**** BEGIN LOGGING AT Fri Apr 08 02:59:58 2011 Apr 08 07:51:19 morning Apr 08 08:33:03 morning all Apr 08 11:33:50 hi obi Apr 08 11:34:02 hi woglinde! Apr 08 12:43:56 I am having an image that busybox ends up with user 1000 instead of root and thus failing to run as root Apr 08 12:44:08 someone has an idea what might be wrong there? Apr 08 12:44:32 oe-core or oe Apr 08 12:44:36 oe Apr 08 12:45:09 Might your UID on the host be 1000? (which might indicate a problem with fakeroot) Apr 08 12:45:48 mwester-laptop: yes Apr 08 12:51:49 mwester-laptop: any idea what might be wrong with it? Apr 08 12:52:46 I'm building an image to see if the same happens here, but it will take some time (laptop is very slow). Apr 08 12:54:10 You can check the do_install (and releated) log files for busybox to see if there are any msgs that might help. Apr 08 12:54:43 What host, btw? Apr 08 13:07:13 it happens here too Apr 08 13:07:20 didn't had time to debug it why Apr 08 13:07:39 ubuntu 10.04 x86_64 here Apr 08 13:08:08 hm copy or tar operation Apr 08 13:08:40 as I said I don't know, the uid is wrong in the tar Apr 08 13:09:05 it must be some commit in the last two weeks, it was working fine before Apr 08 13:09:20 hm Apr 08 13:09:26 bitbake one maybee Apr 08 13:18:22 otavio: which bitbake verision are you using? Apr 08 13:19:10 http://git.openembedded.org/cgit.cgi/bitbake/commit/?id=ee39de6856b628df792c4a56966528796fbe2c8e Apr 08 13:21:35 any good way to override /etc/issue ? Apr 08 13:22:05 cd oe; find . -name issue Apr 08 13:42:18 ynezz, as expected: Apr 08 13:43:05 check_data_file_clashes: Package ...-issue wants to install file .../etc/issue.net But that file is already provided by package * base-files Apr 08 13:43:08 angstroem is overriding it Apr 08 13:44:55 woglinde, any way to easily change it? Apr 08 13:46:06 ? Apr 08 13:46:16 make an issue file your distro Apr 08 13:46:20 +for Apr 08 13:47:29 woglinde, yes, override it to say some more useful info Apr 08 13:51:21 you should read about .bbappend Apr 08 13:51:39 make your own layer using it and tweak the stuff there Apr 08 13:52:09 sakrah.homelinux.org/blog/2010/11/bblayers-bbappend Apr 08 13:52:37 ynezz jupp I would answer the same Apr 08 13:54:00 ynezz, thanks! Apr 08 13:59:03 Assuming the permissions problem is for those using bitbake maste,r its likely the fakeroot change in bitbake. it was fixed for oe-core, but broke use of fakeroot in OE. It needs to be fixed, in the meantime you can revert the commit or back up to 1.12. Apr 08 13:59:06 ah, ynezz already linked it :) Apr 08 13:59:47 hmm, a first pass fix might not be too difficult Apr 08 14:00:58 kergoth_: testing Apr 08 14:01:11 kergoth_: is someone working to fix it on OE? Apr 08 14:01:29 well, it will almost certainly be either me or RP that fixes it Apr 08 14:01:34 i haven't gotten to it, and i assume he hasn't either Apr 08 14:01:38 i'll see if i can get it done this week Apr 08 14:01:48 its a change poky made, but they broke oe compat in doing it Apr 08 14:01:55 erm, get it done today Apr 08 14:01:56 that is Apr 08 14:02:11 the thing is, oe-core/poky runs hte entire bitbake build under pseudo, and turns it on and off using an env var Apr 08 14:02:23 this means install/package share the same info, so permissions from install get properly into packaging, etc Apr 08 14:02:31 *large* improvement over our