**** BEGIN LOGGING AT Sun Sep 10 02:59:56 2006 Sep 10 03:54:18 somebody had a question for me? Sep 10 03:54:27 [08:00] RP: I will pass that question on to mithro or mickeyl Sep 10 04:10:39 03lenehan 07org.oe.dev * r64c313f3... 10/ (1 packages/gawk/gawk_3.1.4.bb): gawk 3.1.4: Remove spurious .debug directories from the main packages. Sep 10 04:10:45 03lenehan 07org.oe.dev * r2be19e7c... 10/ (1 packages/gawk/gawk_3.1.4.bb): gawk 3.1.4: Run sanitize.py over the recipe. Sep 10 04:10:51 03lenehan 07org.oe.dev * r468b9951... 10/ (1 packages/gawk/gawk_3.1.4.bb): Sep 10 04:10:51 gawk 3.1.4: Use the update-alternatives class rather than manually Sep 10 04:10:51 performing the update-alternatives actions. Sep 10 04:21:32 03lenehan 07org.oe.dev * r6dc1b774... 10/ (1 packages/libxml/libxml2_2.6.22.bb): libxml2 2.6.22: Remove spurious .debug directories from the non -dbg packages. Sep 10 04:53:43 03lenehan 07org.oe.dev * r9069334d... 10/ (1 packages/postfix/postfix_2.0.20.bb): postfix 2.0.20: Remove spurious .debug directories from the non -dbg packages. Sep 10 05:45:51 * v8jlene is away Sep 10 06:09:16 good morning Sep 10 07:31:09 koen: here? Sep 10 08:18:06 hey koen Sep 10 08:18:22 koen: I updated the wiki to add BitTest: http://www.openembedded.org/wiki/BitTest Sep 10 08:19:26 koen: with a presentation similar to the GettingStarted Sep 10 08:20:07 good morning all Sep 10 08:29:33 Hi! Sep 10 08:29:42 hi psokolovsky Sep 10 08:30:01 cyrilRomain: Good work! Glad you with zecke stirred that stuff. Sep 10 08:31:44 psokolovsky: I'll try to integrate sanitization into the bittest content_checker Sep 10 08:39:13 hi zecke Sep 10 08:39:29 zecke: http://www.openembedded.org/bittest Sep 10 08:56:59 ~botmail for zecke: your tinderbox is broken, it reports 'red' for a build that succeeded Sep 10 09:18:57 mtn is slow today Sep 10 09:19:56 psokolovsky: you'll like the fact that 0.30 will give a huge performance boost on pull and commit :) Sep 10 09:20:21 koen: nice! Sep 10 09:40:16 koen: http://www.openembedded.org/bittest updated with bittest_single Sep 10 10:14:44 mtn: fatal: St11logic_error: database.cc:845: invariant 'I(cancel_size < pending_writes_size)' violated Sep 10 10:14:44 mtn: this is almost certainly a bug in monotone. Sep 10 10:14:46 oops Sep 10 10:28:44 psokolovsky: http://rafb.net/paste/results/Eoy9kk54.html Sep 10 10:32:06 koen: I see... Well, I restarted it once, and it continued from where it stops (123 revs), but it happened again and it's still at 123. Keeping trying. Sep 10 10:33:35 you'll have to rebuild mtn with '<=' instead of '<' Sep 10 10:33:57 I installed a build of head to get the speed improvements at the same time :) Sep 10 10:34:10 * koen is running 'mtn db regenerate_rosters' now Sep 10 10:34:53 http://rafb.net/paste/results/DbQOPD39.html for the curious Sep 10 10:38:57 http://www.openembedded.org/~koen/mtn-head-20060910.bz2 for the adventurous Sep 10 10:39:25 koen: what's size of that? Sep 10 10:39:35 1.8MB Sep 10 10:39:42 oh, I see Sep 10 11:09:30 koen: Downgrading to mtn 0.28 also helped Sep 10 11:33:48 koen|away: I doubt the tinderbox is broken, which build do we talk about? Sep 10 11:39:42 and my irssi is broken :( Sep 10 11:43:32 zecke: I have had an issue this morning with source_checker: http://rafb.net/paste/results/n2twIU63.html Sep 10 11:44:18 which OE branch is that? Sep 10 11:44:22 zecke: the workaround I used is to return False instead of raising the exception Sep 10 11:44:27 zecke: .dev Sep 10 11:44:40 * cyrilRomain checks its config file though Sep 10 11:45:29 03pfalcon 07org.oe.dev * re716e6a2... 10/ (1 classes/gpe.bbclass): gpe.bbclass: Provide default SECTION assignment. Sep 10 11:45:34 03pfalcon 07org.oe.dev * rd2fd513e... 10/ (1 packages/cbrpager/cbrpager_0.9.14.bb): cbrpager: Add SECTION. Sep 10 11:45:39 03pfalcon 07org.oe.dev * r8d322364... 10/ (3 files in 2 dirs): cherokee: Add SECTION. Sep 10 11:45:55 03pfalcon 07org.oe.dev * rda5b9fb5... 10/ (1 packages/dia/dia_0.94+0.95pre3.bb): dia: Add SECTION. Sep 10 11:45:59 http://rafb.net/paste/results/2MrmSE11.html Sep 10 11:46:00 03pfalcon 07org.oe.dev * r49b091d7... 10/ (1 packages/evince/evince_0.5.2.bb): evince: Add SECTION. Make DESCRIPTION more useful. Sep 10 11:46:05 03pfalcon 07org.oe.dev * r9d806f6a... 10/ (1 packages/freeciv/freeciv_2.0.8.bb): freeciv: Update SECTION. Sep 10 11:46:11 03pfalcon 07org.oe.dev * r94746608... 10/ (1 packages/gftp/gftp_2.0.18.bb): gftp: Add SECTION & DESCRIPTION. Sep 10 11:46:16 03pfalcon 07org.oe.dev * rf2880639... 