41 zecke: Not as far as I know, no Nov 12 15:28:46 :( Nov 12 15:28:53 zecke: Which is somewhat irritating Nov 12 15:29:17 zecke: This box's had uptimes of ~4 months.. it'd be interesting to see just how many tb go through it Nov 12 15:29:35 any clue if the NSLu2 can do 12mbit of of multicast traffic? Nov 12 15:30:05 zecke: it should be able to Nov 12 15:30:13 zecke: especially if you de-underclock it Nov 12 15:30:24 koen: together with 12mbit on USB? Nov 12 15:30:49 hmmm Nov 12 15:31:06 zecke: with your dvb stick? Nov 12 15:31:27 koen: yes Nov 12 15:31:47 zecke: you can pull a few Mbyte/s from a usbdisk over ethernet Nov 12 15:32:48 6-7mb with samba Nov 12 15:33:16 so 12mbit from usb -> 12mbit eth shouldn't be a problem Nov 12 15:33:42 Some of this fetcher code is just insane :-( Nov 12 15:34:30 hehe Nov 12 15:34:48 RP: remove fetcher from this sentenc Nov 12 15:35:18 zecke: I was wondering about the word "some" :-/ Nov 12 15:35:27 * chouimat learned the hardware that KDE svn checkout over samba on a slug == bad idea ... Nov 12 15:35:34 s/hardware/ hard way Nov 12 15:35:54 zecke: The good news is after beating up this code like this, we should be able to add generic (pre)mirrors support Nov 12 15:38:39 * koen prepares another go at an mtn fetcher Nov 12 15:38:57 koen: can angstrom build uclibc? Nov 12 15:39:06 NAiL: with a patch, yes Nov 12 15:39:13 koen: fetchers should be a lot easier to write now - much less voodoo involved... Nov 12 15:39:17 NAiL: set ANGSTROM_MODE = "uclibc" in local.conf Nov 12 15:40:04 NAiL: see http://linuxtogo.org/gowiki/AngstromuClibc Nov 12 15:40:37 koen: thanks :) Nov 12 15:41:41 koen: what about ppc? :) Nov 12 15:41:56 NAiL: waiting for my Efika to build for ppc Nov 12 15:42:03 ah, ok Nov 12 15:42:34 I'd like to actually test stuff :) Nov 12 15:42:46 I've got a PPC nas that I want to build packages for. Would be a lot easier if it had glibc Nov 12 15:43:58 aren't there online embedded boards we can use? Nov 12 15:44:23 likewise: like serial console? Nov 12 15:44:28 zecke: ssh Nov 12 15:44:46 zecke: typically, ssh into a server, to which the board is connected Nov 12 15:45:18 zecke: not really exciting though... no flashing leds nearby. takes the charm of embedded engineering :-) Nov 12 15:45:39 hehe Nov 12 15:45:52 NAiL: can you boot the TS-101 from anything else but Flash? Nov 12 15:46:45 Dunno. I haven't looked at the inner workings yet. It uses U-Boot, and I think it might support booting from ide Nov 12 15:47:01 I don't have serial on it yet, so it's kinda hard to find out. Nov 12 15:50:54 NAiL: OK, thanks. I think I still have some credit left from a returned item at one web shop. They have the ts-101 as a special offer now. Nov 12 15:51:34 likewise: ah, cool Nov 12 15:51:48 I'm working with the norwegian distributor to modify the firmware Nov 12 15:52:09 I've got a couple of unreleased firmware versions that has telnet enabled Nov 12 15:56:15 I have a perfectly good gcc-3.4.4 compiler + binutils, how can I tell OE not to build one but to use mine? I've tried the ASSUME_PREFERRED but OE insists on build it's own tools. Nov 12 15:56:58 s/ASSUME_PREFERRED/ASSUME_PROVIDED Nov 12 15:57:36 btw I have a similar question :) Nov 12 15:57:56 I added ASSUMEPROVIDED += "ncurses-native" and it still builds the native version for me, same with the virtual version Nov 12 15:58:09 yeah, something is borked Nov 12 15:58:10 s/ASSUMEPROVIDED/ASSUME_PROVIDED/ Nov 12 15:59:11 it takes so long to build the tools and so little time to build the image files... Nov 12 15:59:45 T0mW: actually I use a prebuilt toolchain, and the assume provided thing worked for gcc Nov 12 15:59:50 it is not trying to build libc or gcc for me Nov 12 16:00:18 it must be my syntax or where I put the assume-provided then. I put it into the build/conf/local.conf Nov 12 16:00:37 I have it in the local.conf as well Nov 12 16:01:16 * koen recommends reading http://www.openembedded.org/user-manual&dpage=ch05s05 Nov 12 16:01:38 is there any concept of known good releases in the monotone / OE build methodology? Nov 12 16:02:30 koen: I did follow that link btw :> Nov 12 16:02:46 aem, I mean, when I setup my environment I was following this article Nov 12 16:03:21 maybe instead of ASSUME-PROVIDED="virtual/arm-linux-gcc-3.4.4" I need to say ASSUME_PROVIDED="virtual/arm-linux-gcc-3.4.4-r4", the full name of what it tries to build? Nov 12 16:04:14 well I have ASSUME_PROVIDED += " virtual/${TARGET_PREFIX}gcc " Nov 12 16:04:33 ok Nov 12 16:05:07 hmm funny, I did not set the TARGET_PREFIX variable anywhere Nov 12 16:05:10 and it still works for me Nov 12 16:07:51 trying it without the "-3.4.4" Nov 12 16:08:25 TARGET_PREFIX is an OE variable Nov 12 16:08:39 mgross: no, what would known good mean with +4000 packages and 40 machines and 30 distributions Nov 12 16:08:50 koen: right, but his point was that he didn't define it. Nov 12 16:09:14 T0mW: my point was that OE defines it in conf/bitbake.conf Nov 12 16:09:23 hhe Nov 12 16:09:23 heh Nov 12 16:09:46 zecke: s/packages/recipes/ Nov 12 16:09:57 koen: we generate +4000 packages ;) Nov 12 16:10:14 we can generate... Nov 12 16:11:21 btw it's a rally cool system you guys came up with Nov 12 16:13:16 would be nice if manufacturers would build their environment using OE, usually you get some dev-board shipped along with the weirdest distro you could imagine Nov 12 16:13:34 sometimes really terrible and sick stuff Nov 12 16:13:43 Jin^eLD: http://linuxdevices.com/news/NS2986976174.html :) Nov 12 16:13:44 would be so much easier with OE Nov 12 16:13:49 would be nice if they gave us a reasonable kernel source with their dev boards Nov 12 16:14:02 not some ancient hacked up source Nov 12 16:14:38 koen: very nice! Nov 12 16:14:45 sigh, I guess line dev support starts "here" Nov 12 16:14:57 s/line/linux/ Nov 12 16:15:40 T0mW: well, what I am usually disappointed about are the prebuilt toolchains and a prebuilt set of libraries that you sort of have to use Nov 12 16:16:06 Jin^eLD: yeah, like when they give you a 2.95.3 dev Nov 12 16:16:09 right now I am struggling to create a package which would use those libs, ended up with seating LEAD_SONAME stuff and so on, very confusing Nov 12 16:17:19 Jin^eLD: don't misunderstand, I've done a bit with 2.95.3, it was good compiler, but Nov 12 16:18:47 ...it's old, and as far as I remember it had some problems, no Nov 12 16:18:48 ? Nov 12 16:19:01 this where I've found OE to be useful. I still have to get a 2.6.17 kernel working fully, but