**** BEGIN LOGGING AT Sat Sep 16 02:59:57 2006 Sep 16 07:50:34 morning Sep 16 07:51:18 hi Sep 16 08:17:52 morning all Sep 16 09:38:11 good morning all Sep 16 09:53:42 koen, morning Sep 16 09:53:51 hey leoncamel Sep 16 10:20:46 hi, can anybody explain to me why this "bitbake angstrom-gpe-image" fails on libsdl? http://en.pastebin.ca/172817 Sep 16 10:20:47 a "bitbake libsdl-x11" completed successfully just before that. Sep 16 10:37:35 pH5, hmm, not sure. I "bitbake opie-image" and "bitbake gpe-image" fails. Sep 16 10:38:38 pH5, it seems there are a lot of modification on OpenEmbedded core system. and there are a lot of packages broken. Sep 16 10:55:49 bitbake's output is not very helpful, there are no ERROR messages at all. I'm not even sure that the libsdl problems in the DEBUG messages are the reason for bitbake's failing silently Sep 16 11:16:02 pH5: the last line of your paste bin is a DEBUG and may be ERROR in some cases Sep 16 11:16:13 bitbake fails to build virtual/libsdl because I masked out *opie*, so libsdl-qpe fails because of missing dependencies. Sep 16 11:16:26 but I have set PREFERRED_PROVIDER_virtual/libsdl to libsdl-x11 Sep 16 11:16:28 * cyrilRomain looks at the bitbake/lib/bb/taskdata.py file Sep 16 11:17:11 * cyrilRomain line 352:359 Sep 16 11:19:49 cyrilRomain: thanks, but doesn't that mean that whatever called add_provider in this case should have printed an error message after the NoProvider exception was raised? Sep 16 11:20:05 I think bitbake hacker is needed to explain why add_provider is called with external = False in your case Sep 16 11:20:43 /summon bitbake hacker Sep 16 11:21:24 pH5: I'm afraid not to know bitbake enough to know if such behaviour is intended or not :-/ Sep 16 11:22:15 morning guys Sep 16 11:22:26 pH5: is that bitbake 1.6 or TRUNK? Sep 16 11:22:28 cyrilRomain: thanks for that hint anyway. at least I know what happens now, if not why. Sep 16 11:22:34 good morning mickeyl Sep 16 11:22:42 that is bb trunk Sep 16 11:22:53 ah. i guess RP will be interested then Sep 16 11:24:28 pH5: ah, it seems there is a # FIXME comment in the code Sep 16 11:24:41 line 528:532 Sep 16 11:25:18 for now it just raise and error but I guess there will be a bb.msg.error or simething like that Sep 16 11:25:31 s/and/an/ Sep 16 11:29:19 hey Sep 16 11:29:33 hi zecke Sep 16 11:29:36 cyrilRomain: where should we put bittest? I think you would like to be able to commit to it Sep 16 11:30:05 zecke: Koen is right, it is more a OETest than Bittest Sep 16 11:30:30 zecke: I can send you the patch so that you can review it, and change things you don't like Sep 16 11:30:59 cyrilRomain: some modules are more OETest the 'framework' is not :) Sep 16 11:31:12 zecke: btw, I'm wondering what a generic bittest would consist in: what kind of test should it perform ? Sep 16 11:31:15 cyrilRomain: we could put it back to the bitbake repo and you get yourself a berlios account Sep 16 11:31:41 cyrilRomain: Generic BitTest == Generic Framework to access files Sep 16 11:32:00 cyrilRomain: tests will be always specific Sep 16 11:32:05 zecke: ok, so there is no pb then since I only modified the modules :) Sep 16 11:33:38 RP: how current is your qemu page? Sep 16 11:33:44 RP: I would like to run poky here Sep 16 11:35:05 hi zecke Sep 16 12:12:24 koen|away: NetBSD has outscored us :) Sep 16 12:12:34 koen|away: 160.000 revs Sep 16 12:14:35 heh. i didn't know we had a competition running Sep 16 12:14:44 ????? Sep 16 12:15:42 hrw|gone: is it me or did neither cacko nor pdaxrom do any releases since ages? Sep 16 12:15:55 mickeyl: I didn't know that as well Sep 16 12:16:00 hehe Sep 16 12:16:13 mickeyl: but they have 160.000 so competition ended before it started ;) Sep 16 12:16:21 mickeyl: define "ages" ;) Sep 16 12:16:22 mickeyl: sash was hired away Sep 16 12:16:33 chouimat: more than 6 months Sep 16 12:17:58 mickeyl: :) Sep 16 12:22:25 * mickeyl comments to http://www.oesf.org/forums/index.php?act=ST&f=72&t=21055 Sep 16 12:25:31 hey, folks. it seems the sysvinit-2.86-r28 is broken ? Sep 16 12:41:55 hey. folks. here is the log of sysvinit-2.86-r28: http://rafb.net/paste/results/5ioEyL26.html Sep 16 12:47:00 zecke: maybe can we split bittest into bittest and oetest, where oetest will hold __all_test__ variable and the modules (?) Sep 16 12:47:59 leoncamel: do_package runs do_install a second time, and sysvinit's do_install obviously doesn't clean out the image folder Sep 16 12:48:24 I think we should remove do_install from PACKAGEFUNCS in classes/package.bbclass as do_package is called immediately after do_install anyway Sep 16 12:53:04 pH5, so what is