10/ (5 files in 2 dirs): gimp: Add SECTION. Sep 10 11:47:00 cyrilRomain: I have tested md5sum checking using Qt2.3.10 Sep 10 11:47:33 cyrilRomain: I wonder if I ever comitted a wrong qt2.3.10 md5sum and I wonder if I didn't fix that up some time ago... Sep 10 11:48:02 cyrilRomain: please do md5 or md5sum on this tar and check if the first position is different Sep 10 11:49:01 zecke: 1f7ad30113afc500cab7f5b2f4dec0d7 qt-embedded-2.3.10-free.tar.gz Sep 10 11:49:32 zecke: is my tarball broken ? Sep 10 11:50:07 cyrilRomain: See above, I think I have provided a good explanation ;) Sep 10 11:50:43 cyrilRomain: 1f vs. af so I have pushed a wrong md5sum for QtE and never pushed the corrected number (it must bitrot on ewi) Sep 10 11:50:59 zecke: ok Sep 10 11:51:41 cyrilRomain: have you tried bittest_single as well? Sep 10 11:51:53 zecke: either way, the code should be corrected to catch the exception or simply return False Sep 10 11:52:09 zecke: yes, and added to the wiki _after_ I sent you the link :[ Sep 10 11:52:21 cyrilRomain: hehe, see my rafb.net paste for a patch Sep 10 11:54:20 zecke: ah, I also have a request for bittest Sep 10 11:54:54 zecke: I have noticed the report is sent only at the end of the test. Is it possible to send it by package ? Sep 10 11:55:44 zecke: like build report in fact Sep 10 11:57:06 zecke: bittest_single is very easy to use and is good to introduce bittest :) Sep 10 11:57:50 njs: Have you seen this: mtn: Fataler Fehler: St11logic_error: database.cc:845: Invariante 'I(cancel_size < pending_writes_size)' verletzt Sep 10 11:58:40 cyrilRomain: currently this is not possible Sep 10 11:59:35 zecke: ok, np Sep 10 12:00:57 * cyrilRomain now looks at content_checker to integrate sanitize Sep 10 12:02:06 zecke: btw, why HOMEPAGE and LICENSE are set to "unknown" in bitbake.conf ? why not simply remove them ? Sep 10 12:03:16 zecke: so that we only deal with getVar returning None (not None _and_ unknown) Sep 10 12:06:01 zecke: http://rafb.net/paste/results/Eoy9kk54.html Sep 10 12:06:47 koen|away: hehe Sep 10 12:07:04 update to head or downgrade to 0.28 Sep 10 12:07:06 koen: which build should be green but is red? Sep 10 12:07:19 the one from a few hours ago Sep 10 12:07:28 'bootstrap-image' succeeded, but is red Sep 10 12:07:31 which is 'the one' Sep 10 12:07:38 maybe the tinderbox doesn't like bitbake trunk Sep 10 12:07:56 koen: How often http://www.angstrom-distribution.org/repo/repo.php is updated? Sep 10 12:08:13 http://ewi546.ewi.utwente.nl/tinderbox/showlog.pl?machine_id=111&logfile=20060910105312.log Sep 10 12:08:15 koen: no, certainly tinderbox does not like multi threaded builds :) Sep 10 12:08:24 psokolovsky: everytime I add something to the feeds Sep 10 12:08:45 ok Sep 10 12:08:57 psokolovsky: it helps to bump PR for a section change :) Sep 10 12:09:09 koen: I do. Sep 10 12:09:57 koen: was thinking if to do or not, but decided, that it better make it propogate faster, even if require extra rebuilds ;-( Sep 10 12:10:58 psokolovsky: if the resulting .ipk will be different -> bump PR Sep 10 12:11:07 ok Sep 10 12:20:59 oh joy Sep 10 12:21:00 mtn: fatal: St11logic_error: database.cc:845: invariant 'I(cancel_size < pending_writes_size)' violated Sep 10 12:21:03 mtn: this is almost certainly a bug in monotone. Sep 10 12:21:14 NAiL: scroll up Sep 10 12:22:09 downgrade to 0.28? Sep 10 12:23:05 NAiL: use mtn 0.28 Sep 10 12:24:18 mmkay, thanks Sep 10 12:25:07 zecke: your patch almost worked: s/TestItem.generate_error/TestCase.generate_error/ ;) Sep 10 12:26:53 hmmm Sep 10 12:26:57 hehe Sep 10 12:27:10 how do I turn the newly debugging spew of in bitbake trunk? Sep 10 12:30:35 zecke: btw: http://www.openembedded.org/sources/bitbake-qa-0.1.tar.bz2 Sep 10 12:30:48 hehe Sep 10 12:31:16 koen: I had to laugh about Berlios and Disk-Space as I had to wait a week until the SVN had space again Sep 10 12:31:24 damn Sep 10 12:31:31 until the SVN disk had space again Sep 10 12:31:41 I had 'loads of space' in the first draft :) Sep 10 12:31:52 then I remembered about the svn issue Sep 10 12:32:36 ouch, qemu breaks the source_checker: http://rafb.net/paste/results/202tPQ82.html Sep 10 12:33:18 and then bittest become crazy Sep 10 12:33:42 HMM Sep 10 12:33:59 hmm, likely to be a missing variable Sep 10 12:34:08 but it should not go crazy Sep 10 12:34:40 zecke: is multithreaded code called from bittest ? Sep 10 12:34:42 koen: Once I have an idea where to put it, I will do svk smerge -I to the new home Sep 10 12:34:59 cyrilRomain: maybe, I have not tried trunk Sep 10 12:35:07 cyrilRomain: but only patch checker could call it Sep 10 12:38:16 morning Sep 10 12:38:32 zecke: hum.. I tried with packages/qemu/*.bb and choose 'temporary' and it worked ... Sep 10 12:39:13 hmm Sep 10 12:39:35 cyrilRomain: So we have two issues here: Auto Answering prompts and BitTest stopping to work after errors? Sep 10 12:40:14 zecke: no I Ctrl-C after the lines that say 'could not read file' (that actually exists) Sep 10 12:40:27 but auto answering prompt yes Sep 10 12:40:59 --non-interactive Sep 10 12:41:13 for svn co sounds right, we should update the FETCHCOMAMND_svn Sep 10 12:42:22 yep Sep 10 12:44:42 damn, the wrong key again Sep 10 12:45:13 zecke: why HOMEPAGE and LICENSE are set to "unknown" in bitbake.conf ? why not simply remove them ? Sep 10 12:45:50 right, we can remove it from bittest Sep 10 12:46:02 I would not remove it from bitbake.conf in