I can save the dev time using OE for my system buildup. Nov 12 16:19:20 Jin^eLD: I was pretty good system, quirky Nov 12 16:19:30 darn, cannot type today Nov 12 16:19:33 s/I/it/ Nov 12 16:19:35 :) Nov 12 16:19:49 I slept most of yesterday and last night Nov 12 16:20:02 drat, it is still building the gcc Nov 12 16:20:48 btw, can OE/bitbake be setup for an SMP system? Nov 12 16:21:15 Jin^eLD: the latest development version of bitbake can Nov 12 16:22:19 cool, I should try it then.. Nov 12 16:22:43 is that 1.6.1 ? Nov 12 16:22:52 zecke: is that 1.6.1 ? Nov 12 16:23:06 T0mW: hehe, no as 1.6.1 is not the latest Nov 12 16:23:10 ok Nov 12 16:23:18 T0mW: an no as we somehow follow the old kernel stable numbering scheme Nov 12 16:23:19 using that now on AMD X2 Nov 12 16:23:31 T0mW: 1.7.3 is latest unstable Nov 12 16:23:49 heh, I wonder if I'm ready to go unstable Nov 12 16:23:53 oops I need to send the release notes... Nov 12 16:24:39 unstable really means unstable, it can work, it can be broken, it could stay broken for weeks Nov 12 16:25:45 hello Nov 12 16:26:43 anyone here could explain me what's wrong with this rejected patch hunk ? Nov 12 16:26:48 *************** obj-$(CONFIG_EDAC) += edac/ Nov 12 16:26:52 *** 68,73 **** Nov 12 16:27:04 obj-$(CONFIG_EISA) += eisa/ Nov 12 16:27:08 obj-$(CONFIG_CPU_FREQ) += cpufreq/ Nov 12 16:27:12 obj-$(CONFIG_ARCH_MXC) += mxc/ Nov 12 16:27:16 obj-$(CONFIG_MMC) += mmc/ Nov 12 16:27:20 obj-$(CONFIG_INFINIBAND) += infiniband/ Nov 12 16:27:24 obj-$(CONFIG_SGI_SN) += sn/ Nov 12 16:27:28 --- 68,74 ---- Nov 12 16:27:32 obj-$(CONFIG_EISA) += eisa/ Nov 12 16:27:36 obj-$(CONFIG_CPU_FREQ) += cpufreq/ Nov 12 16:27:36 obj-$(CONFIG_ARCH_MXC) += mxc/ Nov 12 16:27:36 + obj-$(CONFIG_DPM) += dpm/ Nov 12 16:27:36 obj-$(CONFIG_MMC) += mmc/ Nov 12 16:27:36 obj-$(CONFIG_INFINIBAND) += infiniband/ Nov 12 16:27:37 *sigh* Nov 12 16:27:37 obj-$(CONFIG_SGI_SN) += sn/ Nov 12 16:27:45 valLongfield: maybe use http://rafb.net/paste/ Nov 12 16:28:03 or this one http://pastebin.com/ Nov 12 16:28:15 or http://pastebin.ca/ Nov 12 16:28:20 Jin^eLD: ok, I'll let OE build the stupid cross compiler Nov 12 16:28:39 ok sorry Nov 12 16:28:41 T0mW: I still wonder why it did not work for you.. Nov 12 16:29:10 well. in the meantime a have another problem: sh: ipkg-build: command not found Nov 12 16:29:23 I did compile the ipkg-utils and ipkg and also did a bitbake -c stage ipkg-utils Nov 12 16:29:38 hmm. but it's not in the staging directory Nov 12 16:29:44 ...why? Nov 12 16:29:49 Jin^eLD: wait, something was different that time, it said building gcc-cross-initial, worked for a bit, then skipped down to glibc Nov 12 16:33:36 so here is the .rej of the patch: http://pastebin.com/822605 I can't understand what's wrong there Nov 12 16:33:58 Jin^eLD: did you inherit the package_ipk ? Nov 12 16:34:41 T0mW: I think so.. let me recheck. it worked before... but before I compiled everything with soft FP and then it turned out that the brebuilt stuff was compiled with hard FP, so I ended up cleaning everything, started to recompile and for some reason it stopped to work Nov 12 16:34:51 Jin^eLD: I think I ran into that the other day, my distro file now has: INHERIT += "package_ipk" Nov 12 16:35:10 I have that as well