**** BEGIN LOGGING AT Sat Sep 09 02:59:56 2006 Sep 09 04:01:24 03lenehan 07org.oe.documentation * re109add8... 10/ (3 files in 3 dirs): Sep 09 04:01:24 usermanual SECTION: Document the SECTION variable and start to put a Sep 09 04:01:24 description to some of the sections. Still a bit of a mess but it's probably Sep 09 04:01:24 better than nothing. Sep 09 04:36:28 03lenehan 07org.oe.dev * r02661077... 10/ (1 packages/gd/gd_2.0.33.bb): Sep 09 04:36:28 gd 2.0.33: Don't search for the X11 headers. Without this hosts that have Sep 09 04:36:28 X11 headers in /usr/include/X11 and not in /usr/X11R6/include will get a Sep 09 04:36:28 cross-compile badness error from the compiler from tying to use the host Sep 09 04:36:28 includes. Sep 09 06:47:30 hi, folks. how can I trace the dependency of package ? Sep 09 06:49:02 bitbake -s Sep 09 06:49:07 i think Sep 09 06:49:32 but that doesnt mean it will give you the correct deps Sep 09 06:52:21 emte, OK. I want trace the dependency of libX11. it seems it broken. Sep 09 06:53:35 libX11 would come from a package Sep 09 06:53:52 you need to know which one and then follow that one Sep 09 06:54:05 but like i said it may more may not give you correct deps Sep 09 06:54:09 or* Sep 09 06:54:30 it depends on the Maintainers and if/what they put in the depends feilds Sep 09 07:02:38 hmm, is there anybody would like talk the dependency of "libx11"? Sep 09 07:03:28 glibc :P Sep 09 07:04:18 emte, hmm. I think the libXrender circle dependency of "libx11".. Sep 09 07:05:20 emte, hmm, what do you mean "glibc" ? Sep 09 07:05:52 well if you dont have glibc i dont think you can even compile/run anything X based Sep 09 07:07:06 emte, hmm, yes. I did compile the glibc by OE. Sep 09 07:07:36 emte, I bitbake gpe-image. and it failed. so, I am trying to find where is wrong ? Sep 09 07:07:55 ah Sep 09 07:09:07 emte, hmm, what do you think about it ? Sep 09 07:09:27 i cannot say i have had a gpe image fail in quite a while Sep 09 07:18:15 emte, hmm, would like give me some advice about it ? Sep 09 07:22:29 without knowing what your actually problem is ... Sep 09 07:22:34 actual* Sep 09 07:50:50 morning all Sep 09 08:06:13 03rpurdie 07org.oe.dev * r2313adb8... 10/ (1 packages/linux/linux-openzaurus_2.6.17.bb): linux-oz-2.6: poodle compile fixes Sep 09 08:06:37 hi Sep 09 08:06:53 RP: we need to enable multiple LUNs on scsi subsystem Sep 09 08:08:00 otherwise tosa do not see all ports in my cardreader Sep 09 08:11:21 RP: I will take care of it later Sep 09 08:11:41 hi koen Sep 09 08:14:04 good morning all Sep 09 08:15:40 heh...morning Sep 09 08:18:50 ~lack mtn pluck Sep 09 08:19:10 ~lart mtn pluck Sep 09 08:19:11 * ibot drops a baby grand on mtn pluck Sep 09 08:19:15 03hrw 07org.oe.oz354x * r56a77851... 10/ (1 packages/linux/linux-openzaurus_2.6.17.bb): Sep 09 08:19:15 linux-oz-2.6: poodle compile fix Sep 09 08:19:15 Revision: 2313adb82b2a9e641f7a5b1f87dec5b5148bf9b5 from .dev Sep 09 08:25:12 RP: pushed Sep 09 08:25:20 03hrw 07org.oe.oz354x * r69d57d1f... 10/ (4 files in 3 dirs): linux-openzaurus 2.6.17: enable Multiple LUNs for tosa and collie Sep 09 08:25:25 03hrw 07org.oe.dev * rc520216d... 10/ (4 files in 3 dirs): linux-openzaurus 2.6.17: enable Multiple LUNs for tosa and collie Sep 09 08:25:34 ~lart mtn pluck again Sep 09 08:25:35 * ibot strangles mtn pluck again with a doohicky mouse cord Sep 09 08:29:15 morning hrw and koen Sep 09 08:29:50 hey RP Sep 09 08:32:20 I want 'mtn takepatch' Sep 09 08:32:45 mtn log -rASDFG --last 1 --diffs|mtn takepatch - Sep 09 08:39:00 hmm. poodle's kernel is still broken :-( Sep 09 08:39:30 ;( Sep 09 08:39:36 RP: then I will not touch my poodle Sep 09 08:40:03 hrw: I'll get it fix. It has to be something silly... Sep 09 08:57:46 cu Sep 09 08:58:43 * koen waves to zecke Sep 09 08:59:00 hey Sep 09 08:59:10 I think I will hack fontconfig now Sep 09 08:59:14 ji zecke Sep 09 08:59:18 and hi even :) Sep 09 08:59:25 hey Sep 09 08:59:36 RP: have you seen my poky 'request'? Sep 09 08:59:44 zecke: I have, thanks :) Sep 09 09:00:18 RP: http://bugzilla.gnome.org/show_bug.cgi?id=350973 <- fun with floats Sep 09 09:02:30 koen: You mention what I was thinking as I was reading it :) Sep 09 09:05:08 :) Sep 09 09:23:17 ~lart fink Sep 09 09:23:17 * ibot whacks fink upside the head Sep 09 09:23:44 hi, is there somewhere a page that explains how to use bitbake to compile my own appliucations? Sep 09 09:24:29 I don't think there is such a doc Sep 09 09:24:34 since bitbake isn't a real compiler Sep 09 09:25:11 Koen: So how do I do that? Sep 09 09:25:37 like I said, bitbake isn't a real compiler Sep 09 09:26:05 YoG: you create a couple of 'tasks' which depend on each other Sep 09 09:26:19 YoG: and then you ask bitbake to execute a task and it will execute all the dependencies Sep 09 09:26:45 zecke: ok, how do I do that? Sep 09 09:26:49 YoG: or do you want to add your application to OpenEmbedded Sep 09 09:27:02 YoG: taking a look at the BitBake manual sounds like a clever idea Sep 09 09:27:19 zecke: in openembedded? Sep 09 09:27:22 YoG: and then taking a look at base.bbclass (an example in the bitbake directory) is another awesome idea Sep 09 09:27:41 YoG: ask you ask for plain bitbake, you should look into bitbake Sep 09 09:27:53 BitBake != OpenEmbedded Sep 09 09:28:27 for OE, consult our excellent user manual: http://www.openembedded.org/user-manual&dpage=ch02s04 Sep 09 09:28:42 zecke: ok, will do, thanx Sep 09 09:30:17 03rpurdie 07org.oe.dev * rf46b8ba3... 10/ (1 packages/linux/linux-openzaurus_2.6.17.bb): linux-oz-2.6: poodle locomo oops fix Sep 09 09:33:16 hrw|gone: fixed ;-) Sep 09 09:33:23 * RP -> back later Sep 09 09:33:33 WTF fink and MacPorts suck donkey ass Sep 09 09:34:27 zecke: amen to that Sep 09 09:42:29 03koen 07org.oe.documentation * r96afe58b... 10/ (1 usermanual/usermanual.xml): usermanual/usermanual.xml: use bitbake 1.6 branch Sep 09 10:28:23 hi all , yesterday I hrw told me that I wouldnt be able to use an atheros cardbus in my simpad, why is that? Sep 09 10:29:52 koen? Sep 09 10:30:10 koen: ? Sep 09 11:36:35 koen: we need to be careful to not suck as much as MacPorts and fink Sep 09 11:37:21 we already suck more by needing case sensitive file systems Sep 09 11:37:42 koen: well, it is glibc Sep 09 11:38:10 koen: port install gimp2 Sep 09 11:38:21 koen: and installing aalib fails on patching Sep 09 11:38:37 I have gimpOSX Sep 09 11:39:02 which reminds me, I should install gtk 2.10 to have 'native' rendering Sep 09 11:44:08 damn I need to shave Sep 09 11:44:13 03pH5 07org.oe.dev * r38f8d73b... 10/ (23 files in 16 dirs): many recipes: change pcre to libpcre in DEPENDS Sep 09 11:45:30 koen: yeah downloading of atk fails... Sep 