current situation Apr 08 14:02:33 nice Apr 08 14:02:58 but in doing so, it stopped running tasks under fakeroot for OE, which doesn't run bitbake under pseudo, so we hit problems with install tasks that change ownership/permissions Apr 08 14:03:06 so going to add another var to let it do either/both Apr 08 14:03:36 cool, just let me know, I could test it during weekend Apr 08 14:03:45 (if you happen to fix it that fast) Apr 08 14:03:55 go go kergoth Apr 08 14:04:05 give me about 5 minutes Apr 08 14:04:11 :D Apr 08 14:04:54 hmm, i see two possible implementations. first pass might not be ideal performance wise, we'll see Apr 08 14:05:04 but it's only one additional var check in exec_func Apr 08 14:07:36 ynezz: cd bitbake; git fetch https://github.com/kergoth/bitbake master; git merge FETCH_HEAD Apr 08 14:07:44 ynezz: then set FAKEROOT = "fakeroot" in bitbake.conf or local.conf Apr 08 14:08:19 and see if that takes care of it. could just add a test $(id -u) -eq 0 || oefatal "aiee" to a do_install ;) Apr 08 14:08:29 hrm, actually, thats not right Apr 08 14:08:29 dangit Apr 08 14:08:33 * kergoth_ mutters Apr 08 14:08:39 that will run *all* shell tasks under fakeroot Apr 08 14:08:44 well, it won't hurt Apr 08 14:08:48 but its not the same as the previous behavior ;) Apr 08 14:08:52 still, worth testing Apr 08 14:08:58 i'll work on a better resurrection of old behavior Apr 08 14:09:02 as well as a deprecation message Apr 08 14:10:04 note that its also possible to run tasks under fakeroot using the new stuff. the new stuff uses env vars, and its possible run something under fakeroot via the env, by setting LD_LIBRARY_PATH + LD_PRELOAD Apr 08 14:10:13 but you need to know the exact soname of the fakeroot library Apr 08 14:10:25 if we *know* we're using fakeroot-native for everything, that could work Apr 08 14:10:37 ah Apr 08 14:11:39 e.g. LD_PRELOAD=libfakeroot.so LD_LIBRARY_PATH=${STAGING_LIBDIR_NATIVE}/libfakeroot Apr 08 14:11:40 or whatever Apr 08 14:11:47 if you set FAKEROOTENV = that Apr 08 14:12:15 okay, i did the first pass in that initial 5 minutes i asked for, but i went over for the proper fix ;) Apr 08 14:17:49 hm, it's still same Apr 08 14:18:00 you set FAKEROOT=fakeroot in local.conf? Apr 08 14:18:05 ah, damn :) Apr 08 14:18:05 So, re portmap, why did it need to list fakeroot do_install anyhow? Apr 08 14:18:11 good question Apr 08 14:18:19 Couldn't we just fix the recipes Makefile for whatever silly reason it needs that? Apr 08 14:18:23 also, i don't think every use of fakeroot was correctly pulling in fakeroot-native Apr 08 14:21:11 * kergoth_ is personally of the opinion that it'd be nicer to be explicit about our permissions declaratively the way rpm does in its spec files, and avoid this whole pseudo/fakeroot mess Apr 08 14:24:08 kergoth_: fakeroot has -i/-s options which allow to save state across tasks Apr 08 14:24:39 i know Apr 08 14:24:46 we never had any ways to use it Apr 08 14:25:01 also, there was more to the switch than that, i think. ask richard and mark hatle Apr 08 14:25:09 * kergoth_ doesn't know much about the pseudo stuff Apr 08 14:25:30 hm, it's still same, even with the FAKEROOT set Apr 08 14:26:02 i verified this improved version of the fix here using a test command checking id -u Apr 08 14:26:08 pushing in a second, hold Apr 08 14:26:44 yeah Apr 08 14:28:49 hm.. fakeroot