Nov 12 16:35:22 so it must be something different...hmm Nov 12 16:36:32 I cleaned out all ipkg stuff and rebuilt it, seems to be ok again Nov 12 16:37:26 next thing I have to tackle is the locales Nov 12 16:41:40 T0mW: locales... btw, that's not really OE related, but did you ever use nl_langinfo() ? Nov 12 16:41:49 no Nov 12 16:41:51 I have a weird problem with it. Nov 12 16:41:59 well, then you won't know the answer I guess :) Nov 12 16:42:04 heh Nov 12 16:44:17 crap, yesterdays problem is haunting me again... Nov 12 16:44:46 /opt/rootfs/lib/libpthread.so.0 uses hardware FP, whereas .libs/libsqlite3.so.0.8.6 uses software FP Nov 12 16:45:01 I set TARGET_FPU = "hard" Nov 12 16:45:21 why is sqlite3 being compiled with soft FP? Nov 12 16:45:22 Jin^eLD: is there a hard FPU? Nov 12 16:45:44 T0mW: I guess not, but they did the prebuild libs with hard FP, actually I will mail them and ask why Nov 12 16:45:46 Jin^eLD: you have a hardware FPU ? Nov 12 16:45:48 Jin^eLD: did you change it somewhere during the build? Nov 12 16:45:55 but right now I have to compile with hard FP Nov 12 16:46:13 'hard' means having an actual fpu *or* emulating it in kernel Nov 12 16:46:23 koen: not that I would know... my distro/machine/local.conf file are not changing it, well - the local.conf sets it to hard, rest is not changing Nov 12 16:46:26 like the i386 coprocessor on x86 Nov 12 16:46:50 or pentium / amd cpu Nov 12 16:46:50 I think they are emulating it, afaik ARM9 does not have a hard FPU Nov 12 16:47:01 the arm9 on my desk has Nov 12 16:47:06 really? hmm Nov 12 16:47:09 wow Nov 12 16:47:13 cirrus ep9312 :) Nov 12 16:47:13 koen: how to figure that out? Nov 12 16:47:15 koen maverick? Nov 12 16:47:18 the arm920 I have doesn't Nov 12 16:47:20 chouimat: yes Nov 12 16:47:24 koen: will /proc/cpuinfo tell? Nov 12 16:47:56 Jin^eLD: most likely it won't have a fpu Nov 12 16:48:05 koen: FYI, I tried to build ixp4xx/angstrom/uclibc but it fails during uclibc: http://pastebin.ca/246751 Nov 12 16:48:19 koen: well I also think that it does not have one.. however I have to use the prebuild libraries, and they seem to be compiled with hard FP Nov 12 16:48:54 the question is... why is my TARGET_FPU setting not taken? when it starts building, this summary - it shows the "hard" target Nov 12 16:49:38 likewise: yes it currently requires you to build the kernel by hand before uclibc :( Nov 12 16:51:22 hmm. well, any idea how to figure out what it is still building the soft FP version? Nov 12 16:52:41 the OE Build Configuration: shows TARGET_FPU = "hard" Nov 12 16:58:46 03rpurdie * r652 10bitbake/lib/bb/fetch/ (__init__.py cvs.py git.py local.py svk.py svn.py wget.py): Nov 12 16:58:46 Fetchers: Add forcefetch method for handling of SRCDATE now or tag=master. Move Nov 12 16:58:46 md5 code from wget.py into the fetchre core. Start generating md5 stamps for all Nov 12 16:58:46 downloads including scm tarballs. When a download is accessed, touch the md5 Nov 12 16:58:46 file to allow easy creation of source mirrors Nov 12 16:59:37 source mirrors should now be easy :) Nov 12 17:25:29 Hmm. I've broken bitbake :-/ Nov 12 17:41:31 hail mickeyl Nov 12 17:42:34 hey pb_! how are things? Nov 12 17:42:37 hi pb Nov 12 17:44:05 mickeyl: very