your patch ? Sep 16 12:54:09 leoncamel: http://en.pastebin.ca/172979 Sep 16 12:54:18 hi lrg Sep 16 12:55:01 hey pH5 Sep 16 12:58:05 pH5, but, others packages build successfully. Sep 16 12:59:17 cyrilRomain: well, this will give us more trouble Sep 16 12:59:45 cyrilRomain: e.g. loading of modules is complex already, searching in more place :) Sep 16 12:59:45 in more places will not make it more easy Sep 16 13:00:00 leoncamel: those are packages that happily do_install twice. sure, perhaps that failing ln in sysvinit should have an -f parameter, but still calling do_install twice shouldn't be needed. Sep 16 13:00:01 zecke: hum ok Sep 16 13:02:05 pH5, hmm, is this bug introduced by this patch: http://www.openembedded.org/viewmtn/diff.psp?id1=db2f843f48cece5e25cf2e654db269ba3dba5192&id2=fd4f8197fc893ac473a9b7436ba5268b5e0da70d&fname=classes/package.bbclass Sep 16 13:06:20 leoncamel: yes. To me it looks like the addition of do_install to PACKAGEFUNCS part was an unwanted change Sep 16 13:06:29 kergoth: looking at the above patch I see you have changed from PACKAGEFUNCS ?= to = again Sep 16 13:06:37 kergoth: why did you do that? what was your reasoning? Sep 16 13:16:15 zecke: it is more complex than using 'sys.path.append' to oetest/modules directory in a oetest.py script that basically call the bittest's start_testing function ? Sep 16 13:16:54 or even bittest.py itself even Sep 16 13:18:21 kergoth: and can I remove do_install from PACKAGEFUNCS again? Sep 16 13:28:53 cyrilRomain: no, but this was too complex already... Sep 16 13:30:22 pH5, zecke: so what is the best solution ? Sep 16 13:43:29 zecke: http://www.o-hand.com/~richard/qemu.html Sep 16 13:43:45 zecke: but check out the qemu scripts in poky as they're more up to date :-/ Sep 16 13:44:51 morning all Sep 16 13:44:51 RP: do I need to compile qemu separately? Sep 16 13:45:16 hey RP Sep 16 13:45:22 zecke: qemu-native from poky will work just fine (as long as you left the sdl bits enabled) Sep 16 13:45:32 hi RP Sep 16 13:45:59 pH5: Just looking at that bitbake issue. Its a tricky one :) Sep 16 13:46:44 pH5: Was an alternative provider of virtual/libsdl built? Sep 16 13:48:42 I now masked out libsdl-qpe to work around this. Before that run, libsdl-x11 was already built. After that for name in names part in shell.py/build, libsdl-qpe is somehow in the list of unresolved providers, thus td.add_unresolved fails Sep 16 13:48:43 RP: does qemux86 deserve updates? Sep 16 13:49:08 zecke: qemux86's framebuffer doesn't work properly yet :-( Sep 16 13:49:38 It will boot, just X won't start Sep 16 13:49:40 RP: oh well, the Gtk+ + Xoo should work Sep 16 13:49:55 zecke: qemuarm works Sep 16 13:50:36 pH5: Is the problem that is errored when it shouldn't or that it didn't error when it should? Sep 16 13:53:05 RP: http://pastebin.ca/173031 <- could you please apply this? Sep 16 13:53:59 RP: two problems. first, something adds libsdl-qpe to the list of unresolved build targets (and I have its build dependencies masked out) although PREFERRED_PROVIDER is set to libsdl-x11, and then add_unresolved raises NoProvider, but there is no ERROR message Sep 16 13:54:52 * RP suspects there's a case where there are no providers of a given depend which should error instead of generate a debug warning Sep 16 13:55:09 pH5: The first problem sounds like a potential metadata issue in OE Sep 16 13:56:16 zecke: ok, although I wasn't aware poky used gcc 4 with binutils 2.15 :) Sep 16 13:56:47 RP: Ubuntu Dapper Drake Sep 16 13:56:58 RP: using qemux86 Sep 16 13:57:08 zecke: Strange as I run dapper myself :) Sep 16 13:57:09 RP: and binutils-cross used my host compiler Sep 16 13:57:26 RP: my gcc is stranger than yours? Sep 16 13:57:57 RP: virtual/libsdl resolves to libsdl-x11 just fine as long as I mask out the other provider (which is unbuildable because of missing dependencies) Sep 16 13:58:06 zecke: Perhaps I default to gcc 3.x or something :} Sep 16 13:58:55 I wonder why glibc-initial 2.4 gets build :( Sep 16 13:59:07 pH5: What it should do is work out the other provider is unbuildable and drop it (silently) just leaving you with libsdl-x11 Sep 16 14:00:32 zecke: applied, thanks Sep 16 14:00:59 RP: it somehow drops libsdl-x11, too, if libsdl-qpe is available but unbuildable. I think you should be able to reproduce this if you mask out .*opie.