OE Sep 10 12:46:28 zecke: ah, but we still need in bittest to determine when a variable is not set Sep 10 12:46:31 mickeyl: have you seen the request of help? Sep 10 12:46:38 no Sep 10 12:46:47 zecke: so we need to test for None and 'unknown' Sep 10 12:46:54 mickeyl: when people do import msg Sep 10 12:47:32 mickeyl: msg will be always loaded but we want to have a global/static variable Sep 10 12:47:36 hey mickeyl Sep 10 12:47:54 zecke: in python ? Sep 10 12:47:58 * mickeyl lacking context Sep 10 12:48:15 mickeyl: ah okay. We have a bb.msg module Sep 10 12:48:34 mickeyl: and one is able to set a debug level and we would like to set the debug level globally/statically Sep 10 12:48:48 mickeyl: but globals() __builtins__ are scoped by module Sep 10 12:49:03 why not putting it into the module scope directly? Sep 10 12:49:04 mickeyl: and we don't want to pass an object/instance to bb.msg.note Sep 10 12:49:11 a module is acting perfectly as a global static variable. Sep 10 12:49:22 mickeyl: okay again Sep 10 12:49:25 bb.msg.setLevel(foo) Sep 10 12:49:26 ? Sep 10 12:49:42 mickeyl: yes, but we will end up with many different levels Sep 10 12:49:49 mickeyl: and someone doing import msg Sep 10 12:49:56 mickeyl: will have a level != foo Sep 10 12:53:14 mickeyl: BTW: http://rafb.net/paste/results/NaHW1k63.html Sep 10 12:53:51 mickeyl: do CONFIG+=final and compile with -funit-at-a-time and your binary will be smaller, carry less symbols and be generally faster/optimized Sep 10 12:54:13 wicked. that works? Sep 10 12:54:31 mickeyl: what part of that? Sep 10 12:54:49 mickeyl: if you do not have clashing static variables/method names Sep 10 12:54:58 (the -funit-at-atime9 Sep 10 12:54:59 ) Sep 10 12:55:04 interesting. Sep 10 12:55:22 btw., we should just enforce fully qualified imports Sep 10 12:55:37 being able to import bb.msg and msg is a weakness in python imo Sep 10 12:56:01 ah and then it will be imported exactly once? Sep 10 12:56:05 yes Sep 10 12:56:16 we can probably even catch non-fully qualified imports by checking __file__ Sep 10 12:56:20 or __name__ Sep 10 12:57:06 zecke: -funit-at-a-time and your include only affect C++ programs, right? Sep 10 12:59:15 no Sep 10 13:00:24 it is a normal gcc option Sep 10 13:00:51 if you throw your complete program in one file to it, it in theory can change function invocations etc Sep 10 13:00:59 from my tests the code is at least smaller Sep 10 13:03:38 cyrilRomain: can you push to OE? Sep 10 13:11:12 zecke: no Sep 10 13:21:13 zecke: http://projects.linuxtogo.org/frs/?group_id=14 Sep 10 13:22:30 03freyther 07org.oe.dev * r6519d461... 10/ (1 packages/qte/qte-common_2.3.10.inc): packages/qte/qte-common_2.3.10.inc: Correct the md5sum Sep 10 13:26:10 hey mikearthur_home Sep 10 13:26:23 hey koen Sep 10 13:26:34 thats actually my laptop, just about to disable autostart Sep 10 13:26:35 s Sep 10 13:29:51 morning Sep 10 13:41:12 hi all ... I'm looking for Nils Faerber - does anyone know his IRC nick? Sep 10 13:43:13 pgfeller: 'nilsf', but he's rarely on IRC Sep 10 13:43:43 koen: thanks; then I'try by e-mail - but is good to know anyhow. Sep 10 13:43:58 ~seen nilsf Sep 10 13:44:02 nilsf <~nils@dns.kernelconcepts.de> was last seen on IRC in channel #handhelds.org, 407d 19h 26m 9s ago, saying: 'koen: Hmm... ;)'. Sep 10 13:46:20 psokolovsky: I'll put some packages with fixed sections in the angstrom feeds in ~30 minutes Sep 10 13:46:43 koen: more on the way ;-) Sep 10 13:54:14 03pfalcon 07org.oe.dev * r15b56950... 10/ (3 files in 2 dirs): xchat: Add SECTION & DESCRIPTION. Typo fixes. Sep 10 13:54:19 03pfalcon 07org.oe.dev * rab12a514... 10/ (1 packages/qiv/qiv_1.9.bb): qiv: Add SECTION. Sep 10 13:54:26 03pfalcon 07org.oe.dev * ra30dc5a2... 10/ (1 packages/rox/rox-filer_2.5.bb): rox-filer: Add SECTION. Sep 10 13:54:32 03pfalcon 07org.oe.dev * r28b38c05... 10/ (1 packages/nogravity/nogravity_2.0.bb): nogravity: Add SECTION & DESCRIPTION. Sep 10 13:54:37 03pfalcon 07org.oe.dev * rf659bb6b... 10/ (1 packages/intercom/intercom_0.15.bb): intercom: Add SECTION. Sep 10 13:54:50 03pfalcon 07org.oe.dev * rb04966a0... 10/ (1 packages/horizon/horizon_cvs.bb): horizon: Add SECTION. Sep 10 13:55:01 03pfalcon 07org.oe.dev * r2ac38518... 10/ (1 packages/gomunicator/gomunicator_cvs.bb): gomunicator: Add SECTION. Sep 10 13:55:07 03pfalcon 07org.oe.dev * rbed2795f... 10/ (5 files in 2 dirs): epdfview: Add SECTION. Sep 10 13:55:12 03pfalcon 07org.oe.dev * r62363ab6... 10/ (3 files in 2 dirs): Add SECTION & DESCRIPTION. Sep 10 13:58:57 03pfalcon 07org.oe.dev * r17328d82... 