good, thanks. how are you? Nov 12 17:44:06 hi woglinde Nov 12 17:44:58 pb_: struggling with a cold and unbelievable tight deadlines (you might heard of the OpenMoko project), but excited at the same time Nov 12 17:48:01 http://dot.kde.org/1163332807/ Nov 12 17:48:29 yeah. finally Nov 12 17:49:52 03rpurdie * r653 10bitbake/lib/bb/fetch/ (__init__.py cvs.py svk.py svn.py): Fetchers: More cleanups, error handling fixes, fix SRCDATE handling to account for PN 'overrides' Nov 12 17:55:03 mickeyl: aha, very good. I hadn't previously heard of OpenMoko but google reveals to me what it is. Nov 12 17:55:14 the cold and deadlines don't sound so good though Nov 12 17:56:32 pb_: yeah, *shrug* i'm working as fast as I can. thank god no one's really pushing me harder than me, which is ok, i think. I'm finally glad you and me work on the same "side" of UIs ;) Nov 12 17:57:03 :-) Nov 12 17:57:52 * mickeyl going to a birthday party... would rather like to stay in bed Nov 12 17:57:58 see you guys Nov 12 17:59:07 mickey_away: later Nov 12 18:03:22 03rpurdie * r654 10bitbake/lib/bb/fetch/svk.py: svk.py: Remove unused variables Nov 12 18:55:49 03rpurdie * r655 10bitbake/ (4 files in 2 dirs): taskdata.py: Improve abort flag handling, fixing several bugs Nov 12 18:58:34 03rpurdie * r656 10bitbake/bin/bitbake: bitbake: Add calls to expandKeys, fixing issues in bug #1294 Nov 12 19:12:10 03rpurdie * r657 10bitbake/lib/bb/runqueue.py: runqueue.py: raise build failure exceptions instead of trapping them to allow stack traces to be seen Nov 12 19:13:27 03rpurdie * r658 10bitbake/lib/bb/runqueue.py: runqueue.py: Add sanity check for zero length task list Nov 12 19:43:14 koen|awa1: I am happy to report that your changes made the problem with jards go away. Thanks a lot! Got it installed now. Nov 12 19:48:52 morning Nov 12 19:49:11 RP: nice to see fetcher changes - I will work on source mirror soon Nov 12 19:56:02 getting gnome running on 112M RAM device is not for me. Nov 12 19:56:39 hrw with compiz? Nov 12 19:57:20 woglinde: no Nov 12 19:57:31 woglinde: this machine is from 2000/2001 - gfxcard has 4M ram Nov 12 19:59:01 hihi Nov 12 19:59:27 woglinde: its nice webpad Nov 12 19:59:38 I'll try icewm or kde on it Nov 12 20:03:03 sound fun Nov 12 20:03:29 default debian start too much stuff on it Nov 12 20:05:39 hra ah that one from the blog Nov 12 20:05:49 yes Nov 12 20:05:55 h default debian shouldnt start anyrhing Nov 12 20:06:02 hrw: go with gentoo Nov 12 20:06:21 chouimat 10 days waiting for compiling? Nov 12 20:07:01 only 10? Nov 12 20:07:03 woglinde: icecream and/or distcc where invented for what ??? your dog? Nov 12 20:07:26 its 400MHz transmeta Nov 12 20:07:44 hrw: build everything on a fast computer like you would do with OE :) Nov 12 20:07:54 icream distcc implies that you have a compile farm with an enegery factory Nov 12 20:08:22 chouimat: finally I will build something OE based for it, now want debian Nov 12 20:08:22 you should use OE to create a a webpad distro Nov 12 20:08:22 *g* Nov 12 20:10:02 hrw: what extra services does it have running? unless you some heavy tasks, there shouldn't be many tasks left in minimal etch install Nov 12 20:10:22 yurk debian ... piece of crap :) Nov 12 20:10:24 suihkulokki: cups, lpr, avahi