* in BBMASK and then try some action on virtual/libsdl Sep 16 14:01:00 zecke: It will build 2.3.5 after that - I noticed the issue but didn't get around to fixing it Sep 16 14:01:48 pH5: ok, I'd best have a look at it, thanks Sep 16 14:02:47 RP: how well does clutter perform on iwmmxt hardware? Sep 16 14:03:06 zecke: Since it needs GL, probably not well at all Sep 16 14:10:41 03rpurdie * r610 10bitbake/lib/bb/taskdata.py: taskData.py: Only mark external targets as external Sep 16 14:11:31 03rpurdie * r611 10bitbake/ (bin/bitbake lib/bb/runqueue.py): Fix -f force option when used without -b Sep 16 14:12:07 03koen 07org.oe.dev * rb24a1121... 10/ (1 classes/package.bbclass): Sep 16 14:12:07 package.bbclass: remove do_install *again* to fix all the failed builds people are reporting Sep 16 14:12:07 ~lart kergoth for introducing this regression Sep 16 14:12:07 * ibot slams kergoth against a large cement Tux for introducing this regression Sep 16 14:12:54 ~botmail for rwithby: rb24a1121... should fix the util-linux problems you were seeing Sep 16 14:16:49 RP: can bitbake disptach jobs to other bitbakes yet? Sep 16 14:17:34 is there an easy way to find out what makes bitbake think it has to build perl? Sep 16 14:18:11 zecke: Not yet. Patches welcome ;-) Sep 16 14:19:11 pH5: You could try -l Depends Sep 16 14:19:25 Hi, I am trying to compile http://granule.sf.net on OE. My current work is at http://oz.leggewie.org/granule/ I made some progress, but now I am stuck. Maybe somebody can help me? building granule requires access to the assa lib. the problem seems to be that for example assa/GenServer.h is not where OE expects it to be. Can I fix this with RDEPENDS instead of DEPENDS or do I need to patch the source to make it look for the include files elsewhere? I Sep 16 14:19:30 pH5: We really need to improve the logging and make the domains useful now... Sep 16 14:21:16 Laibsch: http://rafb.net/paste/results/hDEm6q36.html Sep 16 14:22:07 RP: thanks, that did the trick Sep 16 14:22:14 koen: why does abiword DEPEND on perl? Sep 16 14:22:55 it seems perl cannot be built if there is no perl-native staged Sep 16 14:22:58 pH5: that should be perl-native Sep 16 14:23:16 Thanks koen. What does this autotools_stage_all do exactly so that maybe next time I can apply it myself? Sep 16 14:23:17 pH5: the bootsplash code (cdump) is now written in perl Sep 16 14:23:33 Laibsch: copy .so and .h files to staging/ Sep 16 14:23:41 I see. Sep 16 14:24:05 Laibsch: for properly autotools packages, that is Sep 16 14:24:13 autotooled* Sep 16 14:24:19 Let's see ;-) Sep 16 14:26:59 I get "NOTE: preferred version 0.92 of dbus not available (for item dbus-0.23.1-osso5)" - are "items" providers or versioned providers or can they be both? Sep 16 14:28:20 pH5: I think we have some limited form of versioned depends which is what is showing up there Sep 16 14:50:24 clutter's video-cube doesn't work :( Sep 16 15:00:37 hi Sep 16 15:01:21 Laibsch: did it work? Sep 16 15:01:51 Compiling now. Sep 16 15:10:33 zecke: can you help me in simpad stuff: which oe kernel 2.6 should i take (simpad-2.6.conf define handhelds-pxa-2.6 PXA->xscale???) Sep 16 15:12:04 Hello Sep 16 15:12:16 hey sirfred Sep 16 15:12:32 hi sirfred, hi zecke :) Sep 16 15:15:49 03koen 07org.oe.dev * r4cdccae5... 10/ (1 packages/abiword/abiword.inc): abiword.inc: fix up DEPENDS Sep 16 15:16:02 pH5: it looks like the abiword maintainer wasn't paying attention to the new RDEPENDS stuff Sep 16 15:22:49 koen: thanks :) Sep 16 15:35:59 /usr/bin/ccache: invalid option -- S ... i guess CC is broken, but i cannot find the failure. any tips or tricks how to fix this? Sep 16 15:41:48 balle: no idea Sep 16 15:42:02 balle: what would be 'S'? Sep 16 15:44:10 i dont know. cannot find a -S option in the makefile, but therefore the help output of ccache at the end oO Sep 16 15:46:58 balle: well, currently it is Garbage-In, Garbage-Out Sep 16 15:49:11 balle: you do not give any information, nobody can give an answer Sep 16 15:51:49 balle: in which context did you get that error ? Sep 16 15:52:19 well thats the only error it get from the log files. configure is writing broken makefiles. in the bb file i am running do_configure() { oe_runconf --enable-client-only } Sep 16 15:52:47 i am trying to bitbake bacula filedaemon Sep 16 15:54:28 zecke: can you help me in simpad stuff: which oe kernel 2.6 should i take (simpad-2.6.conf define handhelds-pxa-2.6 PXA->xscale???) Sep 16 15:55:02 zecke: with handhelds-sa-2.6, actual cvs date if got some error (usb-net.c), should i take a older SRCDATE? Sep 16 15:58:08 balle: sorry no clue. maybe can you have a look at the configure options used in the bacula gentoo ebuild: http://gentoo-portage.com/app-backup/bacula Sep 16 15:59:17 i am a little bit confused that i cannot find the "ccache -S" anywhere in the source producing the error in the log.do_configure... "magic"? :) Sep 16 16:02:22 