10/ (1 packages/dejagnu/dejagnu_1.4.4.bb): dejagnu: DESCRIPTION: Add note of (add your epithet) dejagnu guts. Sep 10 14:05:48 hrw: thanks for the line about cardBus two days ago Sep 10 14:05:59 stired me in the right direction Sep 10 14:20:19 psokolovsky: http://www.angstrom-distribution.org/repo/?action=details&pnm=intercom Sep 10 14:32:15 ok, first integration of sanitize.py into bittest content_checker succeed Sep 10 14:32:43 koen: cool, thanks! Sep 10 14:33:44 it automatically creates .sanitized files. For example when checking lua_5.0.2.bb, lua_5.0.2.bb.sanitized is created Sep 10 14:35:20 and when rules that are not respected, a warning or error is dumped in the test report in 'Test Comment' field Sep 10 14:47:13 example: http://ewi546.ewi.utwente.nl/tinderbox/showlog.pl?machine_id=113&logfile=20060910170215.log Sep 10 14:50:56 cyrilRomain, looks scary ;-). You really should convince zecke to add bittest to main BB distribution. Unless there will be "bittest" command like "bitbake", it still will be underused. Sep 10 14:53:15 psokolovsky: yes. Bittest was available on svn two days ago. and disappear in part because of my reinvent-the-wheel sanitize script Sep 10 14:54:01 psokolovsky: I hope zecke will make it available again Sep 10 14:54:17 psokolovsky: I also added a wiki page Sep 10 14:54:49 cyrilRomain: well, there're much stuff in BB repo ;-). the point is to make it go into distro. I hope that will happen gradually. Sep 10 14:55:21 cyrilRomain, yes, I saw. will hope to go over it when time permits Sep 10 14:56:48 * cyrilRomain think about koen's idea to automatically mail MAINTAINER on .bb bad content Sep 10 15:00:33 cyrilRomain: lol Sep 10 15:11:40 psokolovsky: you can get bittest from http://projects.linuxtogo.org/frs/?group_id=14 or http://www.openembedded.org/sources/bitbake-qa-0.1.tar.bz2 Sep 10 15:12:08 koen: ok Sep 10 15:12:55 what smtp server would you use to do that ? Sep 10 15:15:07 cyrilRomain: call 'mail' on the host? Sep 10 15:16:05 through os.system ? Sep 10 15:16:27 gosh I don't have mail command Sep 10 15:17:25 koen: I'm trying to use the smtplib python module, but get 'connection refused' Sep 10 15:18:47 ah, the ciabot has smtpserver=localhost Sep 10 15:20:01 I tried to do that, but it fails on smtp = smtplib.SMTP('localhost') :( Sep 10 15:20:33 hum.. my firewall should not block it. Sep 10 15:20:49 I'm wondering if i need a smtp daemon or something like that Sep 10 15:21:09 ~lart me for ignoring so much about mail protocols and tools Sep 10 15:21:10 * ibot whips out a hot clue gun and makes sure that cyrilromain is stuck to the floor for ignoring so much about mail protocols and tools Sep 10 15:36:45 ah, installing send mail and starting daemon solved it :) Sep 10 15:40:23 great it works :) We can even put the sanitized file in attachement :) Sep 10 15:40:50 cute Sep 10 15:57:01 hey people Sep 10 16:00:43 hey mithro Sep 10 16:18:47 'morning' all Sep 10 16:19:13 yo RP Sep 10 16:19:22 NOTE: package hentges-gpe-image-oz3.5.4.1-gpe2.7: completed =D Sep 10 16:33:16 anyone know if there is a way to tell linux to not swap out certain applications? Sep 10 16:34:56 (or rather swap out at last choice) Sep 10 16:35:04 mithro: It seems there is something called mlockall() Sep 10 16:35:33 mickeyl: still around? Sep 10 16:35:41 RP: is there a way to apply that to applications kind of like nice? Sep 10 16:35:51 we need a "swap nice" type thing Sep 10 16:36:34 mithro: I don't know - various scripts seem to exist Sep 10 16:37:28 mithro: google on mlockall and noswap Sep 10 16:42:10 mithro: It looks like they broke inheritance of the flag so it might not be possible to set it for a running process in recent kernels :-/ Sep 10 16:42:25 :/ Sep 10 16:47:03 Is there a simple way to do the equivalent of: OPENSLUG_STANDARD_RDEPENDS -= to exclude packages included by default? Sep 10 16:48:34 += works but I'd like -= so that I can just edit my local.conf to remove kernel modules I don't want without editing the openslug.conf (or openzaurus.conf) file Sep 10 17:13:24 -= would be a welcome addition IMO Sep 10 17:13:39 but there are subtle issues that need to be considered here Sep 10 17:13:48 * mickeyl notes for the BitBake session @ OEDEM Sep 10 17:14:19 rewrite bitbake to use 9pfs :) Sep 10 17:15:17 9pfs ? Sep 10 17:15:59 chouimat: the plan9 stuff Sep 10 17:16:27 koen cool Sep 10 17:16:40 do you know the movie "plan9 from outer space"? Sep 10 17:16:51 since 2.6.15-ish linux supports it as well Sep 10 17:16:55 mickeyl: never seen it Sep 10 17:16:58 it's hilarious Sep 10 17:17:03 it's sooo bad Sep 10 17:17:16 but funny at the same time Sep 10 17:17:30 ~bon appetit Sep 10 17:17:36 somebody said bon appetit was smacznego. Guten Appetit. Eet Smakelijk. God Appetitt. Buon Appetito. Buen apetito Bom Apetite. buen apetito Sep 10 17:17:54 mickeyl: a friend of mine has a nice game to play while watching the movie ... each time you see a flying saucer landing or taking off ... you drink a shooter of vodka :) Sep 10 17:22:18 chouimat: http://www.stil.by/images/606.jpg Sep 10 17:22:52 koen hehe Sep 10 19:11:24 mickey|dinner: If you have a moment sometime, I'd like to run some ideas about shell.py in trunk past you Sep 10 19:11:57 I'm here for a couple of minutes, go ahead Sep 10 19:12:27 mickeyl: How much have you been following trunk? (the taskdata/runqueue changes) Sep 10 19:12:39 RP: I'm