Nov 12 20:11:02 hrw waht? Nov 12 20:11:09 lpr and cups istn standard Nov 12 20:11:12 suihkulokki: and harddisk is also from 2000/2001 Nov 12 20:11:26 woglinde: standard if you will install 'desktop environment' Nov 12 20:11:36 fscking gnome Nov 12 20:11:49 yeah, that would be from gnome Nov 12 20:12:29 hrw hrhrhrhrhr Nov 12 20:12:45 just install without selecting any task and apt-get any app you need (firefox, xorg-xserver & icewm ) Nov 12 20:14:11 suihkulokki: sure Nov 12 20:14:28 suihkulokki: firefox and debian? dont make me laugh please.. Nov 12 20:14:33 :F Nov 12 20:14:37 1.5.0.x available only Nov 12 20:14:43 2.0b2 only for i386 Nov 12 20:14:56 no any sign of newer firefox or iseweasel Nov 12 20:15:31 hrw: Do you think it makes sense for openembedded-essential to suggest or recommend python-psyco? Nov 12 20:16:02 it make Nov 12 20:16:48 I need to talk with some DD about getting this metapackage into debian Nov 12 20:16:58 That would surely be nice. Nov 12 20:17:18 Can you also propose it to Ubuntu? Nov 12 20:17:39 if it will be in Debian then it will also spread into ubuntu Nov 12 20:17:48 if not then fsck ubuntu Nov 12 20:17:52 ;D Nov 12 20:18:13 hrw: wanna talk to me about it? :) Nov 12 20:18:21 suihkulokki: you are DD? Nov 12 20:18:35 hi greentux Nov 12 20:18:36 hrw: yes Nov 12 20:18:50 suihkulokki: ;D Nov 12 20:19:51 hrw: https://wiki.ubuntu.com/MOTU/Packages/Candidates Nov 12 20:19:55 I've thought about packaging bitbake and someking of metapackage of OE dependencies Nov 12 20:20:00 Just list it and it will be included faster Nov 12 20:20:14 suihkulokki: There was somebody else Nov 12 20:20:16 but last time I checked bitbake was moving fast Nov 12 20:20:28 suihkulokki: I believe his nick was Stefan_Schmidt or something Nov 12 20:20:40 suihkulokki: Maybe you want to cooperate. Nov 12 20:20:52 suihkulokki: bitbake package would be nice Nov 12 20:20:53 Talked with him in the channel yesterday or Friday about it. Nov 12 20:20:56 hi hrw Nov 12 20:21:00 Laibsch: yesterday Nov 12 20:21:37 hrw: let me know how the mirror works out. Hopefully the code changes will assist with it :) Nov 12 20:22:31 RP: what do you think about marking fetchers archives (SCM ones) in some way? this way cronjob could remove 'yesterday' fetchings.. Nov 12 20:23:01 RP: also mode for source mirror to autocreate archives from scm trees stored in CVS/SVN/GIT_DIR Nov 12 20:23:11 hrw: We generate md5 stamps for SCM achives now. Each time they're accessed we touch the md5 file Nov 12 20:23:42 RP: I was thinking about spliting source mirror stuff into few parts: Nov 12 20:23:47 hrw: So the mirror should just need to look at the timestamps of the md5 files and remove anything not touched for a while Nov 12 20:23:50 1. normal wget fetch Nov 12 20:24:00 2. sanedates scm archives Nov 12 20:24:06 3. eachday scm archives Nov 12 20:24:11 4. scm trees Nov 12 20:24:21 1 and 2 will get stored forever Nov 12 20:24:26 3 removed each next day Nov 12 20:24:36 4 used to regenerate 3 on request? Nov 12 20:24:50 hrw: How about the idea above? Nov 12 20:25:10 RP: sounds quite ok Nov 12 20:25:35 It has the advantage of being simple and now coded into bitbake :) Nov 12 20:26:24 204G of diskspace for testing :) Nov 12 20:26:48 /dev/sda3 218G 