09 11:45:41 koen: as their mirrors are totally wrong Sep 09 11:45:52 hmmm Sep 09 11:46:00 I need to shave as well Sep 09 12:12:17 drat, fink declared itself obsolete Sep 09 12:12:34 * koen runs update.pl Sep 09 12:20:47 hi all ! Sep 09 12:23:45 Hi! Sep 09 12:24:01 cyrilRomain: Hope you got mail and saw my commit. Sep 09 12:26:06 psokolovsky: yep, just few seconds ago :) Sep 09 12:29:13 this isn't productive Sep 09 12:29:20 psokolovsky: thanks for your help ! (it is indeed even more readable without oditc) :) Sep 09 12:29:25 I can stream video from my tivo to my computer Sep 09 12:29:47 now I can watch tv wile I "work" Sep 09 12:31:10 cyrilRomain: Thank you for nice tool. My last suggestion would be to rename it to be have more specific name, so it can be put on PATH. bbsanitize, oesanitaize, oebb*, bboe*, whatever ;-). Give it a thought ;-) Sep 09 12:31:21 koen: ^ what do you think Sep 09 12:31:23 psokolovsky: psst: if you want you can add your name in the script :) Sep 09 12:32:19 cyrilRomain: ok ;-) if I'll do more significant hacking, I'll do ;-) Sep 09 12:32:30 afk in the moment Sep 09 12:32:34 well i like bbsanitize Sep 09 12:33:24 oe-stylize? Sep 09 12:36:29 psokolovsky: if the name change, please update the brief description (first lines of the scripts) accordingly Sep 09 12:43:23 cyrilRomain: integrate it to BitTest Sep 09 12:43:42 zecke: yes, I'll try to :) Sep 09 12:44:22 zecke: for now, I'm discovering bittest and try to "improve" the source-checker Sep 09 12:44:59 :) Sep 09 12:45:01 zecke: by "improve" I mean not to download packages and cvs/svn/.../ repositories, just checking there are right here Sep 09 12:45:35 cyrilRomain: hehe, source checker is intended to download to check availability of the source :) Sep 09 12:45:59 zecke: well, I download header only of tarball, which is a lot faster :) Sep 09 12:46:26 cyrilRomain: yes, but it won't check the md5sum and can't be used to mirror :) Sep 09 12:46:31 zecke: but it does not check for tarball integrity Sep 09 12:46:34 zecke: yep Sep 09 12:46:43 cyrilRomain: create your own module :) Sep 09 12:47:05 mickeyl still has to show me how to load modules by name (again) Sep 09 12:47:12 anyway, maybe the best is to frequently use the fast check once a week, and one big test per month for example Sep 09 12:47:37 cyrilRomain: oh well, there is the '-c' option for wget Sep 09 12:47:52 zecke: my own module ? ah.. I started to add a check_availability method in fetch/*.py Sep 09 12:47:52 cyrilRomain: the idea is to remove the complete source once a month Sep 09 12:48:05 hmm Sep 09 12:48:25 zecke: ah thanks for the wget tips (better that downloading the first 32bytes) Sep 09 12:48:27 hey mikearthur Sep 09 12:48:41 alright koen Sep 09 12:48:44 hmm, I don't get why I waste my time creating frameworks nobody uses... Sep 09 12:48:51 I should remove bittest Sep 09 12:49:13 zecke: dual license it and sell it to mvista Sep 09 12:49:46 zecke: removing bittest ? 8-( Sep 09 12:49:51 ;) Sep 09 12:50:27 zecke: should I understand I'd rather create a module derving from a bittest class ? Sep 09 12:51:09 cyrilRomain: I think we do not need a check_availability method in the fetchers Sep 09 12:51:21 cyrilRomain: we only need another FETCHCOMMAND, UPDATECOMMAND for bittest Sep 09 12:51:33 cyrilRomain: we do not need a sanitize.py with a 'custom' parser Sep 09 12:51:45 ah, it was the easiest way to integrate in my "don't-know well bittest' point of view Sep 09 12:51:47 cyrilRomain: all this handcrafting should be avoided Sep 09 12:52:10 zecke: yes Sep 09 12:52:21 cyrilRomain: that is the point. And my conclusion is to not release my Free Software any more Sep 09 12:52:34 zecke: to ba fair, I had to feeling to reinvent the wheel when creating the sanitize script Sep 09 12:52:41 cyrilRomain: hehe Sep 09 12:52:53 zecke: but did not have much time fully learning bittest Sep 09 12:52:54 :-/ Sep 09 12:53:01 cyrilRomain: BitTest does parsing and loading for you Sep 09 12:53:10 cyrilRomain: and provides a way to report results to a TinderBox Sep 09 12:53:18 cyrilRomain: it is easy :) Sep 09 12:54:11 zecke: I don't see clearly how to retrieve the list of .bb data (I mean the whole list of a .bb file, not just a variable) Sep 09 12:54:51 cyrilRomain: there is a testrun.conf and then you set BBFILES :) Sep 09 12:55:23 zecke: I mean not the list of .bb file, but the list of all the data included in one .bb file Sep 09 12:55:54 so that I can sort it like in sanitize.py Sep 09 12:55:55 cyrilRomain: 'def test(self, file_name, file_data):' Sep 09 12:56:08 cyrilRomain: file_data is a updated bb.data dictionary Sep 09 12:56:21 zecke: ah cool :) Sep 09 12:56:22 cyrilRomain: you can access the data, using bb.data.getVar Sep 09 12:56:55 zecke: bb.data.getVar give the value of _one_ variable, right ? Sep 09 12:57:02 cyrilRomain: The pro for a custom line based parser is you can do linting on the files before the BitBake parsers sees it Sep 09 12:57:18 cyrilRomain: the pro for using the BitBake parser is: You use the data as it is seen by BitBake Sep 09 12:57:34 cyrilRomain: take a look at data.py it has documentation ;) Sep 09 12:58:03 zecke: yes, that why I tend to say that sanitize work on the format, while bitest work on the content Sep 09 12:58:16 zecke: ok thanks Sep 09 12:58:33 cyrilRomain: but the amount of format checking you do is quite small Sep 09 12:58:41 e.g. whitespace after '\' Sep 09 12:58:43 * cyrilRomain hope to find some time because holidays end tomorrow :-( Sep 09 12:59:19 zecke: yes, that's why I would like that someone tell me all the rule to check Sep 09 13:00:10 I really don't like OE shipping a reinvented wheel Sep 09 13:00:34 e.g. 99% of the stuff you do is already done by the content checker Sep 09 13:00:44 :-/ Sep 09 13:01:03 1% belongs into the parser(s) to have a style check Sep 09 13:01:21 but sanitize.py is much easier to use Sep 09 13:01:57 and was easier to write (at least until I better know bittest to change my mind) Sep 09 13:02:23 cyrilRomain: easier than writing one 'test' method and returning a result? Sep 09 13:02:45 koen: so what? provide a default testrun.conf Sep 09 13:02:50 I would say 'faster' not easier in fact Sep 09 13:03:45 koen: so bittest should be removed? Sep 09 13:04:48 I think it is like bitbake. You can always do thing more quickly by hand, but a good tool is always better but it needs time to learn Sep 09 13:05:39 hum maybe not a good example, because I can't do what bitbake do more quicky by hand ;) Sep 09 13:05:55 cyrilRomain: don't worry Sep 09 13:06:28 It would have been nice to get feedback, specially on the README but I understand it is too complicated to read a file when you can write