again Apr 08 14:28:52 ynezz: okay, try dropping teh change you have now and re-do the fetch/merge from my github Apr 08 14:33:48 kergoth_: does our bitbake support extract param in SRC_URI? Apr 08 14:34:11 kergoth_: rpm-native fails in do_patch here when it was unpackage by OE bitbake Apr 08 14:34:16 unpacked Apr 08 14:34:52 not sure what you're referring to there. Apr 08 14:35:01 for us, unpacking is an oe thing, nto a bitbake thing, as we use the bb.fetch module Apr 08 14:35:04 SRC_URI = "http://www.rpm5.org/files/rpm/rpm-5.4/rpm-5.4.0-0.20101229.src.rpm;extract=rpm-5.4.0.tar.gz \ Apr 08 14:35:06 oe-core uses fetch2, which does unpacking Apr 08 14:35:10 i know Apr 08 14:35:15 i'm saying its not bitbake that does it for us Apr 08 14:35:23 so if it was supported at all, it'd have to be done in base.bbclass Apr 08 14:35:26 pretty sure the answer is "no" Apr 08 14:35:31 but i wanted to clarify :) Apr 08 14:36:32 but if it's in base.bbclass than why does it work with poky's bitbake and not with oe's (same metadata) Apr 08 14:37:06 it won't, unless you're setting BBFETCH2=1 Apr 08 14:37:23 and even then its questionable Apr 08 14:37:41 again, for oe-core/poky, its bitbake that does unpacking Apr 08 14:37:49 that's not the case for oe, as we use fetch, not fetch2 Apr 08 14:38:29 ynezz: any luck? Apr 08 14:39:32 yep, it works now (tm) Apr 08 14:39:35 thanks Apr 08 14:39:42 awesome Apr 08 14:40:26 kergoth_: in both cases there is BBFETCH2=True Apr 08 14:40:35 JaMa: it s houldn't be. Apr 08 14:40:38 * mwester-laptop kills build in progress, removes tmpdir, and updates everything... ;) Apr 08 14:40:39 your oe setup is wrong. Apr 08 14:40:43 oe-core is not oe Apr 08 14:41:00 kergoth_: to be sure.. I'm using oe-core Apr 08 14:41:03 FAKEROOT has to be set to fakeroot, let me add to oe Apr 08 14:41:07 oe-core is COMPLETELY DIFFERENT Apr 08 14:41:16 oe-core uses fetch2, and bitbake does unpacking Apr 08 14:41:21 you need to be clear about these things. Apr 08 14:41:26 oe and oe-core are completely different Apr 08 14:42:01 I know, I was expecting that you remember that I'm using oe-core/meta-oe/meta-efl/meta-shr for a while Apr 08 14:42:14 but sorry I didn't say it first Apr 08 14:42:58 well, i use oe-core most of hte time too, but that doesn't mean its all i use :) Apr 08 14:43:08 i also do mentor's product, and upstream oe, and.. ;) Apr 08 14:44:09 anyway, i'm not sure if the rpm extraction is in bitbake master or still in poky's bitbake Apr 08 14:44:11 03Chris Larson  07master * r262e3bc9fa 10openembedded.git/conf/bitbake.conf: Apr 08 14:44:11 bitbake.conf: set FAKEROOT=fakeroot Apr 08 14:44:11 This is needed for current bitbake to spawn 'fakeroot' flagged tasks. Apr 08 14:44:11 Signed-off-by: Chris Larson Apr 08 14:44:13 so to go back to original question, when I'm using oe-core + oe bitbake, should extract= param work? Apr 08 14:44:41 looks like it, yes, but there may well be fixes in poky's bitbake that they haven't pushed to master yet Apr 08 14:44:54 see line 700 of lib/bb/fetch2/__init__.py Apr 08 14:45:13 elif file.endswith('.src.rpm') or file.endswith('.srpm'): Apr 08 14:45:13 if 'extract' ? urldata.parm: Apr 08 14:45:14 unpack_file = urldata.parm.get('extract') Apr 08 14:45:16 looks promising Apr 08 14:45:20 (ignore the funky char) Apr 08 14:45:33 src.rpm Apr 08 14:45:35 but like i said, could