afraid not closely Sep 10 19:13:00 i roughly looked at the multithreading Sep 10 19:13:12 mickeyl: Basically, the shell could take advantage of them. Its now the only user of buildProvider and I'm therefore thinking of moving that code into the shell Sep 10 19:13:20 I'll show you my current diff Sep 10 19:14:24 http://www.rpsys.net/openzaurus/temp/bitbake_shell_runqueue.patch Sep 10 19:15:00 My concern is the exception handling as I don't think I have this code right... Sep 10 19:15:55 cooker.configuration.abort is " ! -k on command line" ? Sep 10 19:16:14 mickeyl: I think so, yes Sep 10 19:16:40 action = store_false, so yes Sep 10 19:17:06 That was to fix a slight flaw in runQueue with the shell :) Sep 10 19:18:12 The shell does already use the runqueue/taskdata via buildProvider but we no longer need buildProvider in bitbake and each user can optimise its usage better if they call runQueue directly Sep 10 19:18:18 ok, i have no idea about the TaskData entity, but other than that it looks good on first glance. Perhaps we need to add more except's at the end of the thing that formerly was buildProviders, but we'll see which errors are thrown sooner or later Sep 10 19:18:53 mickeyl: Are you ok with my committing that and then we can clean up any exceptions we need to? Sep 10 19:19:05 of course. buildProvider was _the_ high level abstraction for building a provider. now that the task queue is there, it's the preferred thing Sep 10 19:19:07 ya, go ahead Sep 10 19:19:13 The exception handling is my main concern :) Sep 10 19:19:31 i'm going to update one of my workstations to trunk and play with it Sep 10 19:19:55 It doesn't help that the bottom end of runqueue is only hald finished but it has the right idea and the calling conventions won't change Sep 10 19:20:10 Playing with it would be appreciated. I' Sep 10 19:20:14 I'm sure it has bugs :) Sep 10 19:20:47 i like it. the shell | cli border was more or less totally arbitrary when I started the shell. it gets more refined now and eventually Sep 10 19:21:03 we have a real kind of API Sep 10 19:21:05 for bb Sep 10 19:21:13 That's my aim - small modules for things Sep 10 19:21:22 fetchers, providers, taskdata, runqueue Sep 10 19:21:35 and the cache :) Sep 10 19:21:38 yeah Sep 10 19:21:48 I just destroyed BBParsingStatus :-/ Sep 10 19:21:53 oops ;) Sep 10 19:22:17 Its become cacheData which makes more sense given how it was being used Sep 10 19:22:17 i also can foresee splitting the shell into individual modules implementing commands that can be used by a) the terminal shell and b) some forthcoming GUI Sep 10 19:22:47 That sounds useful, maybe as a commands module Sep 10 19:23:06 It'd be worth you seeing whether the shell would benefit from keeping taskdata around Sep 10 19:23:30 taskdata is kind of clever as you can append metadata to it and it shouldn't break Sep 10 19:24:14 runqueue is different and is a once per build entity Sep 10 19:24:42 ah, i see. that's cool. perhaps with all of your changes the shell stops being an artificial poker in the BB state. i always felt this to be _very_ hackish. Sep 10 19:25:15 mickeyl: I just see it as a different source of commands Sep 10 19:25:43 mickeyl: something like 'mtn automake stdio' for the bitbake shell? Sep 10 19:25:56 automate* Sep 10 19:25:56 mickeyl: We lack some larger configuration container. I'm not happy about the use of cooker in places... Sep 10 19:26:13 koen: ya. but without the overhead of calling it as a binary Sep 10 19:26:34 I'm also aware cooker.status is totally misnamed now (its really a cacheData entity) Sep 10 19:27:12 mickeyl: or via adaptive middleware in mobile systems? Sep 10 19:27:31 koen: heh. that's a good name. where did you stumble over that? ;) Sep 10 19:27:37 RP: that's right. we might need one BuildConfigurator and one SessionConfigurator Sep 10 19:27:38 :D Sep 10 19:27:57 mickeyl: I guess the next big change would be to refactor cooker into some kind of bitbake status entity which either the commandline or shell could use Sep 10 19:28:16 mickeyl: BuildConfigurator is basically what taskData Sep 10 19:29:32 It generates the task list and keeps track of failed tasks. You then generate a runqueue to say which entities out of taskdata to build Sep 10 19:29:33 right Sep 10 19:29:43 . o O (bitbake buildfarms) Sep 10 19:29:46 so we could add a config dict to the taskData ? Sep 10 19:30:09 mickeyl: I see that as being more part of SessionConfigurator Sep 10 19:30:14 RP: you could even write a 'bbrsh' for do_configure :) Sep 10 19:30:33 mickeyl: We need something taskdata would be a part of :) Sep 10 19:30:39 RP: ya, right Sep 10 19:30:53 mickeyl: It would also contain cacheData Sep 10 19:31:25 mickeyl: Basically cooker renamed and restructured? Sep 10 19:31:35 RP: yes Sep 10 19:31:48 we might take out one or the other entry then Sep 10 19:32:24 mickeyl: As an illustration of the relationship between runqueue and taskdata, look at what happens when a task fails (and abort isn't set) Sep 10 19:32:54 mickeyl: Basically, runqueue marks the task as failed in taskdata then builds a new queue :) Sep 10 19:33:00 slick! Sep 10 19:33:16 mickeyl: It does all this transparent to the user (be it the shell or the commandline) :) Sep 10 19:33:20 did you rip out all of the (bogus) task graph stuff? Sep 10 19:33:53 mickeyl: We still use that for the tasks at the file level :-/ Sep 10 19:34:35 mickeyl: With some work, we could remove it though and parse it into a form useable by taskdata directly though Sep 10 19:35:04 RP: yeah. at this point, i'm not sure whether it would overcomplicate the taskdata too much Sep 10 19:35:49 taskdata needs to be fed a list of tasks and their dependencies. Currently, it reads the taskgraph for a given file Sep 10 19:36:14 Whether we could do this differently, I'm not sure. I know a lot of the taskgraph functions are unused Sep 10 19:36:30 yeahttp://www.pathname.com/~quinlan/software/diffsplit/diffsplit Sep 10 19:36:32 ups Sep 10 19:36:33 yeah, even Sep 10 19:37:09 ~seen mickeyl Sep 10 19:37:25 mickeyl is currently on #oe (1d 2h 40m 7s) #openezx (1d 2h 40m 7s) #opie (1d 2h 40m 7s). Has said a total of 73 messages. Is idling for 51s, last said: 'yeah, even'. Sep 10 19:37:30 mickeyl: The main thing is it works atm :) Sep 10 19:37:39 mickeyl: cool, I was looking for diffsplit a while ago :) Sep 10 19:37:40 RP: yeah Sep 10 19:37:51 koen: i need it on my slug Sep 10 19:37:55 koen: will add it to OE Sep 10 19:38:01 hi Sep 10 19:38:04 hey hrw Sep 10 19:38:05 evening hrw Sep 10 19:38:08 mickeyl: do you have a few minutes to discuss something regarding the libvncserver package in oe? Sep 10 19:38:12 mickeyl: and add it to the patch class? Sep 10 19:38:20 koen: what for? Sep 10 19:38:24 mickeyl: I am hoping to really thin out bin/bitbake itself even more. We've moved a lot of code into modules but we can do better. Sep 10 19:38:44 pgfeller: yeah, go on. 7 minutes left, then i need to spend the rest of the evening with my wife :) Sep 10 19:39:00 mickeyl: Next on the hit list is the dependency graph generation code zecke wrote. I'd like that to use taskdata for consistency Sep 10 19:39:12 mickeyl: greet Sabine then ;) Sep 10 19:39:17 Once we clean that up, bin/bitbake should be pretty empty :) Sep 10 19:39:19 hrw: will do, thanks Sep 10 19:39:37 RP: yeah that'd be nice. we should add a couple of diagrams how the parts interact to the bb manual eventually Sep 10 19:39:39 mickeyl: would it be possible to split the .bb for libvncserver & x11vnc? Sep 10 19:39:46 mickeyl: sabine is coming to OEDEM as well? Sep 10 19:40:00 RP, mickeyl, koen: I have another thing to discuss on/before oedem Sep 10 19:40:11 mickeyl: as I would like to maintain the libvnc part - but I lack the skills/knowledge to take care of x11vnc Sep 10 19:40:16 pgfeller: it could be possible and for me it even has a sense Sep 10 19:40:19 koen: it's undecided yet. Sep 10 19:40:47 pgfeller: sure, we can split it. it was just in one recipe, because x11vnc is built in the same process Sep 10 19:40:57 no problem with doing an .inc for both Sep 10 19:41:07 hrw: what do you want to discuss on/before? Sep 10 19:41:15 pgfeller: remember about sdlvnc which is in it too? Sep 10 19:41:57 mickeyl: usage of @openembedded.org aliases, 'OpenEmbedded' name, logo on businesscards Sep 10 19:42:05 hrw: yep ... I would like to reduce my responsibility to libvncsever - as this is what I understand @ the moment Sep 10 19:42:31 hrw: *nod* that fits into two sessions Sep 10 19:42:50 mickeyl: I started own business finally (will write about it to ML during week) and need to print some businesscards Sep 10 19:43:18 OpenHRW ? Sep 10 19:43:27 HaeRWu Sep 10 19:43:47 hrw: since there is no tradename or anything neither on the name nor on the logo, you're free to use it. Sep 10 19:43:49 * koen starts OpenUmbaUmba Sep 10 19:43:51 koen: I avoid 'Hrw' in name because of Human Rights Watch organization Sep 10 19:44:47 hrw: we should add both topics to the schedule for "OE and commercial" section Sep 10 19:44:50 need to run now Sep 10 19:44:56 cu mickeyl Sep 10 19:45:10 mickeyl: thanks and cu :) Sep 10 19:45:22 RP: can't wait to try your work on bitbake et. al. it's really shaping! Sep 10 19:45:27 koen: HaeRWu is my preferred spelling of Hrw Sep 10 19:45:28 * mickeyl waves Sep 10 19:45:53 hrw: congrats on the business. We really need to talk about several commercial aspects at oedem :) Sep 10 19:47:07 Hmm. I also need to go :- Sep 10 19:47:10 :-/ Sep 10 19:47:26 I think zecke and I will use the phrase 'filthy rich' a lot at OEDEM ;) Sep 10 19:47:31 ;)) Sep 10 19:48:19 * RP wished own business == filthy rich Sep 10 19:54:11 RP: you are not allowed to work on OE for other companies then ohand iirc.. Sep 10 19:54:14 * joshin agrees Sep 10 19:54:57 joshin: to? Sep 10 19:55:05 to rp's wish Sep 10 19:55:09 ah Sep 10 19:55:50 hi woglinde Sep 10 19:56:02 hi hrw Sep 10 19:56:15 *sigh* no time for nothing Sep 10 