a parser using regexp Sep 09 13:07:16 03zecke123 * r587 10bitbake_qa/: trunk/bitbake_qa: This tool is too complicated, I will maintain it in my private tree Sep 09 13:07:39 zecke: if bittest is so great, why didn't you suggest it in the thread on the mailinglist? Sep 09 13:08:10 I wasn't subscribed ;) Sep 09 13:08:27 maybe people just don't bittest Sep 09 13:08:40 * cyrilRomain was a bit surprised JustinP didn't know bittest Sep 09 13:09:29 koen: I think I can't do more than providing a framework, short README and examples Sep 09 13:09:30 either way, I think bittest (or something similar if you want to remove it) should be part of the QA plan Sep 09 13:09:42 cyrilRomain: whatever ;) Sep 09 13:12:23 cyrilRomain: I wonder too, as the depends_checker and source_checker spotted errors in the e17 stuff Sep 09 13:13:51 later guys Sep 09 13:27:31 zecke: Q4 morning Sep 09 13:53:56 koen: what made bittest so complicated? Sep 09 13:54:09 Hello Sep 09 13:54:55 zecke: it requires to much setting up Sep 09 13:55:29 koen: the testrun.conf or setting PYTHONPATH? Sep 09 13:55:34 both Sep 09 13:55:59 'set PYTHONPATH' is also extremely undescriptive Sep 09 13:56:05 set it to what? Sep 09 13:57:45 and why wasn't/isn't it mentioned on the mailinglist? Sep 09 13:58:04 koen: it was mentioned, I have posted results of the tests a couple of times ;) Sep 09 13:58:18 for sanitizing the recipes? Sep 09 13:58:21 I think not Sep 09 13:58:33 for fixing up sections, license fields Sep 09 13:58:41 I think that is sanitizing as well Sep 09 13:59:04 so why haven't you mentioned it in the thread about sanitizing? Sep 09 13:59:52 koen: wasn't subscribed the first week in september? Sep 09 14:00:03 you are now Sep 09 14:01:48 I think I even mailed that bittest can send the test result to a tinderbox Sep 09 14:02:19 but the whole point is void, since you removed bittest Sep 09 14:02:41 yes :) Sep 09 14:02:44 so in the future we won't use any of your stuff since you will delete it anyway Sep 09 14:03:00 koen: no Sep 09 14:03:16 koen: I wonder if I should remove bitdoc as it produces crappy HTML like code Sep 09 14:04:26 make it output docbook, so we can have an automated way for updating the usermanual Sep 09 14:04:54 koen: good idea Sep 09 14:05:51 koen: will do after lunch Sep 09 14:07:05 koen: I do not want to carry old bits in trunk and if OE decided not to use BitTest it deserves no purpose :) Sep 09 14:07:22 and we use SVN anyway Sep 09 14:08:57 * cyrilRomain just created a module for the do-not-download-just-check-for-source source checker Sep 09 14:09:14 cyrilRomain: awesome :) Sep 09 14:09:33 cyrilRomain: the bad thing about bittest is you need to touch bin/bittest to add your module Sep 09 14:09:39 * cyrilRomain now tries to create a module instead of adding methods into fetch/*.py Sep 09 14:09:51 zecke: yes I saw that Sep 09 14:10:05 zecke: I think it only lack little documentation Sep 09 14:10:27 good, tell me what is missing. koen's hint about pYTHONPATH was good already Sep 09 14:11:28 the ' you need to touch bin/bittest to add your module' t Sep 09 14:12:16 cyrilRomain: I wait for mickeyl to help me there Sep 09 14:14:20 zecke: hum... I'd like to avoid code duplication and I'm wondering if it is wise creating a module too-much similar to fetch/*.py Sep 09 14:15:14 cyrilRomain: Let me understand your motivation first Sep 09 14:15:28 cyrilRomain: I think source_checker is already able to do the things you want to do Sep 09 14:15:46 cyrilRomain: e.g. removing *.md5 will already incrementally check source Sep 09 14:15:59 zecke: ah, I'd like to check for source availability _without_ downloading/checkouting packages Sep 09 14:16:46 zecke: because to check all SRC_URI it is just too much time/bandwith to download them all Sep 09 14:17:12 cyrilRomain: well. I think the OE meta-data should carry sha256 and md5sum for every tar/file Sep 09 14:17:19 zecke: I agree we should have such a checker, but we also need a fast-checker Sep 09 14:17:32 cyrilRomain: so checking 'availability' usability can only happen by downloading the file Sep 09 14:18:15 my point is we need both. The current source_checker for a strong check (with integrity) Sep 09 14:18:41 hehe well Sep 09 14:18:42 and a faster one, just to check the author has not changed the SRC_URI for example Sep 09 14:18:56 for KDE your checker will regulary fail as they change the tarballs after releases Sep 09 14:19:18 cyrilRomain: the difference is for the first time you will need bandwidth Sep 09 14:19:23 zecke: yep, but at least we will be quickly warn of such changes Sep 09 14:19:36 cyrilRomain: for the second run of the source_checker you only 'stat' the files Sep 09 14:20:27 cyrilRomain: I do not think a dedicated fast checker is necessary when you need to do rm *.md5 to have the same effect Sep 09 14:20:57 zecke: ah ok. I didn't have so many source then because I didn't remark it don't download what is already downloaded Sep 09 14:21:37 cyrilRomain: wget, cvs and svn and git should be clever here Sep 09 14:21:53 zecke: are they ? Sep 09 14:21:53 cyrilRomain: and at least wget,cvs,svn are clever indead Sep 09 14:22:23 indeed... man my language sucks... Sep 09 14:23:08 cyrilRomain: so the only difference is the first run, if you think this justifies a dedicated check you are right with adding adding check_availability methods to fetch/* and I would use them from a bittest module then Sep 09 14:24:17 cyrilRomain: e.g. thanks to psokolovsky we keep the svn co around as well Sep 09 14:24:31 zecke: ok. My check_availability methods is less justified if on the second run things are not downloaded again Sep 09 14:24:48 be back in less than a hour Sep 09 14:39:48 cyrilRomain: in a setup we would run the source_checker once a day Sep 09 14:40:02 cyrilRomain: it would start by removing *.md5 and then run it Sep 09 14:40:11 cyrilRomain: once a month it could remove the complete dir Sep 09 14:40:19 zecke|food: yes Sep 09 14:41:14 zecke|food: I think it is important to report them to tinderbox and providing a link somewhere so that anyone can see the broken SRC-URI Sep 09 14:42:11 and send a mail to ${MAINTAINER} Sep 09 14:42:18 the first run of my script spotted many broken link Sep 09 14:42:29 koen: excellent idea Sep 09 14:42:39 koen: did one update the default maintainer? Sep 09 14:42:58 no idea Sep 09 14:43:09 I updated the default one for angstrom Sep 09 14:43:12 zecke|food: I know understand why it always download things even after a first run Sep 09 14:43:24 zecke|food: it is downloaded into ./tmp_sources Sep 09 14:43:32 instead of using the source directory Sep 09 14:43:43 zecke|food: did I miss something in testrun.conf ? Sep 09 14:44:17 