well be bugfixes that aren't synced, last i looked there were still a number of those Apr 08 14:45:39 oh my good Apr 08 14:45:52 where are packages that are only src-rpm Apr 08 14:46:52 okay, now everyone can update both bitbake and oe and fakeroot should work Apr 08 14:46:52 03Chris Larson  07master * r00403c4d5a 10bitbake.git/lib/bb/build.py: Apr 08 14:46:52 build: run fakeroot if FAKEROOT is set Apr 08 14:46:52 FAKEROOT contains the fakeroot command to be run. This should not be set for Apr 08 14:46:52 those using pseudo + FAKEROOTENV, but should be set for people like upstream Apr 08 14:46:52 OE, and its value should be 'fakeroot' for them, for compatibility with the Apr 08 14:46:53 old mechanisms. Apr 08 14:46:53 Signed-off-by: Chris Larson Apr 08 14:47:11 kergoth_: this chunk is the same between OE bitbake master and poky master, but on that server I have bitbake-1.12, I'll diff with it Apr 08 14:47:26 oh, 1.12 won't have it, its a recent addition afaik Apr 08 14:47:40 see git blame lib/bb/fetch2/__init__.py Apr 08 14:48:36 hmm, no, it is there, or seems to be Apr 08 14:48:39 strange. Apr 08 14:48:42 diff is best, yes :) Apr 08 14:49:13 looks like there was "unpack" parameter for this before "extract" Apr 08 14:49:20 that explains it Apr 08 14:49:28 thanks for help Apr 08 14:49:38 ahh Apr 08 14:49:48 np, sorry for the confusion Apr 08 14:52:08 renamed in ab05499d1ca98fd20ea7f871313b58bf26236416 Apr 08 14:52:21 ah Apr 08 14:52:25 interesting Apr 08 14:52:36 * kergoth_ hasn't followed fetch2 development very closely Apr 08 14:52:38 maybe worth backporting to 1.12 if 1.12 should be enough for oe-core Apr 08 14:52:59 not sure if it should be or not Apr 08 14:53:07 dont' think we have an official bitbake version requirement for it yet Apr 08 14:54:25 ok, thanks Apr 08 14:57:08 actually, if you'd be willing, ask that question on the oe-core list :) Apr 08 14:57:10 * kergoth_ forgot to do that Apr 08 14:58:20 * JaMa leaving so maybe will forgot too :/ Apr 08 14:58:47 no worries, i'll remember and ask eventually, or someone else will Apr 08 14:58:51 may be a tsc question Apr 08 17:36:38 quiet day? Apr 08 17:37:43 yes Apr 08 20:21:48 too quiet Apr 08 21:03:17 and an eerie silence has descended upon #oe... Apr 08 21:04:05 like a tomb Apr 08 21:04:12 back to work Apr 08 21:04:21 heh Apr 08 21:04:33 troublemaker Apr 08 21:04:36 I'm working on trying to get things better...but its not happening :P Apr 08 21:05:37 * Crofton is backing up a work on progress sd card Apr 08 21:16:23 all of OE must be on a plane :-) Apr 08 21:16:46 <- enjoying cold beer Apr 08 21:17:42 heh thats an idea Apr 08 21:17:55 oe-core is broken and everyone is asleep Apr 08 21:19:42 it is? Apr 08 21:22:07 dp@xora-build:~/Work/oe-core/setup-scripts/build$ cat /home/dp/Work/oe-core/setup-scripts/build/tmp-angstrom_2010_x/work/armv7a-angstrom-linux-gnueabi/linux-libc-headers-2.6.37.2-r0/temp/log.do_install.19445 Apr 08 21:22:07 ERROR: Function 'do_install' failed (see /home/dp/Work/oe-core/setup-scripts/build/tmp-angstrom_2010_x/work/armv7a-angstrom-linux-gnueabi/linux-libc-headers-2.6.37.2-r0/temp/log.do_install.19445 for further information) Apr 08 21:22:19 recursive error messages not at all useful :-) Apr 08 21:23:20 XorA|gone: set FAKEROOT = "" Apr 08 21:24:04 kergoth_: special new feature :-) ? Apr 08 21:24:26 * mwester-laptop