but the gentoo link looks very useful Sep 16 16:03:43 CCACHE = "" in local.conf Sep 16 16:18:30 Slims: http://handhelds.org/~zecke/simpad/linux-2.6/.git my git repo Sep 16 16:18:32 Slims: completely outdated Sep 16 16:18:55 complain at the people that wanted to stick at 2.4er kernel Sep 16 16:22:53 I'm just getting started with an AMD X2 board, when I add 'PARALLEL_MAKE = "-j 4"' to the local.conf, it doesn't seem to make an appreciable difference as to bitbake's performance. Is there something else I need to do to run SMP builds? Sep 16 16:23:58 bitbake is a bit faster on this new board, shaves about 12 minutes off a 48 minute build, but that seems to be the result of the faster clockrate of a single core. Sep 16 16:24:43 T0mW: you could try the unstable branch of BitBake Sep 16 16:24:51 T0mW: it can run multiple builds at once Sep 16 16:25:13 oh, it is not in the version 1.6.1 then? Sep 16 16:25:39 T0mW: well, PARALLEL_MAKE is for 'GNU make' Sep 16 16:26:20 ok, I'll try the unstable. The 48 minute run is too long Sep 16 16:26:21 T0mW: but this will only speed up do_compile, do_configure etc. will still be single threaded Sep 16 16:26:49 but be aware, unstable means unstable Sep 16 16:26:56 heh Sep 16 16:27:31 * T0mW looks at finger scars from other bleeding edge software: his x86_64 bit distro install for instance Sep 16 16:28:18 freyther: i'm not so familiar with git, contain it the hole kernel or yust part of it? Sep 16 16:28:55 Slims: sorry no warranty and support Sep 16 16:29:29 Slims: not familiar with GIT, read a manual. Again I currently have not the time to look into SIMpad. When I called for testers years ago no one was there Sep 16 16:29:47 Slims: now I'm not calling for anyone :) Sep 16 16:31:40 T0mW: you might want to set BB_NUMBER_THREADS as well Sep 16 16:34:27 freyther: ok, i clone your git and looking forward. try and error :-) Sep 16 16:35:02 Slims: I don't want to be mean but you should have some sort of clue about linux kernel. And at least you should be familiar building and getting a kernel Sep 16 16:35:23 freyther: ok, thanks Sep 16 16:35:48 freyther: something like "2", for "two cores" ? Sep 16 16:36:18 T0mW: experiment :) Sep 16 16:36:23 heh Sep 16 16:37:15 T0mW: I could give your X2 shelter and try it for you, If you prefer that Sep 16 16:37:15 RP: do you know something about the Monahans? Sep 16 16:38:03 :P Sep 16 16:40:37 It sure is nice running a dual core, I can do two intensive tasks like transcoding an MPEG video and a bitbake build. Sep 16 16:40:46 hehe Sep 16 16:40:57 chouimat: morning Sep 16 16:41:07 freyther: some linux kernel basics, more hardware know how, just learning by doing Sep 16 16:41:14 T0mW: hi Sep 16 16:41:19 bbl Sep 16 16:41:33 * chouimat is away: food shopping Sep 16 16:52:15 RP: nice bootsplash on poky Sep 16 16:57:59 freyther: last question (at the moment :-) have you still a simpad Sep 16 17:01:21 Slims: yes, and if time would permit I would hack on it Sep 16 17:02:06 Slims: I still need to (re)solder the JTAG, it is lying somehwere on my desk Sep 16 17:51:57 RP: I fear your conf/local.conf is broken in regard to BB Collections Sep 16 17:52:23 # BBFILE_PATTERN_normal = "^${OEROOT}/meta/" Sep 16 17:52:23 # BBFILE_PATTERN_extras = "^${OEROOT}/meta/" Sep 16 17:52:23 # BBFILE_PRIORITY_normal = "5" Sep 16 17:52:23 # BBFILE_PRIORITY_extras = "5" Sep 16 18:02:47 freyther: ?= to = was just an accident, not intended Sep 16 18:04:12 kergoth: ah okay, I have added the '?' but I'm not sure if it was correct Sep 16 18:04:43 there are lots of cases in oe today of ?= being used when its completely useless anyhow (defaults defined in bitbake.conf anyway, so never undefined..) Sep 16 18:04:51 heh Sep 16 18:05:06 kergoth: yes but this case it was about the package_.bbclass stuff Sep 16 18:05:40 kergoth: I have seen, well I could have imagined it, that my 'insanity.bbclass' PACKAGE_FUNC was removed from the list Sep 16 18:06:24 ah Sep 16 18:07:01 insanity.bbclass and package_*.bbclass do inherit package and then manipulate PACKAGE_FUNC Sep 16 18:07:27 so if you have any clue on whether ?