2.9G 204G 2% /a Nov 12 20:30:36 10M/s is nice.. too bad that this is lan only Nov 12 20:38:03 doh.. I specified the files that should go into the package in my .bb file using FILES_${PN} = still I get the message NOTE: the following files were installed but not shipped in any package Nov 12 20:38:07 any idea what I am doing wrong? Nov 12 20:38:37 I anyway thought that it should find the files automatically and that this manual FILES setting is not needed Nov 12 20:38:38 ? Nov 12 20:39:14 it also does find most of the files for the package, but surprisingly not all of them... allthough make install seems to put them all in the right plcae Nov 12 20:39:15 place Nov 12 20:39:17 any ideas? :> Nov 12 20:44:22 hrw, Laibsch, suihkulokki: The bitbake debian package is still available for testing. Nov 12 20:44:24 deb http://www.sicherheitsschwankung.de/~jluebbe/debian unstable/ Nov 12 20:45:01 If we get some success reports we only need to polish it a bit and push into new. Nov 12 20:46:34 stefan_schmidt: dh_make is nice tool. but need cleaning of debian/ dir before you show results to people.. Nov 12 20:46:58 stefan_schmidt: install lintian, linda, devscripts and rebuild with 'debuild' Nov 12 20:48:48 hrw: It's not mine. I only poked Shoragan to package it. ;) Nov 12 20:48:53 Shoragan: ping Nov 12 20:49:08 ah Nov 12 20:49:08 Shoragan: hrw has some suggestions to your bitbake package. Nov 12 20:49:48 hrw: Normally Shoragan knows what he does. Already have some packages in debian. Nov 12 20:50:14 hrw: What's exactly the prblem you see in the debian dir? Nov 12 20:50:22 RP: bitbake -cfetchall bootstrap-image == fetch ipkg and end Nov 12 20:51:07 RP: but bitbake -cfetchall task-base is progressing Nov 12 21:03:41 RP: how to use -cfetchall? 'bitbake -cfetchall world' does not work Nov 12 21:04:36 world should work... Nov 12 21:05:39 NOTE: Resolving missing task queue dependencies Nov 12 21:05:39 [hrw@misc1 celinux-test]$ Nov 12 21:05:56 thats all Nov 12 21:06:16 hrw: Was -c fetchall bootstrap-image on a clean tmp directory? Nov 12 21:06:39 RP: clean Nov 12 21:06:44 hmm. bitbake world appears broken... Nov 12 21:11:54 RP: bitbake --iknowthattrunkisunstableandcankillmydog switch (undocumented) would be nice Nov 12 21:12:09 ERROR: No providers of build target gpe* (for []) Nov 12 21:12:18 bitbake -cfetchall gpe* Nov 12 21:12:41 hrw: bitbake doesn't support wildcards like that Nov 12 21:13:59 RP: it was (or maybe it was in bbshell) Nov 12 21:14:55 hrw: bbshell might, the commandline never has Nov 12 21:15:11 ok Nov 12 21:16:05 hrw: Interestingly, opie-image works, bootstrap-image doesn't :-/ Nov 12 21:20:53 hrw: world builds fail due to a NoProvider exception - my last checkins broke the errors messages (they appear at debug level). The --continue option should work with world Nov 12 21:21:29 APIC error on CPU0: 40(40) Nov 12 21:21:30 suxx Nov 12 21:22:48 For bootstrap-image: "Pruned 1922 inactive tasks, 3 left" looks very wrong :} Nov 12 21:26:20 reboot time Nov 12 21:34:53 03rpurdie * r659 10bitbake/lib/bb/taskdata.py: taskdata.py: Add missing error message for unbuildable targets Nov 12 21:44:31 ~dict jerky Nov 12 21:44:33 Dictionary 'jerky' (1 of 5): aspic, barbecue, beef, beef extract, beef tea, boiled