bets on unintended side-effect Apr 08 21:24:35 in order for bitbake to support pseudo for oe-core and fakeroot for oe, we had make it obey multiple vars Apr 08 21:24:53 FAKEROOTENV sets env vars, FAKEROOT is a command to run the task with Apr 08 21:25:07 oe-core has a remnant FAKEROOT which is just a copy of FAKEROOTENV Apr 08 21:25:13 now that bitbake obeys FAKEROOT, this breaks Apr 08 21:25:37 just need to remove the old unneeded FAKEROOT definition from conf/bitbake.conf in oe-core Apr 08 21:25:38 going to bitbake 1.12 branch is safe? Apr 08 21:25:42 i mentioned it to RP on irc earlier Apr 08 21:25:53 yeah, or you could revert the most recent bitbake commit Apr 08 21:26:01 (which will make it not work for oe, just oe-core) Apr 08 21:26:08 I hadnt realised I was on bitbake master until just after that copy/paste Apr 08 21:26:22 * XorA|gone kicks koen Apr 08 21:27:02 poor koen. Apr 08 21:27:35 kergoth_: FAKEROOT should not be defined in bitbake.conf anymore ? Apr 08 21:27:37 (snicker) Apr 08 21:27:58 khem: in oe-core, it shouldn't. in OE, it has to be Apr 08 21:28:27 kergoth_: did you propose patch for conf/bitbake.conf ? Apr 08 21:28:29 for both Apr 08 21:28:43 how about other FAKEROOTxxx vars Apr 08 21:28:45 oe is already in, since it was harmless in all cases and i have push access Apr 08 21:28:51 OE sets FAKEROOT only Apr 08 21:28:57 oe-core sets FAKEROOTENV and FAKEROOTDIRS Apr 08 21:29:02 right Apr 08 21:29:10 should these additional vars be removed as well Apr 08 21:29:12 i poked rp on irc about oe-core, haven't sent a patch to the list about it Apr 08 21:29:20 ok Apr 08 21:29:22 not sure what vars you're referring to. Apr 08 21:29:33 afaik oe-core sets FAKEROOT, FAKEROOTDIRS, FAKEROOTENV. just need to drop FAKEROOT and its fine Apr 08 21:29:34 FAKEROOTENV FAKEROOTDIRS Apr 08 21:29:44 those are not in oe, those are in oe-core Apr 08 21:29:50 they shouldn't be in oe, they shoudl be in oe-core Apr 08 21:29:54 so they're just fine as they are. Apr 08 21:29:54 yes I am talking about oe-core Apr 08 21:29:59 once again Apr 08 21:30:00 I guess oe is well seyt Apr 08 21:30:06 oe-core MUST SET FAKEROOTENV AND FAKEROOTDIRS Apr 08 21:30:17 yes. oe is fine, oe-core needs FAKEROOT removed Apr 08 21:30:21 like i said earlier :) Apr 08 21:30:25 * khem removes locally Apr 08 21:30:39 yes somehow its taking twice today for me Apr 08 21:30:40 can just set FAKEROOT="" in local.conf too Apr 08 21:30:40 :) Apr 08 21:30:43 hehe Apr 08 21:30:48 well, its friday Apr 08 21:30:53 probably sideeffect of getting sick Apr 08 21:31:02 processor is working at reduced MHz Apr 08 21:31:02 * mwester-laptop prescribes beer for everyone. Apr 08 21:31:08 hmm, BBLAYERS could to with excess /'s stripped Apr 08 21:31:29 khem, try working with a migraine and aphasia Apr 08 21:31:35 simple typo leads to // at end of expressions which doesnt match the regexps Apr 08 21:32:31 * mwester-laptop is still rebuilding the package feeds, just to ensure that the fakeroot problem is exorcised completely... Apr 08 21:32:33 and fixing that seems to make things happier Apr 08 21:32:43 ka6sox-home: right now its just runny nose and watery eyes Apr 08 21:32:45 :) Apr 08 21:33:09 mwester-laptop: is new slugos release done ? Apr 08 21:33:24 No, my paying job has gotten in the way. Apr 08 21:33:27 mwester-laptop: I think post this release you should