= or = is more correct, I would be glad... Sep 16 18:10:55 hell, i dunno. ask me after i've had some caffeine, i just woke up :P Sep 16 18:20:55 freyther: I think we should have a autobuilder that does binary search after an error Sep 16 18:21:34 koen: these kind of autobuilders are called 'werkstudent' Sep 16 18:28:11 freyther: it should spot things like this pretty quickly: http://www.openembedded.org/viewmtn/revision.psp?id=b24a11219bd17dda0e470cb32a2ac44c7a416f82 Sep 16 18:32:34 koen: ah yes, BUT :) Sep 16 18:34:01 freyther: what version of bitbake does .dev require nowadays? Sep 16 18:34:04 koen: Once we have SCM integration, we see which Csets are new since the last built Sep 16 18:34:04 koen: and it would be really easy to revert them Sep 16 18:34:25 1.4.4? Sep 16 18:34:54 freyther: do not revert, flag them and email author + cc oe-bugs@ Sep 16 18:36:42 koen: Just got back from shopping. Well, the compile did not complete. Result is again at http://oz.leggewie.org/granule/ Now I am utterly lost as to what the cause is. There are only a bunch of warnings about redefinition of functions it seems. I am not sure this should be fatal. Sep 16 18:37:55 I will try to get some servers I can abuse for oe building Sep 16 18:37:55 I think I can get two to four P4 machines Sep 16 18:38:42 * chouimat is away: Gone away for now. Sep 16 18:38:54 * chouimat is back. Sep 16 18:57:17 Laibsch: building granule right now Sep 16 19:08:35 Laibsch: CardBox.cpp:329: error: 'ERROR' is not a member of 'ASSA' Sep 16 19:16:45 Laibsch: if you remove all references to ASSA:ERROR granule will compile Sep 16 19:29:22 Laibsch: http://www.angstrom-distribution.org/repo/?action=details&pnm=granule :) Sep 16 19:50:15 pH5: wb Sep 16 20:03:24 koen: GREAT! Sep 16 20:03:38 Have you checked in the bb file? Or where can I get it? Sep 16 20:04:01 I cannot use your ipk unfortunately since I have only a collie. Sep 16 20:04:43 I just edited the source to remove all lines with ASSA:ERROR Sep 16 20:04:56 so I didn't create a patch Sep 16 20:05:13 the assa .bb is almost ready to go in Sep 16 20:06:46 I see. Sep 16 20:06:57 You think they are not necessary? Sep 16 20:07:12 I will create a patch and update th bb file. Sep 16 20:08:58 Laibsch: by the looks of it they just give error ouput Sep 16 20:09:14 Laibsch: so without them the user has less clues what when wrong Sep 16 20:21:14 hello all Sep 16 20:24:39 03koen 07org.oe.dev * r99ec3cff... 10/ (1 packages/gnome/libgnome_2.14.1.bb): libgnome 2.14: add esound to DEPENS Sep 16 20:39:20 freyther: I wondered when someone would comment on the logo :) Sep 16 20:39:40 freyther: The local.conf bit you pasted looks like it lacks a -extras :) Sep 16 20:40:02 yeeeees! bacula-fd is compiling and working :D what a nice way to backup my zaurus Sep 16 20:40:23 RP: oh yeah...kernel...I'll test that now Sep 16 20:41:06 JustinP: What did you think was faster out of interest? Sep 16 20:41:12 balle: where did the 'ccache -S' came from finally ? Sep 16 20:41:55 dont know. i added CCACHE = "" in the bb file Sep 16 20:42:18 balle: ok Sep 16 20:42:24 good to know Sep 16 20:42:26 RP: web and dates don' Sep 16 20:42:38 RP: don't like to be teleported. Dates exits with BadGlyphError after being teleported Sep 16 20:43:22 *pling* full backup finished :) Sep 16 20:43:36 freyther: This is something to mention to Cwiiis :) Sep 16 20:45:25 RP: yes, I will at one point Sep 16 20:45:45 RP: I plan to show poky/qemuarm+teleport on aKademy... Sep 16 20:45:58 RP: zaurusd I think seemed to respond faster....but it might have just been me Sep 16 20:47:02 freyther: cool :). Its otherwise working ok? Sep 16 20:47:18 JustinP: I'm not aware of any differences but that sounds good :) Sep 16 20:48:58 RP: ack! Sep 16 20:49:08 RP: first boot of new 2.6.17 kernel (using kexec in altboot) Sep 16 20:49:24 RP: Kernel panic - not syncing: VFS: Unable to mount root fs on unknown-block(0,0) Sep 16 20:49:53 RP: does the kernel magically now need a cmdline in kexec? Sep 16 20:49:57 Is that a release image rather than debug? Sep 16 20:50:10 * RP has heard this mentioned once before recently Sep 16 20:51:15 RP: I build my own stuff most of the time (and test tets images from hrw when I can) Sep 16 20:51:22 RP: this is built by me (no modifications) Sep 16 20:51:28 * JustinP tries the new 2.6.16 kernel Sep 16 20:52:00 JustinP: I wonder if it needs a delay before trying to mount he rootfs to give pcmcia time to get going? Sep 16 20:52:23 It does work for me... Sep 16 20:52:50 RP: I'm still compiling qemux86 Sep 16 20:52:50 RP: but teleport works okay to demonstrate it to developers Sep 16 20:52:50 RP: now if contacts would change the GUI if transfered to a 1204x768 display it would rock even more ;) Sep 16 20:53:45 RP: Once I test the new 2.6.16 (and flash if ok) I'll try the 2.6.17 again. Will try manual kexec. Also will rebuild to make sure it's ok. Sep 16 20:53:46 JustinP: rootdelay=5 on the commandline might help Sep 16 20:54:33 freyther: That would be nice - I'm not sure it would be a normal use case though :) Sep 16 20:55:23 hmm Sep 16 20:56:12 ~lart kergoth Sep 16 20:56:12 * ibot beats