20:00:42 woglinde: true, true Sep 10 20:02:33 koen: PING Sep 10 20:02:50 mtn: error: to use this workspace with this version of monotone, its metadata Sep 10 20:02:57 .. Sep 10 20:02:58 mtn: error: with versions of monotone older than 0.30. Sep 10 20:03:07 monotone 0.29 (base revision: format_version "1" Sep 10 20:03:20 koen: so.. migrate_workspace or not? Sep 10 20:13:02 I'm getting tired of monotone's incompatibility with earlier versions of itself Sep 10 20:14:15 hehe Sep 10 20:14:26 thats what software is about Sep 10 20:14:43 * joshin wonders if he worked with woglinde at MSFT Sep 10 20:14:53 hm no Sep 10 20:15:03 you sure have the mindset down. :p Sep 10 20:17:14 hrw: yes, complete all the 3 migrate steps it tells you to Sep 10 20:17:47 joshin: mtn 0.30 works perfectly with older versions Sep 10 20:19:01 ewi546 is running 0.30 and syncing perfectly with monotone.openembedded.org and my home machines (0.28 and 0.29) Sep 10 20:19:27 where is 0.30? Sep 10 20:19:44 Crofton: not released yet Sep 10 20:19:50 heh Sep 10 20:20:06 I was looking on their web site and wondering what you are talking about :) Sep 10 20:20:13 njs build a static binary from mtn head for me today Sep 10 20:21:16 hrw: you'll notice that commit and update are *way* faster Sep 10 20:21:25 2ok Sep 10 20:28:50 koen: too bad that migration will take ~hour Sep 10 20:29:26 Crofton: you have to signup as a beta tester, heh Sep 10 20:29:27 yep, it's like an initial pull Sep 10 20:30:04 hrw: http://www.openembedded.org/~koen/stats.htm Sep 10 20:30:30 that's for .dev Sep 10 20:32:36 koen: select author, count(*)? Sep 10 20:32:57 mtn log | sort | uniq, etc Sep 10 20:33:27 nite Sep 10 20:33:39 and I discovered that ~10% of the revs in .dev are merges Sep 10 20:33:49 if I read the output right Sep 10 20:37:47 hrw: I'm not sure where I'd stand on that actually :) Sep 10 20:38:43 RP: ah Sep 10 20:41:45 551 jbowler@nslu2-linux.org Sep 10 20:41:45 1000 mickeyl@openembedded.org Sep 10 20:41:45 1195 hrw@openembedded.org Sep 10 20:41:45 1869 koen@openembedded.org Sep 10 20:41:52 sqlite> select count(name) as ile, value from revision_certs where name='author' group by value order by ile; Sep 10 20:42:09 koen: much faster then mtn log ;) Sep 10 20:42:16 yeah Sep 10 20:42:30 I wanted to add in the bk stats as well Sep 10 20:42:46 http://oe-devel.bkbits.net:8080/openembedded/stats?nav=index.html Sep 10 20:43:04 btw - nice to see that I;m still in top10:) Sep 10 20:43:24 ~lart winXP for being unusable on 256M of ram Sep 10 20:43:25 * ibot shoves a crumpet down winXP's throat, happy now?! Huh? Want some JAM with that? for being unusable on 256M of ram Sep 10 20:44:06 * RP is well down the list :) Sep 10 20:46:25 * JoeLaptop hugs ibot for saying that Sep 10 20:47:24 shop was out of 256M SO-DIMM sticks ;( Sep 10 20:54:54 03rpurdie * r596 10bitbake/lib/bb/runqueue.py: runqueue.py: Upon failure in abort mode, raise an exception rather than sys.exit Sep 10 20:55:01 wao, downloading all the source takes days ... Sep 10 20:56:46 03rpurdie * r597 10bitbake/lib/bb/msg.py: msg.py: Each -l option increases the debug level of that logging domain by 1 (similar to -D behaviour). Sep 10 21:00:51 03rpurdie * r598 10bitbake/ (bin/bitbake lib/bb/runqueue.py lib/bb/shell.py): Sep 10 21:00:51 lib/bb/shell.py - Use taskData and runQueue directly Sep 10 21:00:51 lib/bb/runqueue.py - Add fn to TaskFailure Sep 10 21:00:51 bin/bitbake - Remove now uneeded buildProvider Sep 10 21:01:50 03rpurdie * r599 10bitbake/lib/bb/cache.py: cache.py: Drop uneeded BUILD_ALL_DEPS variable Sep 10 21:02:03 cu Sep 10 21:02:14 'night hrw|gone Sep 10 21:03:07 good night hrw|gone Sep 10 21:03:33 RP: is BUILD_ALL_DEPS not valid anymore in bb files ? Sep 10 21:03:53 cyrilRomain: bitbake trunk no longer cares about it Sep 10 21:04:08 RP: I still get tons of debugging spew without '-l' Sep 10 21:04:14 RP: ok, I should take that into account in the content checker then Sep 10 21:04:38 koen: For where - commandline, shell, bbimage? Sep 10 21:04:45 s/For/From/ Sep 10 21:04:53 'bitbake abiword-plugins' Sep 10 21:05:03 do_configure and do_compile output to console Sep 10 21:05:44 koen: ah, hmmm :-/ Sep 10 21:06:24 and do_install + do_package Sep 10 21:06:49 I've missed merging a key bit of poky's bitbake... Sep 10 21:09:03 RP: in fact we should simply remove them from bb files (BUILD_ALL_DEPS is not so much used) Sep 10 21:09:39 cyrilRomain: We need backwards comatibility for now Sep 10 21:09:49 cyrilRomain: bitbake 1.4 and 1.6 do use that variable Sep 10 21:10:52 RP: ah ok, so to be taken into account by the content checker then ;) Sep 10 21:11:50 cyrilRomain: yes. Eventually, it will get removed from OE but for now, it stays Sep 10 21:14:10 RP: should the content checker suggests to remove it, or should it suggest the use of another variable/routines for what BUILD_ALL_DEPS was intended ? Sep 10 21:15:34 cyrilRomain: Not yet. Eventually, it can just suggest its removal but we can't