cyrilRomain: I assume I changed the DL_DIR variable Sep 09 14:44:43 either in the bitbake.conf copy or the source_checker itself Sep 09 14:46:04 Anyone here know if there's a performance difference between little and big endian on arm processors? Sep 09 14:46:05 zecke|food: I'll have a look soon Sep 09 14:46:22 joshin: depends on your usage Sep 09 14:46:56 Network server. I would imagine that big would have the advantage there. Sep 09 14:47:21 if you have a arm BE machine for that it would be good to run it in BE Sep 09 14:48:55 Ok, thanks. I'll be able to do some solid benchmarking soon and compare openslug built both ways. Sep 09 14:49:59 zecke|food: btw, 'openembedded' now has enough google juice again Sep 09 14:50:28 koen: MAINTAINER is still oeAThh.org where should we change that to? Sep 09 14:50:46 zecke|food: your OE tree is obsolete Sep 09 14:50:59 03koen 07org.oe.dev * rabdf62cf... 10/ (1 conf/bitbake.conf): bitbake.conf: update default maintainer Sep 09 14:51:06 hehe Sep 09 14:51:11 where did you change it to? Sep 09 14:51:20 openembedded-devel@lists.openembedded.org Sep 09 14:51:40 I still want a openembedded-bugs@, though Sep 09 14:51:47 me too Sep 09 14:55:14 * koen gets some chopsticks Sep 09 14:59:23 zecke|food: not in source checker code, and bitbake.conf looks ok: DL_DIR = "${TMPDIR}/downloads" Sep 09 15:00:15 hum... I didn't set TMPDIR in my local.conf, but I can't figure out where does it creates tmp_sources Sep 09 15:02:26 zecke|food: ah, in bin/bittest itself Sep 09 15:02:37 * zecke|food stops searching Sep 09 15:03:00 zecke: in run_tests() Sep 09 15:03:46 I probably didn't want to mess with the normal dl_dir Sep 09 15:04:23 cyrilRomain: do you think bittest could be usable? Sep 09 15:04:44 zecke: you mean by OE devs ? Sep 09 15:05:02 zecke: yes Sep 09 15:05:33 zecke: as I said, I think it should be part of the QA plan Sep 09 15:05:47 I will improve it a bit in my tree and sync it back to trunk once it is more usable Sep 09 15:05:49 hmm, is there anybody want review this patch ? http://rafb.net/paste/results/eIrcqj35.html Sep 09 15:05:54 zecke: :) Sep 09 15:06:07 cyrilRomain: do you happen to know perl? Sep 09 15:06:22 hmm, maybe it slove this bug : http://bugs.openembedded.org/show_bug.cgi?id=1188 Sep 09 15:06:51 but, maybe it is the perfect solution for it . Sep 09 15:06:54 zecke: not really, but I'm a developer so I'm able to do something with Sep 09 15:07:28 zecke: what do you need in perl ? Sep 09 15:07:45 cyrilRomain: finally improve the tinderbox code :) Sep 09 15:08:10 zecke: ah. I could have a look (I remember there is not so many line of code) Sep 09 15:08:13 but we can do that in python as well Sep 09 15:08:17 just a question of time ;) Sep 09 15:08:51 zecke: do you want to translate perl code to python for thinderbox ? Sep 09 15:09:08 cyrilRomain: no, extend perl code, write new code in python Sep 09 15:09:19 cyrilRomain: I have used tbox3 to not start from scratch Sep 09 15:09:41 cyrilRomain: e.g. I would like to extract PN/PV/PR from the tinder report and store it in new tables Sep 09 15:09:45 to make an easy query page Sep 09 15:10:13 zecke: I see Sep 09 15:10:27 I have been too busy so far Sep 09 15:11:18 zecke: I could have a look but don't promise so far Sep 09 15:12:11 s/could/can/ Sep 09 15:13:59 zecke: for the DL_DIR issue, it is possible to let user choose ? for example putting it into the testrun.conf ? Sep 09 15:14:51 cyrilRomain: well, yes Sep 09 15:15:12 zecke: I think it would be good Sep 09 15:15:42 zecke: and let me know if you think it is worth sharing my source availability checker (wget/cvs/svn/ works, not yet git and svk) Sep 09 15:16:30 worth sure, I have stated my doubts about the need though :) Sep 09 15:16:46 zecke: :) Sep 09 15:17:07 zecke: another unrelated thing, I have a patch for content_checker.py with the section list discussed on the mailing list Sep 09 15:17:31 zecke: I'll send it to you as soon as the discussion end Sep 09 15:17:39 cyrilRomain: awesome Sep 09 15:18:26 * cyrilRomain would like to do literate programming so that the content of the wiki is directly used to generate the python code (at least for such list) Sep 09 15:19:35 s/is/would be/ Sep 09 15:24:34 we all have such wet dreams Sep 09 15:28:12 zecke: dreams can become true: http://en.literateprograms.org/LiteratePrograms:Welcome Sep 09 15:39:30 leoncamel: If your patch worked for you, I suggest you to submit the patch to bugzilla (to the related bug), so that OE devs will review it Sep 09 15:41:52 cyrilRomain, OK Sep 09 15:53:57 * chouimat is away: stuff Sep 09 15:58:31 zecke: another entry for sanity.conf: "bc" Sep 09 15:58:47 CoreDump|afk: for OpenZaurus kernel? Sep 09 15:58:57 nah Sep 09 15:59:22 sanity.conf is only kernel-related? Sep 09 15:59:36 CoreDump|afk: no, I wonder who uses bc and when Sep 09 15:59:41 =) Sep 09 15:59:46 xscreensaver for one Sep 09 16:01:57 morning Sep 09 16:02:26 hey hrw Sep 09 16:02:31 my home machine is now only mine Sep 09 16:02:36 Ania bought laptop Sep 09 16:03:54 =) Sep 09 16:04:18 CoreDump|home: xscreensaver needs bc? why is that? (neugierig) Sep 09 16:04:34 promotion in vobis == ~300-400 pln less then usual Sep 09 16:05:38 zecke: no idea, really. configure is bitching about it when it's missing heh Sep 09 16:06:01 hrw: ouch Sep 09 16:08:52 ~change 2000 pln to eur Sep 09 16:09:02 2,000.00 Zloty (PLN) makes 501.926 Euro (EUR) (from http://www.xe.com/) Sep 09 16:09:31 500 eur for 1.6GHz celeronM/256M/40GB/wifi/cdrw combo with winXP Home is cheap Sep 09 16:09:59 ~lart thunderbird for lack of importing from normal formats Sep 09 16:10:00 * ibot takes thunderbird to the vet for a "special" visit for lack of importing from normal formats Sep 09 16:21:16 thunderbird sucks Sep 09 16:21:31 sadly its the least bad mail client for OSX Sep 09 16:25:50 zecke: I understand now why you did'nt want to use DL_DIR with source_checker. Because bittest do not download packages that are already present in DL_DIR, you have to clean the DL_DIR to check SRC_URI is still valid Sep 09 16:26:10 cyrilRomain: rm *.md5 ;) Sep 09 16:26:26 CoreDump|home: fwiw, bc is already in RequiredSoftware Sep 09 16:26:28 zecke: ok Sep 09 16:26:39 * cyrilRomain tries it Sep 09 16:28:20 zecke: that is a good trick :) Sep 09 16:28:47 03rpurdie * r588 10bitbake/lib/bb/runqueue.py: runqueue.py: Improve error messages (and make get_error_idstring a more global function) Sep 09 16:32:52 cyrilRomain: yes, and now you almost only 'stat' the servers Sep 09 16:33:40 zecke: ? Sep 09 16:34:00 zecke: ah ok, sorru Sep 09 16:34:04 s/sorru/sorry Sep 09 16:34:10 cyrilRomain: it doesn't download the whole file Sep 09 16:34:26 cyrilRomain: it wgets the server