kergoth over the head with a microkernel Sep 16 20:56:13 *g* Sep 16 20:56:19 RP: It will be :) Sep 16 20:56:44 BTW: I think the appropriate name of GNU/HURD is Linux/GNU/HURD Sep 16 20:57:11 as HURD uses Linux drivers, filesystems and other stuff Sep 16 20:57:16 I should propose this to rms Sep 16 20:57:47 RP: Even if I have to design this hardware. Think of a coool gadget like an ipod Sep 16 20:58:12 RP: put a good battery, ARM, WLAN, WiMax, GSM/UMTS, Bluetooth, ZigBee module in Sep 16 20:58:29 RP: when compiling, right? Sep 16 20:58:45 RP: you store/cache your PIM Data, Documents, GnuPG keys, ssh keys there Sep 16 20:58:48 JustinP: Right, add to linux-openzaurus.inc Sep 16 20:59:11 RP: run all your X11 applications on the device and use teleport to beam them to your display you sit in front of Sep 16 20:59:25 freyther: I can see the attraction :) Sep 16 20:59:46 RP: "root=/dev/hda1 rootfstype=ext3 rw rootdelay=5" should work, correct? Sep 16 20:59:50 RP: With WiMax/UMTS you should always be able to synchronize your stuff to a backing system Sep 16 20:59:59 RP: it will be secure as your disk can be encrypted Sep 16 21:00:11 RP: if you lose your hardware you just get a new one and sync again Sep 16 21:01:17 JustinP: yes Sep 16 21:01:28 back in 2002 I have written Opie widgets that adapted Look 'n Feel according to the DesktopWidget size :) Sep 16 21:01:59 freyther: If you have ideas, please do talk to Cwiiis and others as I think they'll be interested :) Sep 16 21:02:31 * RP draws the line at apps - I have enough other bits to worry about ;-) Sep 16 21:03:07 RP: I will have to sort my thoughts, but I need to admit currently the GNOME platform kicks ass Sep 16 21:03:36 RP: specially dates+contacts are really promising Sep 16 21:06:32 freyther: I'd like to think so but I am biased ;-) Sep 16 21:07:22 RP: hmmm....a manual kexec from a booted environment (2.6.16) seems to be working. Ask, but altboot's sound "cat" is failing with 2.6.17 still. Sep 16 21:07:28 ibot: test Sep 16 21:07:36 Test Passed! Sep 16 21:07:40 ~seen lilo Sep 16 21:07:41 haha Sep 16 21:07:54 lilo is currently on #debian (5d 51m 45s) #asterisk (5d 51m 45s) #kde (5d 51m 45s) #handhelds.org (5d 51m 45s), last said: 'hi'. Sep 16 21:08:23 woglinde: hmm. i.mx allows Linux temp reel ;) Sep 16 21:09:30 woglinde: I need to get a D&C Moto Razr Sep 16 21:09:30 RP: Will mallum get a D&C moto razr? Sep 16 21:10:58 JustinP: Strange... Sep 16 21:11:18 freyther: D&C moto razr? Sep 16 21:11:44 rokr Sep 16 21:11:56 rokr e2 Sep 16 21:12:00 razr suckz Sep 16 21:12:15 RIP lilo :\ Sep 16 21:13:49 damn... Sep 16 21:14:04 I need to get my macbook fixed ASAP as this Centrino sucks... Sep 16 21:14:16 uodates Sep 16 21:14:22 check it Sep 16 21:15:18 Sad news about lilo :( Sep 16 21:15:59 RP: oh what did happen? Sep 16 21:16:24 freyther check channel 1 Sep 16 21:16:27 [Global Notice] On the 12th September Rob Levin, known to many as Freenode's lilo, was hit by a car while riding his bike. He suffered head injuries and passed away in hospital on the 16th. For more information please visit #freenode-announce Sep 16 21:17:23 :( Sep 16 21:17:43 yes, very sad Sep 16 21:18:02 RP: so my "alsa works" you mean that the code is all good but actual sound is still not working? Sep 16 21:18:03 hmm, I should get a helmet Sep 16 21:18:08 s/my/by/ Sep 16 21:18:17 JustinP: No, sounds works for me Sep 16 21:18:31 RP: hmmm, no sound for me... Sep 16 21:18:36 * JustinP checks the mixer again Sep 16 21:18:41 JustinP: As in I play a sound and it comes out the speaker/headphones Sep 16 21:18:43 ?? Sep 16 21:18:56 it wasn't working....then I paused and played again and it started working... Sep 16 21:19:25 well, working now Sep 16 21:19:31 :-) Sep 16 21:20:20 RP: volume still not working form other volume controllers, though, and altboot is still failing to play (cat fails...) Sep 16 21:20:38 going to reboot and try kexec from altboot again Sep 16 21:21:43 JustinP: Have the other volume controllers ever worked? Sep 16 21:22:55 RP: not in 2.6, no. It would be nice if they did. I'm testing the GPE volume applet and volume control in mpd. Neither work. Perhaps it's lack of a global volume control in alsa? Sep 16 21:23:05 RP: kexec from altboot dies again...will try flashing Sep 16 21:23:53 JustinP: I suspect something like that, or the OSS mixer emulation is bust Sep 16 21:24:42 Is it normal that "NOTE: package glibc-2.3.5+cvs20050627-r12: task do_populate_staging: started" takes 5+ Minutes in .dev? Sep 16 21:25:22 CoreDump|afk: IIRC yes.... Sep 16 21:25:46 just checking ;) Didn't try a .dev build in ages.. Sep 16 21:29:19 