do that. Its replacements are already where we need them Sep 10 21:15:59 RP: ok Sep 10 21:16:04 RP: thanks Sep 10 21:16:08 Or are they - meta packages. Hmm. I should think about that :) Sep 10 21:18:02 koen: Found the problem - not quite what I thought it was :) Sep 10 21:18:53 03rpurdie * r600 10bitbake/lib/bb/build.py: build.py: Fix debug_level references causing excessive debug output Sep 10 21:20:10 RP: http://rafb.net/paste/results/UGEUoC83.html Sep 10 21:22:31 I'm getting 'exception received' for the stuff that was previously spewing debug Sep 10 21:22:41 koen: Yes, I know why - one second Sep 10 21:24:11 03rpurdie * r601 10bitbake/lib/bb/build.py: build.py: Really fix debug_level references causing excessive debug output (s/general/default - thanks for spotting koen) Sep 10 21:24:40 Right idea, wrong keyword :) Sep 10 21:24:46 yep, that worked Sep 10 21:27:05 I wonder, what the heck with libpng at all? Sep 10 21:27:06 inherit pkgconfig binconfig pkgconfig Sep 10 21:27:18 yeah Sep 10 21:27:24 and that's for autotool'ed package. Sep 10 21:27:29 I committed that before I had my first coffee Sep 10 21:27:47 was autotool support for libpng just recently added? Sep 10 21:27:50 woglinde discovered that it was autotooled after that Sep 10 21:28:09 I suspect we won't need the .pc hack after doing 'inherit autools' Sep 10 21:28:38 I guess too, and that x86 asm issue will go away, etc. Sep 10 21:28:44 will reopen the bug Sep 10 21:28:52 [oe] [Bug 1326] libpng doesn't build for x86 Sep 10 21:40:10 koen: thanks to bitbake (and data.py) I think we can handle the variable order in sanitize (at least the variable defined and used within the bb file) Sep 10 21:40:50 s/I think// even Sep 10 21:41:46 'night all Sep 10 21:47:46 03rpurdie * r602 10bitbake/lib/bb/taskdata.py: taskdata.py: Clean up mesage handling for unbuildable targets - log at debug level unless the target was directly requested Sep 10 21:49:23 koen: Should run a little quiter now ;-) Sep 10 21:49:35 quieter even :) Sep 10 21:50:05 yay! Sep 10 21:50:21 and 'method already seen'? Sep 10 21:50:39 koen: Not my error message - see zecke ;-) Sep 10 21:50:54 koen: I have it commented out locally ;-) Sep 10 21:59:47 ooooh Sep 10 22:00:00 google earth now has my neighbourhood in high res Sep 10 22:01:09 how OE deals with autoconf site files? Sep 10 22:03:30 koen: ^ any hints? Sep 10 22:04:45 psokolovsky: Basically we have a list of known test results which live in the site directory and we feed that to autoconf Sep 10 22:05:02 RP: In what way they are fed? Sep 10 22:05:23 psokolovsky: Read autotools.bbclass - I don't know Sep 10 22:05:29 RP: I just created site file for new arch-os, but don't see it's being used Sep 10 22:05:52 RP: ok, thanks for hint Sep 10 22:10:53 psokolovsky: what kind of arch-os? Sep 10 22:11:03 I thought we had it pretty much covered :) Sep 10 22:12:15 koen: Well, are you ready to add more variety? ;-) I'm hacking mingw32 (native windows) support. lol Sep 10 22:12:27 koen: i586-mingw32msvc Sep 10 22:12:39 qualify as "embedded system"? lol Sep 10 22:12:41 heh Sep 10 22:12:53 I thought kergoth was insane with his arm-wince-pe :) Sep 10 22:13:23 koen: so, someone did that too? cool! Sep 10 22:13:58 kergoth did that when he was still at TI Sep 10 22:14:34 koen: was it released? Sep 10 22:15:03 no idea Sep 10 22:30:35 * v8jlene is back Sep 10 22:31:12 hey v8jlene Sep 10 22:32:12 Hi koen Sep 10 22:35:43 03lenehan 07org.oe.dev * r96cd2afd... 10/ (1 packages/openswan/openswan_2.2.0.bb): openswan 2.2.0: Remove spurious .debug directories from the non -dbg packages. Sep 10 22:43:12 03lenehan 07org.oe.documentation * r7ed612b5... 10/ (1 usermanual/reference/var_section.xml): Sep 10 22:43:12 usermanual SECTION: Make a note that the list is a recomendation only and Sep 10 22:43:12 that any additional/modified items should be discuessed via the OE dev Sep 10 22:43:12 mailing list. Sep 10 23:03:09 03lenehan 07org.oe.dev * r44f6b849... 10/ (1 packages/lilo-sh/lilo-sh_21.bb): Sep 10 23:03:09 lilo-sh 21: Change the SECTION from bootloader to bootloaders to match the Sep 10 23:03:09 recomended sections list. Sep 10 23:03:15 03lenehan 07org.oe.dev * r1381f26b... 10/ (1 packages/lilo-sh/lilo-sh_21.bb): lilo-sh 21: Run sanitize.py over the recipe. Sep 10 23:12:51 03lenehan 07org.oe.dev * r41f4211f... 10/ (3 files in 3 dirs): quagga 0.99.2: Drop old version. Sep 10 23:41:22 03lenehan 07org.oe.dev * r6815c32e... 10/ (4 files in 2 dirs): Sep 10 23:41:22 quagga: Manually sanitize the recipe. The sanitize.py script makes a Sep 10 23:41:22 bit of a mess of quagga.inc so it needed to be done manually. Sep 10 23:41:36 03lenehan 07org.oe.dev * r85e525fe... 10/ (4 files in 2 dirs): Sep 10 23:41:36 quagga: Add a -dbg package to the list of generated packages so that the Sep 10 23:41:36 .debug files are not left unpackaged. Sep 11 00:36:31 * v8jlene is away (gone cycling) Sep 11 02:43:22 night **** ENDING LOGGING AT Mon Sep 11 02:59:57 2006