and recognizes it has the same filesize already Sep 09 16:34:35 doesn't python have a http lib to do a 200 ? Sep 09 16:34:45 zecke: hm.. I'm now having doubt about my availability checher ;) Sep 09 16:34:51 cyrilRomain: if the filesize is less wget errors if the filesize is more the md5sum will be wrong Sep 09 16:35:14 s/about/about the usefulness of/ Sep 09 16:37:25 zecke: and it is possible to change bittest so that it behaves as if md5 were deleted ? Sep 09 16:38:27 cyrilRomain: no, but again. You would run the source checker from within a script that invokes bittest once a day Sep 09 16:38:32 zecke: so that DL_DIR is not really messed up (it just contains all the packages instead of the one we use) Sep 09 16:38:36 cyrilRomain: and before invoking bittest it would remove the md5 stuff Sep 09 16:39:13 zecke:right Sep 09 16:39:53 we could add a 'setup' method to the 'specification' Sep 09 16:40:01 and the source checker could remove the md5 files Sep 09 16:40:56 why not Sep 09 16:43:46 zecke: to report bittest result to tinderbox, what should I do ? (I tried using the same tinderbox settings from my local.conf but without success) Sep 09 16:43:54 03rpurdie * r589 10bitbake/lib/bb/ (7 files in 2 dirs): Convert fetchers to use bb.msg Sep 09 16:44:29 cyrilRomain: the variables are the same Sep 09 16:44:39 cyrilRomain: you need to specify another output format though Sep 09 16:45:15 zecke: ah ok Sep 09 16:46:29 zecke: which one ? Sep 09 16:46:38 uf.. finally done: http://www.hrw.one.pl/projects/ Sep 09 16:46:55 Bonjour Sep 09 16:47:02 * rwhitby is in Toulouse Sep 09 16:47:12 bonjour Sep 09 16:47:13 ca va? Sep 09 16:47:15 zecke: should nslu2-linux move to bitbake 1.6? Sep 09 16:47:22 tres bien merci Sep 09 16:48:01 rwhitby: should is strong word. I think that you can and it is suggested to do Sep 09 16:48:02 zecke: Is there any reason we should be using print statements anywhere? (instead of bb.msg.note) Sep 09 16:48:12 rwhitby: for holidays ? I live in Toulouse Sep 09 16:48:13 rwhitby: I'd use 1.6 Sep 09 16:48:19 and what version Monotone are the OE servers using at the moment? Sep 09 16:48:24 0.29 Sep 09 16:48:28 cyrilRomain: work. I'm here till Wednesday Sep 09 16:48:36 rwhitby: I agree with RP there, Bitbake1.6 should be safe Sep 09 16:49:07 rwhitby: so welcome in this warm place :) Sep 09 16:49:09 hi Sep 09 16:49:17 cyrilRomain: -r TestReportTinder should do Sep 09 16:49:21 zecke: thx Sep 09 16:49:29 cyrilRomain: I was last here 8 years ago. I like Toulouse. Sep 09 16:49:30 TestReportTinder Sep 09 16:49:49 BTW: how can I enable the blinking console cursor again? Sep 09 16:50:08 i ve created a package for tor and repaired the socat package. how to submit that stuff? Sep 09 16:50:38 RP: I have two questions. Does bb.msg.note provide everything we want/need? e.g. regions/regexp matching? Sep 09 16:50:43 balle_: bugs.openembedded.org Sep 09 16:50:51 balle_: bugs.openembedded.org and bitching us Sep 09 16:51:04 With a move from bitbake 1.4 to 1.6, do I need to clear caches? clear all of tmp? Sep 09 16:51:10 but tor is a new package? Sep 09 16:51:11 zecke: It doesn't yet but we will be able to add that - its designed with it in mind Sep 09 16:51:41 RP: I think we should these prints over to bb.msg.note once it provides that what we want? does that sound reasonable? Sep 09 16:51:47 rwhitby: It should wipe the cache itself and tmp shouldn't need clearing Sep 09 16:51:52 RP: thx Sep 09 16:52:09 RP: besides I do see no reason to stay with print (well only print >> f, "foo" but we don't use that anyway) Sep 09 16:52:23 rwhitby: we have finally a versioned cached format :) Sep 09 16:52:29 and i dont want to report a bug, but submit a patch Sep 09 16:52:41 balle_: report a bug, attach a patch Sep 09 16:52:43 zecke: I was thinking the other way - Convert everything to use msg.note with the domain set, the worry about controlling the output :} Sep 09 16:53:06 zecke: I'll change the print statements though as I think they're just a throwback... Sep 09 16:53:12 balle_: new stuff is 'wishlist' bug Sep 09 16:53:19 NOTE: Invalid cache found, rebuilding... - sweet Sep 09 16:53:29 RP: I think it is a matter of touching the code once or twice Sep 09 16:53:43 RP: If you feel confortable with touching it twice I'm fine with that :) Sep 09 16:53:43 rwhitby: It has a version number embedded in and we remembered to change it ;-) Sep 09 16:54:12 zecke: No need to touch the code twice. The domains/regexp mathcing is just enhacing msg.py Sep 09 16:54:12 zecke: it worked but I have a strang issue: http://rafb.net/paste/results/xZnQrC16.html Sep 09 16:55:35 cyrilRomain: you have no TMPDIR in your local conf/testrun.conf? Sep 09 16:56:03 cyrilRomain: it sounds like mkdirhier(None) is called Sep 09 16:56:13 RP: then feel free to hack away Sep 09 16:57:02 hi mickeyl Sep 09 16:57:31 hey RP Sep 09 16:57:33 zecke: no TMPDIR in my conf/testrun.conf Sep 09 16:57:52 * cyrilRomain adds it Sep 09 16:57:54 ah wait a second Sep 09 16:58:11 cyrilRomain: do you have BBPATH set? and is the OE dir inside your BBPATH? Sep 09 16:58:21 bye bye gandalf Sep 09 16:58:32 zecke :) Sep 09 16:58:33 ~lart Qtopia for disabling my blinking cursor Sep 09 16:58:33 * ibot takes a large goose feather pillow and swings it wildly in Qtopia's direction, hitting Qtopia and sending Qtopia flying into the closet for disabling my blinking cursor Sep 09 16:59:05 * mickeyl waves in silence to gandalf Sep 09 16:59:07 zecke: yes Sep 09 16:59:16 now the last service has migrated. Sep 09 16:59:26 cyrilRomain: grep through your bitbake.conf there really should be a TMPDIR defined Sep 09 17:00:45 lol bash caches `pidof app` Sep 09 17:02:01 zecke: conf/bitbake.conf:TMPDIR = "${TOPDIR}/tmp" Sep 09 17:02:35 go to "/home/cyril/OpenEmbedded/bitbake/branches/bitbake-1.6/lib/bb/__init__.py", line 130, in mkdirhier Sep 09 17:02:51 and check what it prints for the argument (check if it is None) Sep 09 17:03:09 and then I assume your BBPATH is wrong Sep 09 17:03:27 zecke: ok Sep 09 17:04:07 hrw: does gnumeric build for you? Sep 09 17:04:46 hrw: s/houndred/hundred/ in http://www.hrw.one.pl/projects Sep 09 17:05:52 CoreDump|home: it is in feeds so it built for me Sep 09 17:06:07 hmm Sep 09 17:06:25 koen: thx Sep 09 17:07:20 zecke: it was TMPDIR which was missing from my conf/testrun.conf Sep 09 17:07:36 cyrilRomain: no, you shadow an error here Sep 09 17:08:28 zecke: well, after eating something I'll have idea more clear Sep 09 17:08:37 cyrilRomain|bbl: take care Sep 09 17:10:30 I'm getting "application.c: In function `gnumeric_application_setup_pixbufs': Sep 09 17:10:30 application.c:569: error: `gnm_cursor_cross' undeclared (first use in this funct Sep 09 17:10:30 ion) Sep 09 17:10:40 any ideas? Sep 09 17:11:59 cu Sep 09 17:29:24 aaaargh Sep 09 17:29:30 * koen stabs libpcre Sep 09 17:32:47 koen: any idea about the gnumeric issue I posted above? Google tells me you had the same problem a year ago =) Sep 09 17:33:03 03rpurdie * r590 10bitbake/ (12 files in 3 dirs): Complete conversion to use bb.msg Sep 09 17:33:04 I think it solved itself Sep 09 17:33:41 :\ Sep 09 17:33:50 "works in .dev" Sep 09 17:34:42 used to work in .oz as well. My build env is likely t be the problem Sep 09 17:41:04 03koen 07org.oe.dev * rc6bcfdb5... 