RP: cat: write error: Invalid argument Sep 16 21:29:33 RP: this is when altboot first starts up. It works in 2.6.16, not in 2.6.17 Sep 16 21:29:41 RP: well, reflashing with 2.6.17 worked...strange Sep 16 21:29:56 JustinP: very strange :-/ Sep 16 21:29:57 CoreDump|afk: altboot kexecing newest 2.6.17 strangely fails with a kernel panic Sep 16 21:30:09 JustinP: Did you try the rootdelay? Sep 16 21:30:17 RP: all versions of 2.6.17 I have tested display this error with altboot Sep 16 21:30:21 RP: your poky has 15 MB of RAM left. A GSM/UMTS server and WebCore should fit there ;) Sep 16 21:30:48 freyther_: cool :) Sep 16 21:30:59 RP: not with altboot, no. I'll grab my newest compile and try altbooting it. As I said before the new 2.6.17 kernel (without rootdelay) works fine when kexec'd from a normal boot and works fine when flashed. Sep 16 21:33:11 JustinP: very strange... Sep 16 21:33:52 indeed Sep 16 21:35:41 grrrrrr glibc is _still_ at do_stage wtf Sep 16 21:36:29 CoreDump|afk: hey Sep 16 21:36:39 hi freyther_ Sep 16 21:37:15 for 20 euros you can be a pirate Sep 16 21:37:17 hehe Sep 16 21:38:54 I thought I could be a pirate for using bittorrent. Sep 16 21:38:54 Yeah just 20 Euro a year and you get a PIRATE in the PIRATE PARTY! Sep 16 21:39:22 joshin: go check www.piratenpartei.de Sep 16 21:39:30 woglinde: oh you have to leave scientology first ;) Sep 16 21:39:32 I have. Sep 16 21:39:52 =D Sep 16 21:39:56 arrr Sep 16 21:43:12 RP: yep, rootdelay=5 allows it to be kexec'd within altboot Sep 16 21:45:23 JustinP: Interesting. Can you try rootdelay=2 ? (the times are in seconds) Sep 16 21:45:35 JustinP: We should probably allow a small delay by defaul Sep 16 21:45:38 t Sep 16 21:46:02 * JustinP rebuilds Sep 16 21:46:37 I'm going to bed now Sep 16 21:47:09 nite Sep 16 21:47:11 zecke Sep 16 21:47:15 me will follow Sep 16 21:47:23 'night zecke Sep 16 21:47:46 election day, today Sep 16 21:48:11 select we can not cast our vote for the pirate party today :( Sep 16 21:48:24 damn Sep 16 21:48:30 s/select/sadly/ Sep 16 21:48:40 we have to cast our vote for real criminals Sep 16 21:48:54 hmm s/criminals/politicians/ Sep 16 21:49:28 hmm last regexp is broken Sep 16 21:49:43 RP: yeah....cating a .raw to /dev/dsp doesn't work..... says invalid argument Sep 16 21:49:49 cannot exchange criminals and politicians Sep 16 21:49:54 RP: I can cat the raw file fine, though Sep 16 21:49:58 RP: write fails Sep 16 21:50:05 =) Sep 16 21:50:23 JustinP: I suspect the default sound parameters changed. Liam would be a better person to ask than me Sep 16 21:50:43 You have no idea how long it took me to find the sox magic to create that lovely "cat'able" raw file =) Sep 16 21:50:52 a "pirate" party in parliament would be really, really nice :) Sep 16 21:51:07 balle hehe :) Sep 16 21:51:28 balle be part of it Sep 16 21:51:48 CoreDump|afk: It might work with some ioctl magic Sep 16 21:52:39 woglinde, one question... are you the woglinde from ccc? Sep 16 21:52:40 hmm? The "cat" trick is supposed o work w/o any tricks. Even works on my notebook heh Sep 16 21:52:42 pH5: I've found your bitbake bug :) Sep 16 21:52:59 balle: hehe Sep 16 21:53:12 CoreDump|afk: I suspect it might be the 32 bit sounds changes. Does your notebook do 32bit? :) Sep 16 21:53:29 Chaos Congress is approaching :) Sep 16 21:53:35 no idea =) The CPU is 32bit yea Sep 16 21:53:36 yeah Sep 16 21:53:51 CoreDump|afk: The Zaurus doesn't either due to the i2S interface but the rest of the chain "supports" it so userspace is getting confused Sep 16 21:54:22 interesting Sep 16 21:54:53 CoreDump|afk: The best person to ask is lrg. I don't claim to understand it all ;-) Sep 16 21:56:13 RP: so...16 bit sound? Sep 16 21:56:22 balle: as of your ccache error. Without pasting a log to pastebin nobody can give an answer Sep 16 21:57:44 freyther_, i turned ccache off and submitted a bacula package to bugzilla, thats building. maybe someone else can find the error or leave it as it is :) Sep 16 21:58:20 balle: well, without a log nobody is capable of identifying anything Sep 16 21:58:39 balle: "I have some kind of error, but I don't say what the error is" won't get us far... Sep 16 21:58:50 yes but the error i pasted, was the only error i got Sep 16 21:59:06 freyther_: I have this line in taskData.py, it causes ph5's bug and I have no idea why I put it there... :) Sep 16 21:59:15 balle: I have missed a paste. All I saw was ccache, gcc fail with 'S' Sep 16 21:59:53 anyway, off to bed now Sep 16 22:00:04 i haven' t pasted it, but the error message wasn't saying anymore Sep 16 22:00:15 g'n8 freyther_ Sep 16 22:00:54 balle: in