meat, bouilli, bouillon, broken, broken off, bully, bully beef, capricious, careening, cataclysmic, catchy, changeable, changing, charqui, chipped beef, chopped-off, choppy, civet, convulsive, decousu, desultory, deviating, deviative, deviatory, different, disastrous, ... Nov 12 21:51:48 hrw|gone: bitbake world should at least give errors now but there is something going wrong with the task depdendencies too causing the bootstrap-image issue Nov 12 21:52:17 ok Nov 12 21:52:26 just started with -D -D -D Nov 12 21:53:06 ERROR: No buildable providers available for required build target tsclient Nov 12 21:53:10 thats world Nov 12 21:53:33 DEBUG: No providers of build target libpanelapplet (for ['/home/hrw/devel/oe/org.openembedded.dev/packages/tsclient/tsclient_0.140.bb']) Nov 12 21:53:41 so broken dependency we have Nov 12 21:53:41 hrw: Yes, I expect a dependency problem somewhere... Nov 12 21:55:26 yes. It also appears the dependency code is correct and we just can't do fetchall that way. I need to give it some thought :} Nov 12 21:55:42 DEBUG: No providers of build target gtk+-2.6.4-1.osso7 Nov 12 21:56:34 started with -c Nov 12 21:56:42 -k I meant Nov 12 21:59:00 File "/home/hrw/devel/bitbake/bitbake/lib/bb/runqueue.py", line 80, in prepare_runqueue Nov 12 21:59:02 depdata = taskData.build_targets[depid][0] Nov 12 21:59:05 IndexError: list index out of range Nov 12 21:59:56 -k support is a bit untested :-/ Nov 12 22:00:03 ;) Nov 12 22:03:11 'night all Nov 12 23:34:09 03rpurdie * r660 10bitbake/lib/bb/taskdata.py: taskdata.py: Add some extra debug to dump_data and optimise depid and rdepid to avoid duplicates Nov 12 23:35:57 03rpurdie * r661 10bitbake/lib/bb/runqueue.py: runqueue.py: Have recrdeptask follow depends as well as rdepends (and rdepends of depends and depends of rdepends) Nov 12 23:37:09 03rpurdie * r662 10bitbake/lib/bb/runqueue.py: runqueue.py: Avoid errors in builds with build failures when using -k option Nov 12 23:55:47 03rpurdie * r663 10bitbake/lib/bb/providers.py: providers.py: Make PREFERRED_PROVIDER_foobar defaults to foobar if available Nov 12 23:56:49 03rpurdie * r664 10bitbake/ (bin/bitbake lib/bb/__init__.py): Bump to version 1.7.4 Nov 13 00:03:33 good night Nov 13 00:13:05 anyone know what size SD SLC-1000 can accept? Nov 13 00:13:37 Zero_Chaos: Standard SD cards Nov 13 00:14:23 RP: I meant can it accept a 4gig Nov 13 00:14:59 Zero_Chaos: People have reported 2GB working, I don't know about 4, sorry Nov 13 00:16:06 RP: well, just to walk through logic... there is not odd limitation for ext2 at 2GB like there is for the FAT/FAT32 border... but does the linux driver (I'm assuming 2.6 kernel) support 4 gig? Any clue? Nov 13 00:18:50 Zero_Chaos: There is a potential issue in that those cards work on 512byte blocks and the maximum number of 512 byte blocks addressable but a 32 bit number is 2GB Nov 13 00:19:56 RP: why did you have to go and bring logic into this :-P Nov 13 00:20:14 RP: thanks for the reality check, I guess I'll wait for someone who has tried. Nov 13 00:20:29 Zero_Chaos: You did ask for logic ;-) Nov 13 00:21:23 RP: I know ;-) Nov 13 00:24:26 'night all Nov 13 02:14:56 hi, is anybody out there using monotone 0.31? **** ENDING LOGGING AT Mon Nov 13 02:59:57 2006