think of using oe-core Apr 08 21:33:41 oh well bills needs to be paid Apr 08 21:33:45 so I understand Apr 08 21:33:50 I am, it seems, to be a Websphere expert by the end of next week. :o Apr 08 21:34:27 At least that's what the customer was told by our management... so, it's a bit of a challenge. Apr 08 21:34:32 :D Apr 08 21:34:32 * khem hasnt programmed userspace for quite a while Apr 08 21:34:59 kergoth_: thanks that explains my broken oe-core build Apr 08 21:35:07 I think there has to be a better phrase to describe the space in which one programs websphere... Apr 08 21:35:27 mwester-laptop: choose one Apr 08 21:35:37 mwester-laptop: outer space :) Apr 08 21:35:46 heh! I like that one! Apr 08 21:36:17 I'm just prejudiced against Java, I guess. I consider it the modern version of COBOL. Apr 08 21:37:23 As Dijksatra mentioned there needs to be a language for dummies Apr 08 21:37:55 so Average Joe can write programs without crashing the system Apr 08 21:38:19 Those days are gone when handful of smart asses only did computers Apr 08 21:38:37 place that only a bit earlier in the phrase Apr 08 21:39:57 Well, that's true. And I do have to admit that C++ is not the ideal OO language at all... Apr 08 21:40:10 At least Java makes more sense as an OO language. Apr 08 21:41:56 I'm tired of software that is built upon languages that encourage bad programming. Apr 08 21:42:25 s/encourage/enable,encourage,allow Apr 08 21:42:53 * mwester-laptop wonders what ka6sox-home thinks of OO Perl, then... :D Apr 08 21:45:17 python is as far towards that kind of stuff as I go... Apr 08 21:45:38 django fanclub member :p Apr 08 21:45:53 ynezz, you do django? Apr 08 21:46:02 oh, nope Apr 08 21:46:30 I took a 1hr seminar on HTML3 and thats the only "web" programming I know Apr 08 21:48:34 I prefer Spark Apr 08 21:48:45 patches sent, restart build :-) Apr 08 21:49:06 but thats pretty hardcore. Apr 08 21:59:06 hmmm Apr 08 21:59:26 I really want to spend more time in both Go and F# Apr 08 21:59:34 i quite like them, though i haven't used either for anything really Apr 08 22:05:26 kergoth_, are you referring to Go! or Google's Go? Apr 08 22:05:55 google's Apr 08 22:06:28 I'll go, bye Apr 08 22:07:04 oookay Apr 08 22:07:09 lately I've been using erlang...like it a lot. Apr 08 22:07:53 golang is enabled in gcc 4.6 Apr 08 22:08:01 it has a frontend now Apr 08 22:08:52 is F# like ML nextgen Apr 08 22:35:37 has anyone tried building networkmanager-openvpn recently Apr 08 22:36:24 there seems to be a version mismatch problem with networkmanager Apr 08 22:36:33 as described here: http://www.mail-archive.com/networkmanager-list@gnome.org/msg17342.html Apr 08 22:55:30 neilm, how did you hit that recipe? Apr 08 22:55:49 omap3-desktop-image Apr 08 22:57:35 I won't be home until next week, but I can poke it them Apr 08 22:57:36 then Apr 08 22:57:40 that recipe failed complaining about no members named import and export Apr 08 22:58:17 I'm trying to bump right now to see if it will work Apr 08 23:04:51 bump to networkmanager-openvpn_0.8.3.995 appears to have worked Apr 08 23:46:36 sed: can't read sed: No such file or directory??? Apr 08 23:47:01 Im going to guess my gcc is non functional then :-( Apr 08 23:49:28 * XorA|gone goes to sleep and sees what appears at end of night Apr 08 23:52:34 GL **** ENDING LOGGING AT Sat Apr 09 02:59:57 2011