general. It would be interesting to know which package fails, which configuration you have, which versions of gcc where chosen and then paste the log that lead to the error Sep 16 22:01:05 RP: which line? Sep 16 22:01:34 freyther_: item = dataCache.pkg_fn[fn] Sep 16 22:01:45 line 394 Sep 16 22:03:03 item shouldn't change there if I understand what its doing correctly Sep 16 22:03:17 so why did I add that? :-/ Sep 16 22:03:28 balle yes I am member of the ccc e.v Sep 16 22:03:29 . Sep 16 22:03:37 RP: fn is not the filename but the name of the package? Sep 16 22:03:48 only ccc not ccc berlin Sep 16 22:04:09 freyther_: fn is the filename Sep 16 22:05:01 freyther_: I'll remove it as its wrong. If that breaks something else, so be it, I'll remember why it was there :} Sep 16 22:05:24 woglinde, good to know. me too. from www.chaostal.de Sep 16 22:05:35 hmm, I'm still not familiar with taskdata Sep 16 22:05:45 RP: ah okay we map from filename to PackageName Sep 16 22:06:01 RP: as we could have two files with the same package name Sep 16 22:06:09 woglinde, i guess you'll be in berlin on the next c3? Sep 16 22:06:22 balle I will be in Berlin Sep 16 22:06:23 freyther_: Yes. but task can and should be something like virtual/libsdl and we change it back to libsdl in this case... Sep 16 22:06:35 but dont know yet, If I will attend the congress Sep 16 22:06:47 RP: we have had task-bootstrap.bb und task-bootstrap-union Sep 16 22:06:50 nite all Sep 16 22:07:00 I have to get some sleep Sep 16 22:07:04 RP: and task-bootstrap-union set its PN to task-bootstrap Sep 16 22:07:31 freyther_: That sounds broken to me... Sep 16 22:08:13 RP: rootdelay=2 works Sep 16 22:08:14 RP: yes, I'm just guessing about the reasoning of the code Sep 16 22:08:34 freyther_: It must be something like that... Sep 16 22:08:38 RP: shall I try 1? Sep 16 22:08:52 JustinP: If you have time, that'd be useful :) Sep 16 22:09:17 JustinP: Since its marginal, even a small delay will perhaps work Sep 16 22:09:41 RP: oh well, as you add self.add_build_targe(fn, item) Sep 16 22:10:09 RP: and then have an empty line and mess with item, using my AI I can assume item = .. line is bogus Sep 16 22:10:30 freyther_: The ordering does look suspect, doesn't it... Sep 16 22:11:05 self.add_build_target(fn,item), item = pkg_fn[fn], self.mark_External_target(item) Sep 16 22:11:26 RP: it does look odd, but I'm not familiar with your code :( Sep 16 22:11:53 freyther_: keep in mind its in a loop as well so add_build_target can get called after item = pkg_fn[] as well Sep 16 22:12:14 freyther_: I feel I'm losing the plot... Sep 16 22:12:39 RP: why do you add all eligble items (add_tasks, add_buid_target) Sep 16 22:12:54 RP: but only mark the last eligble as external? Sep 16 22:13:34 I will call my first born, "FNID" it just sound cute... Sep 16 22:13:40 freyther_: It marks the item as external and item isn't supposed to have changed Sep 16 22:14:05 ah right, remove the line and see what breaks Sep 16 22:14:10 freyther_: but the mark_external code is compartively recent... Sep 16 22:14:33 03rpurdie * r612 10bitbake/lib/bb/taskdata.py: taskdata.py: Remove a seemingly incorrect line of code fixing issues reported by pH5. This code was likely added for a reason so lets see what breaks... Sep 16 22:15:26 RP: great :) Sep 16 22:15:45 freyther_: I profiled it and managed to speed it up significantly - 80% of the time was in one function :) Sep 16 22:16:34 RP: which one? Sep 16 22:16:44 80% that does sound much Sep 16 22:17:11 freyther_: I'm embarrased to say ;-) Sep 16 22:17:18 freyther_: matches_in_list() Sep 16 22:17:47 freyther_: I removed its main user as its a very slow function. I didn't realise it was being hit as often... Sep 16 22:17:59 Its a throwback from when taskdata had an SQL backend Sep 16 22:18:15 hehe Sep 16 22:18:17 RP: ok, rebooting with rootdelay=1 Sep 16 22:18:45 nite Sep 16 22:19:42 RP: yep, rootdelay=1 works fine too Sep 16 22:19:52 RP: looks like it's some very small timeout problem Sep 16 22:20:12 RP: I'm off to elsewhere, thanks for the working 2.6.17 :-) Sep 16 22:20:19 JustinP: great, thanks :) Sep 16 22:20:45 CoreDump|afk: perhaps try a 16 bit .raw? (I have no idea what sox magic you did or whether the .raw is correct) Sep 16 22:20:45 JustinP: I'll update the .inc - its always been slightly surprising it worked in the first place without that really! Sep 16 22:20:55 RP: heh. Sep 16 22:30:50 'night a;; Sep 16 22:30:55 all even Sep 16 22:36:32 .oO( jmp .night ) Sep 16 23:23:35 Anyone build irssi successfully lately? Sep 17 01:16:04 morning ~! **** ENDING LOGGING AT Sun Sep 17 02:59:57 2006