10/ (1 packages/libpcre/libpcre_4.4.bb): pcre: attempt to repair the damage introduced Sep 09 18:18:47 03coredump 07org.oe.oz354x * r21efd18f... 10/ (3 files in 2 dirs): pkgconfig: Fix a big which caused orbit2-native's configure to fail, thanks koen, hrw and Xora! Sep 09 18:19:10 ~stab pkgconfig Sep 09 18:19:13 * ibot runs at pkgconfig with an origami Swiss Army knife, and inflicts a nasty paper cut. Sep 09 18:21:30 * CoreDump|afk gets something to eat, later Sep 09 18:30:30 RP: ping Sep 09 18:33:02 hey woglinde Sep 09 18:33:39 zecke: hum.. I'm quite sure TMPDIR must be set into the testrun.conf Sep 09 18:34:29 zecke: in bin/bittest, line 136: the data given is not 'data' but 'test_config' Sep 09 18:34:38 ~seen RP Sep 09 18:34:56 cyrilRomain: see, TMPDIR is in bitbake.conf Sep 09 18:34:57 rp is currently on #gpe #oe #openzaurus #handhelds.org. Has said a total of 106 messages. Is idling for 1h 37m 54s, last said: 'hi mickeyl '. Sep 09 18:35:18 cyrilRomain: bitbake.conf should be loaded Sep 09 18:35:35 cyrilRomain: and TMPDIR should be present then, this is what I find scary Sep 09 18:35:38 yes, but bitbake.conf data is keep in the 'data' variable Sep 09 18:35:54 and testrun.conf data in the 'test_config' variable Sep 09 18:36:34 hi koen :) Sep 09 18:38:06 zecke: I tested that (see print lines) http://rafb.net/paste/results/6f752a33.html Sep 09 18:38:26 okay need to think about it then Sep 09 18:38:46 zecke: and I get that: http://rafb.net/paste/results/cerBIn49.html Sep 09 18:40:03 let me think about it Sep 09 18:48:12 zecke: I have a question though (sorry if I interrupt you in more important tasks): why is there bitbake_qa/conf/bitbake.conf ? This file differs from the org.opemembedded.org/conf/bitbake.conf. If it is normal, bittest users should be warned to set BBPATH to /stuff/conf/bitbake.conf when using bittest and to org.openembedded.dev/conf/bitbake.conf when using bitbake Sep 09 18:49:15 cyrilRomain: we set timeouts for wget and smaller retries Sep 09 18:49:27 cyrilRomain: I have synced with the OE bitbake.conf Sep 09 18:51:53 zecke: ok. Then user have to change BBPATH when using bittest otherwise he/she will not use thoses different timeout, right ? Sep 09 18:52:50 cyrilRomain: right. current BitTest is intended to run multiple configurations at once Sep 09 18:53:07 in contrast to a conf/local.conf to build just one machine+distribution Sep 09 18:53:33 as koen finds that hard to understand I currently write a more simple frontend using the data from conf/local.conf Sep 09 18:53:50 yes Sep 09 18:54:20 with HIG compliant dialogs? Sep 09 18:54:39 sure Sep 09 18:54:49 but sadly they are shown off-screen Sep 09 18:57:41 ah, the fabled off-screen rendering of gtk 2.12? Sep 09 19:05:20 zecke: ok. I think the wiki need something like a 'GettingStarted' page for bittest and how to report to tinderbox (the latter page has seemingly been lost when wiki was transfert to openembedded.org) Sep 09 19:05:26 mickeyl: ping have 1 min for priv message? (about wellenreiter and some wireless questions) Sep 09 19:05:35 zecke: I'll write something for that Sep 09 19:06:23 darkschneider: sure. go ahead Sep 09 19:06:30 tx Sep 09 19:07:01 mickeyl is alive! Sep 09 19:07:05 hey mickeyl Sep 09 19:07:09 hi ho Sep 09 19:08:22 when is this mac trade show in france again? Sep 09 19:08:55 there's a big mac thing the 12th Sep 09 19:09:01 ah right Sep 09 19:09:32 maybe core 2 duo powerbooks Sep 09 19:09:34 ehm Sep 09 19:09:38 macbook pros Sep 09 19:09:46 *nod* Sep 09 19:09:49 and core duo mac mini Sep 09 19:09:53 mickeyl: is there a 'which' implementation in python? Sep 09 19:09:54 I'm waiting for that event Sep 09 19:10:06 zecke: no, nothing stock Sep 09 19:10:36 os.system('which') / Sep 09 19:10:37 ? Sep 09 19:11:41 should be a one liner iterating over PATH Sep 09 19:11:56 mickeyl: like bb.which ;) Sep 09 19:13:21 zecke: will you ask Robert about a Hotel recommendation? Sep 09 19:14:12 mickeyl: I can, I planned to call the lab though Sep 09 19:14:26 mickeyl: and hrs.com is quite nice as well Sep 09 19:14:51 zecke: do we know where exactly the event happens? Sep 09 19:15:04 mickeyl: did you get a reply as well? Sep 09 19:15:12 yep Sep 09 19:15:45 mickeyl: http://www.trolltech.com/company/newsroom/events/allevents/devdays2006/VenueAcommodations Sep 09 19:16:19 I'm sure I won't book at the hilton :} Sep 09 19:16:33 zecke: let's book the hilton :D Sep 09 19:16:42 mickeyl: funny, I even know where this hotel is Sep 09 19:16:54 * koen waits for a "2 developers hauled of by security while screaming 'gpl it!'" newsstory Sep 09 19:17:00 mickeyl: about five bus stops away from ostbahnhof Sep 09 19:17:44 mickeyl: you like leffe radieusse as well? Sep 09 19:17:55 koen: that's ok for sure Sep 09 19:17:57 koen: "Beavis and Butt-Head clones arrested while screaming 'GPL it'" Sep 09 19:18:09 zecke: let me book the hilton for us. I will make it a "Betriebsausgabe" Sep 09 19:18:18 mickeyl: good, I picked 24 bottles today :) Sep 09 19:18:27 koen: wow :D Sep 09 19:18:44 mickeyl: oh, no, I can't cover that :} Sep 09 19:18:50 zecke: I'm sure you can :) Sep 09 19:19:07 i need to lower my taxes ;) Sep 09 19:19:12 mickeyl: lol Sep 09 19:23:17 mickeyl: any news on openezx? Sep 09 19:24:19 koen: i'm afraid not. harald and me are working on the same project atm., so I know how things are goin on.... i don't expect any substantial improvements before next year Sep 09 19:24:44 mickeyl: I was just curious since harald mentioned you talked about it Sep 09 19:26:34 koen: yeah, we talked (shortly) about it. let me say the following. although we are not directly working on openezx, at the end of the year we should have something that openezx benefits from... Sep 09 19:28:32 someone needs to work on the AP/BP stuff though Sep 09 19:28:48 and i know neither harald nor me are in the (timely) position to do that Sep 09 19:29:27 * mickeyl hears Sabine calling ... "leave that damn machine and come to me" Sep 09 19:33:04 http://www.stadtplandienst.de/fastlink.asp?key=cb284721666d80178c5eb4774fae18d8 Sep 09 19:33:13 Haidenau-Platz ist unten rechts Sep 09 19:33:21 und ostbahnhof ein wenig weiter rechts Sep 09 19:34:05 oh weiter unten Sep 09 19:35:41 8259 revs in .dev Sep 09 19:37:49 any of you guys know much about procmail rules? Sep 09 19:38:26 I got mine via googling :) Sep 09 19:38:32 oop wrong window :) Sep 09 19:38:37 yeh, I've tried, can't work out why this won't work :s Sep 09 19:38:44 http://mikearthur.co.uk/.procmailrc Sep 09 19:38:57 everything seems to get matched by FreeCycle Sep 09 19:39:51 mikearthur: http://rafb.net/paste/results/0mFmmb71.html Sep 09 19:40:11 koen: cool Sep 09 19:45:08 * chouimat is back. Sep 09 19:45:14 chouimat: wb Sep 09 19:50:44 koen thanks ... Sep 09 19:52:15 http://www.openembedded.org/~koen/stats.htm Sep 09 19:52:58 I'm not certain if that includes merge nodes or not Sep 09 20:28:50 mickey|bbl: isn't there any or all in python? Sep 09 20:55:34 okay Sep 09 20:55:58 where should I push the improved bittest_qa to? Sep 09 20:56:02 release a tar? Sep 09 21:00:00 23:09 < zecke> okay Sep 09 21:00:00 23:09 < zecke> where should I push the improved bittest_qa to? Sep 09 21:00:01 23:09 < zecke> release a tar? Sep 09 21:00:18 haha, why does openzaurus.inc sets DISTRO? Sep 09 21:01:50 zecke: put it in svn Sep 09 21:02:04 or push it into .dev in contrib/ Sep 09 21:09:10 03rpurdie * r591 10bitbake/ (bin/bitbake lib/bb/msg.py): Add bitbake -l option to specify debug domains to show Sep 09 21:11:05 koen: is a invocation of bittest_single easy enough when bitbake is in your PATH and you have set BBPATH? Sep 09 21:11:16 I think so Sep 09 21:11:35 holy crap Sep 09 21:11:42 trunk is emulating -D Sep 09 21:12:31 koen: More like adding some kind of controls for the logging ;-) Sep 09 21:13:05 -l Cache is much more useful than -D -D if you have a cache problem ;-) Sep 09 21:13:49 http://rafb.net/paste/results/iiSB7M91.html Sep 09 21:14:08 (bbimage -n Angstrom-bootstrap-image-test-20060909-ipaq-pxa270 -t tar -e /home/koen/OE/monotone/org.openembedded.dev/packages/angstrom/angstrom-bootstrap-image.bb) Sep 09 21:15:11 if debug_level[domain] >= level:| KeyError: Parsing Sep 09 21:15:13 http://handhelds.org/~zecke/bitbake-qa-0.1.tar.bz2 Sep 09 21:15:17 looks like a bug in the new code Sep 09 21:15:48 zecke: why isn't it in svn? Sep 09 21:16:00 it is in svn Sep 09 21:16:03 zecke: Your python is better than mine - how do I make a function run to init a variable in msg.py, without turning it into a class? Sep 09 21:16:45 koen: not in a public one though Sep 09 21:16:54 zecke: I want to run set_debug_level(0) to init the debug_level variable :-/ Sep 09 21:17:14 RP: oh well, you can init it on load? Sep 09 21:17:25 let me take a look Sep 09 21:17:50 zecke: yes, init on load would be fine Sep 09 21:18:04 RP: then treat it like a static variable? Sep 09 21:18:19 zecke: yes Sep 09 21:18:37 zecke: Although perhaps we should rethink this... Sep 09 21:18:46 write set_debug_level(0) into the file Sep 09 21:18:59 somewhere (I gues sbelow the def set_debug_level) Sep 09 21:19:23 zecke: That doesn't work quite the way I expected Sep 09 21:19:27 RP: but there is a corner case in my python knowledge :) Sep 09 21:19:38 RP: the question is how often this module gets loaded Sep 09 21:19:43 zecke: its not just me then :) Sep 09 21:19:56 RP: I think import msg will not be 'cached' Sep 09 21:20:27 zecke: Most things don't even import, they just call bb.msg Sep 09 21:20:44 RP: as it gets imported from within __init__.py Sep 09 21:21:24 RP: I think we need to store the debug level at a more prominent place Sep 09 21:21:43 zecke: agreed. Where? Sep 09 21:22:06 zecke: Taking into account this stuff gets called anonymously a lot :-/ Sep 09 21:22:17 about [<--- here --->] let me experiment Sep 09 21:22:47 koen: You and your bugs. I thought I had all the edge cases covered ;-) Sep 09 21:23:08 set_debug_level(0) at the start of bbimage would solve all the problems Sep 09 21:23:16 (the cooker also calls that) Sep 09 21:24:46 do that first Sep 09 21:24:57 I will try to play with the globals keysword Sep 09 21:27:55 03rpurdie * r592 10bitbake/bin/bbimage: bbimage: Add call to init logging variable Sep 09 21:28:01 zecke: thanks Sep 09 21:28:19 RP: mailing OE list first about bittest Sep 09 21:28:37 RP: that seems to solve it, thanks Sep 09 21:28:48 NOTE: package angstrom-bootstrap-image-1.0: completed Sep 09 21:29:13 zecke: why not host bb_qa at openembedded.org? Sep 09 21:29:15 koen: Excellent :) Sep 09 21:29:31 zecke: 'I'll be interested to learn more about it :) Sep 09 21:34:24 03rpurdie * r593 10bitbake/lib/bb/build.py: build.py: Add interactive flag for tasks to optionally disable console redirection. Default to maintain backwards compatibility Sep 09 21:42:10 wow, that tor recipe looks mighty nice for a 'first' Sep 09 21:42:38 Has anyone noticed poky's devshell? Sep 09 21:42:46 I did Sep 09 21:43:01 and the addition of apt :) Sep 09 21:43:33 We could add devshell to OE now I updated bitbake :) Sep 09 21:43:42 cool Sep 09 21:43:57 RP: how do you get around the hardcoded arch list in dpkg? Sep 09 21:44:01 RP: and require unreleased version of BitBake? Sep 09 21:44:20 zecke: This is where we need to file a backport request :) Sep 09 21:44:33 I know it works with 1.4 ;-) Sep 09 21:44:44 koen: I think it can't be too hard to patch that... Sep 09 21:44:52 koen: kergoth is working on that, not me - I don't know Sep 09 21:45:00 RP: do you have patches to dpkg already? Otherwise I would start with that now Sep 09 21:45:06 kergoth: you are around? Sep 09 21:45:22 I have no idea where he's at Sep 09 21:45:43 * RP gets other work to deal with :-/ Sep 09 21:45:51 RP: I looked at teaching OE to create .debs in the past (easy) and letting dpkg support all of OE's archs (messy) Sep 09 21:46:16 koen: I know. Sep 09 21:47:11 I have lots of grand visions. Some coincide with OH's plans, some perhaps don't although none conflict. They just need time... Sep 09 21:47:36 that's why we have OEDEM :) Sep 09 21:47:48 indeed :) Sep 09 21:48:55 I did play with the idea of basing angstrom on poky last week :) Sep 09 21:50:17 That'd be an interesting choice :) Sep 09 21:51:19 I *think* I've found a way to do some clever branching using monotone: http://linuxtogo.org/gowiki/AngstromRoadmap Sep 09 21:52:03 that would allow angstrom to have a small subset of packages, but 'propagate' will update the bits in untested/ Sep 09 21:52:36 That assumes you make changes in .dev? Sep 09 21:53:21 yes Sep 09 21:54:41 I like the idea of splitting .dev into multiple BBCOLLECTIONS more Sep 09 21:54:54 kind of liek jhbuild modules Sep 09 21:54:59 like* Sep 09 21:55:37 I can see that happening although perhaps not the way most do Sep 09 21:56:34 For example, I can it being possible to write a parser which would translate ubnutu src debs into bb recipes Sep 09 21:58:30 didn't we have an srpm backend eons ago? Sep 09 21:59:49 RP: having bitbake backends for srpm/dsc/ebuild/jhbuild/GAR would be sweet Sep 09 22:00:10 we need to force kergoth to complete BitBake2 Sep 09 22:00:29 zecke: Personally, I don't think that's going to happen Sep 09 22:00:54 koen: An each would result in a separate set of metadata Sep 09 22:01:22 koen: I think it could be very powerful, especially if you support packages staging with each Sep 09 22:01:57 yep Sep 09 22:02:10 s/packages/packaged/ Sep 09 22:06:37 'night all Sep 09 22:08:17 RP: hmm, this is an interesting scope issue Sep 09 22:08:23 even globals() is per module Sep 09 22:09:19 zecke: yes :-/ Sep 09 22:11:18 Hello Sep 09 22:11:27 hi sirfred Sep 09 22:11:32 RP: Hey! Sep 09 22:11:35 RP: Seen my mail? Sep 09 22:12:05 even __builtins__ a per module Sep 09 22:12:13 sirfred: I have now - good news :) Sep 09 22:12:34 RP: I think so. It's not an spectacular speed, but I think it's better than before. Sep 09 22:12:43 And it's the IMAGEON making the work. ;-) Sep 09 22:12:54 sirfred: did you get profiling working? Sep 09 22:13:01 RP: Yes, before this change. Sep 09 22:13:13 RP: And I got ~30% in w100fb_sync. Sep 09 22:13:16 sirfred: If we have to wait for it do we gain anything - this is the question Sep 09 22:13:37 With mardy's Xw100, we do as X can be doing other things but I'm not sure about the kernel Sep 09 22:13:54 RP: But, I was scrolling all the time. It's reasonably that w100fb_sync is lasting that much. Sep 09 22:14:06 sirfred: yes, I agree Sep 09 22:14:08 RP: Take into account that we're just waiting for the IMAGEON to finish. Sep 09 22:14:34 RP: I would like you to test the patch. I think the speed is significantly better Sep 09 22:14:58 sirfred: I will try but I can't promise when :-( Sep 09 22:15:00 RP: And yes, I agree with the fact that perhaps a fb accelerated driver is not very useful. But, once we have it, ... Sep 09 22:15:09 RP: I will pass that question on to mithro or mickeyl Sep 09 22:15:27 zecke: ok, thanks. At least its not just me :) Sep 09 22:15:43 zecke: It might be __init__.py should call it Sep 09 22:15:53 RP: we could have a default value in __init__.py Sep 09 22:16:01 sirfred: One thing the kernel could do is scrolling as that would speedup the console a lot Sep 09 22:16:14 zecke: snap :) Sep 09 22:16:17 but we risk having two users of bb.msg.note with two different debug levels Sep 09 22:16:24 RP: But now, also font drawing is faster. Sep 09 22:16:32 zecke: yes :-/ Sep 09 22:16:40 RP: And we don't need to sync, because all the operations are being made into hardware. Sep 09 22:16:42 actually thinking Object Orientated we would need to pass an object around Sep 09 22:16:50 the big bad Bitbake instance Sep 09 22:16:57 sirfred: I'm really not sure you can say that about the syncing Sep 09 22:17:16 zecke: I was trying to avoid that Sep 09 22:17:23 yep Sep 09 22:17:38 RP: What do you think? Sep 09 22:17:45 it wasn't written with that in mind :( Sep 09 22:18:00 sirfred: As I said before, we need to find out what that sync protects Sep 09 22:18:25 RP: Why do you think it is protecting any other thing that graphic operations order? Sep 09 22:18:31 sirfred: There is some memory lurking which is bothering me, not sure what though :-/ Sep 09 22:18:36 sirfred: yes, I worry that it does Sep 09 22:19:00 RP: You can take a look at the other hardware drivers. Sep 09 22:19:22 RP: I don't think they're intending sync in another meaning other than waiting for hardware operations to finish. Sep 09 22:19:25 sirfred: I'd ask the framebuffer maintainer Sep 09 22:19:36 RP: That's a good idea. Sep 09 22:19:45 sirfred: If we alter the framebuffer address instead of redrawing it, we wouldn't even need to blit the fonts for scrolling ;-) Sep 09 22:20:30 RP: But you cannot offset the address forever. Later or sooner you will have to scroll, don't you? Sep 09 22:20:58 sirfred: Yes, but you only have to redraw once every several lines instead of every time = speedup Sep 09 22:21:33 RP: That could be an improvement, but perhaps the result is not as stable as we would like. Sep 09 22:21:42 zecke: We're slowly making it more OO orientated but the problem would be flowing from bitbake into OE Sep 09 22:21:52 RP: Another improvement I was thinking about was copying the entire font to the graphic offscreen memory. Sep 09 22:21:56 sirfred: I think it would work very well Sep 09 22:22:34 (the scrolling). The fonts in offscreen memory might help but it depends whether we can use the downtime Sep 09 22:22:53 I don't think hardware accell in the kernel can :-( Sep 09 22:23:08 RP: I have to leave now. Sep 09 22:23:22 RP: Anyway, perhaps if you try the patch... could change your mind. ;-) Sep 09 22:23:25 sirfred: ok, thanks for letting me know - i'll try to test Sep 09 22:23:31 Good night. Sep 09 22:23:38 'night sirfred Sep 09 22:24:41 'night all Sep 09 22:25:27 'night cyrilRomain|zZz Sep 09 22:28:04 cyrilRomain|zZz: good nite Sep 09 22:49:24 03rpurdie * r594 10bitbake/lib/bb/providers.py: providers.py: Improve findBestProvider debug messages to explain the infamous 'preferred version 2.4 of glibc not available' messgae Sep 09 23:14:28 03rpurdie * r595 10bitbake/ (bin/bitbake lib/bb/cache.py lib/bb/shell.py): Turn BBParsingStatus into CacheData and move to cache.py Sep 10 01:39:07 * v8jlene is back Sep 10 02:15:20 03lenehan 07org.oe.dev * r7a92a5cb... 10/ (9 files in 4 dirs): make: Add version 3.81 for both make and make-native. Sep 10 02:30:18 03lenehan 07org.oe.dev * r0cce8d25... 10/ (4 files in 2 dirs): Sep 10 02:30:18 libpng 1.2.12: Disable the use of x86 assembler code. When building Sep 10 02:30:18 for i586 or i686 it attempts to use the assembler code but does not Sep 10 02:30:18 link it in, so either the assembly code needs to be disabled or the Sep 10 02:30:18 makefiles patch to link in the assembly code. Closes #1326. Sep 10 02:48:07 03lenehan 07org.oe.dev * r8d7a934e... 10/ (1 packages/conserver/conserver.inc): conserver: Remove spurious .debug directories from the main packages. Sep 10 02:48:23 03lenehan 07org.oe.dev * rf7b20af1... 10/ (4 files in 2 dirs): Sep 10 02:48:23 conserver: Fix up the PR's which I needed to bump them in the main .bb Sep 10 02:48:23 files, not in the .inc file like I did in the last commit. Sep 10 02:48:23 03lenehan 07org.oe.dev * r046ba1dd... 10/ (4 files in 2 dirs): conserver: Run sanitize.py over all the files. Sep 10 02:56:39 03lenehan 07org.oe.dev * rf7624d7e... 10/ (1 site/i586-linux): site i586-linux: Add required entry for sudo. **** ENDING LOGGING AT Sun Sep 10 02:59:56 2006