**** BEGIN LOGGING AT Wed Jun 29 02:59:56 2011 Jun 29 06:09:30 good morning all of u Jun 29 07:49:02 good morning Jun 29 07:58:32 ant_work: I am unable to enable the touchscreen on mini6410 board ,have u worked on the touchscreen module Jun 29 08:11:24 morning all Jun 29 08:15:59 good morning Jun 29 08:16:52 erwt: hello. I don't know your board. Fwiw I have enough issues with ads7846 touchscren as well ;) Jun 29 08:19:02 ant_work: what are you facing to with ads7846? Jun 29 08:19:28 we used corgi_ts before. this new driver is very noisy Jun 29 08:19:46 so noisy that calibration does fail Jun 29 08:20:31 and that's strange: there is indeed a 'filtering' inside Jun 29 08:25:01 ant_work: i am using friendly arm board mini6410 Jun 29 09:53:18 i m working on changing screen resolution with fbdev on MIPS dev3210 board for last 3-4 days... but still I m failling... can anyone help me.... how to change screen resolution with fbdev? Jun 29 09:53:29 hm webserver down again? Jun 29 09:53:39 hm its slow again Jun 29 09:54:23 sanket sorry I told you what I know Jun 29 09:54:49 sanket_: there's no point in just asking the same question over and over again. the answer is still the same: use fbset, and if your kernel doesn't support mode changing then you will have to fix it until it does. Jun 29 09:55:06 pb ;) Jun 29 09:55:17 didnt know he is asking now everyday Jun 29 09:56:02 heh. well, maybe not _every_ day, but I'm pretty sure he asked yesterday and on monday. Jun 29 10:00:31 hej, sources.oe.org still active, or how can I get a file hosted there ? Jun 29 10:08:16 jo mickeyl Jun 29 10:18:20 mlip: ping ka6sox-away I think is the usual method Jun 29 10:18:34 he's probably not around atm though Jun 29 11:00:17 bluelightning, ty Jun 29 14:00:55 quick question about RRECOMMENDS: is it possible for a layer to 'veto' installation of a package which is RRECOMMENDed by a recipe? Jun 29 14:01:30 I want to turn off the qt4-embedded demos, examples, unused libs etc, but it seems wrong to edit the qt recipe. Jun 29 14:01:35 iirc, oe-core doesn't support that Jun 29 14:01:49 you can do it in oe.dev but I don't think the feature was ever backported to oe-core Jun 29 14:02:32 if you're using oe.dev then the variable you need to set is BAD_RECOMMENDATIONS. Jun 29 14:02:49 if you're using oe-core then I guess you need to backport the appropriate bits of package_ipk.bbclass and then proceed as before :-} Jun 29 14:02:58 Using oe-core. Jun 29 14:03:52 It seems crazy that you'd get lots of packages inappropriate for an embedded rootfs installed with qt4-embedded, with no way to turn them off. Jun 29 14:04:01 celston you dont want to package it or you dont want to build it? Jun 29 14:04:07 Hi everybody Jun 29 14:04:22 or you dont want the package to be installed? Jun 29 14:04:34 I don't mind if it's built and packaged, but I'd rather not put it in the rootfs. Jun 29 14:04:43 is there a problem with cloning openembedded? Jun 29 14:04:54 I got some errors when I tried to clone it Jun 29 14:04:55 celston: if they're grossly inappropriate then they probably oughtn't to be in Recommends in the first place Jun 29 14:05:01 $ git clone git://git.openembedded.net/openembedded Jun 29 14:05:01 Cloning into openembedded... Jun 29 14:05:01 remote: Counting objects: 432195, done. Jun 29 14:05:01 remote: Compressing objects: 100% (123105/123105), done. Jun 29 14:05:01 error: inflate: data stream error (incorrect data check)KiB/s Jun 29 14:05:02 fatal: pack has bad object at offset 20888398: inflate returned -3 Jun 29 14:05:03 mgundes server is again under high load Jun 29 14:05:04 fatal: index-pack failed Jun 29 14:05:16 mgundes: please clone it from github Jun 29 14:05:36 RRECOMMENDS_${PN} = "${LIB_PACKAGES} ${OTHER_PACKAGES}" Jun 29 14:05:44 Where OTHER_PACKAGES is a list of ALL QT packages. Jun 29 14:05:50 crumbs Jun 29 14:05:52 I think that's just bogus Jun 29 14:06:09 I lie, it's not dbg and dev packages. Jun 29 14:06:35 pb_ what do you mean with github? sorry I am not good at in git Jun 29 14:06:43 can you write url Jun 29 14:06:53 mgundes: where did you find the instructions you are using for cloning? Jun 29 14:07:19 anyway: git clone git://github.com/openembedded/openembedded.git Jun 29 14:07:52 ok thanks pb_ Jun 29 14:08:58 Thanks guys, I'll look for BAD_RECOMMENDATIONS. How likely is it that oe-core would accept this feature? Jun 29 14:09:27 dunno, maybe 50/50? Jun 29 14:10:03 I think the chances are better than even that they would accept some form of that feature, but I'm not sure whether they would take it as is. Jun 29 14:10:28 pb_, unfortunately I got same error Jun 29 14:10:29 $ git clone git://github.com/openembedded/openembedded.git Jun 29 14:10:30 Cloning into openembedded... Jun 29 14:10:30 remote: Counting objects: 432195, done. Jun 29 14:10:30 remote: Compressing objects: 100% (123120/123120), done. Jun 29 14:10:30 error: inflate: data stream error (incorrect data check)MiB/s Jun 29 14:10:31 fatal: pack has bad object at offset 2707357: inflate returned -3 Jun 29 14:10:32 fatal: index-pack failed Jun 29 14:10:58 ah Jun 29 14:11:00 mgundes thats look like something os broken Jun 29 14:11:03 at your side Jun 29 14:11:04 maybee Jun 29 14:11:07 in that case it would appear that something is broken on your local host Jun 29 14:11:16 hm Jun 29 14:11:23 can you clone other repos successfully? Jun 29 14:11:24 filesystem Jun 29 14:11:39 I can try with another user Jun 29 14:14:13 hmm same error with another user Jun 29 14:14:14 actually I already have another openembedded sources and I wanted to get it again Jun 29 14:14:42 may this be reason? Jun 29 14:18:28 I think you are right, I could not clone another project git repository too Jun 29 14:20:48 anyway, I will reboot my system and try again then, thanks a lot Jun 29 14:37:44 hi again Jun 29 14:38:00 i wish i could, from within gmail, get a link to the current email within the gmane interface for the list it's in Jun 29 14:38:08 i'm dreaming, i know, but it sure would be handy Jun 29 14:38:47 useful, indeed Jun 29 14:38:58 did anyone have an opinion on http://permalink.gmane.org/gmane.comp.handhelds.openembedded/46236 ? Jun 29 14:39:47 guh, i hate computers at times Jun 29 14:43:54 kergoth: yeah, seems reasonable. I doubt it will even have much performance impact. Jun 29 14:44:55 yeah i agree. i'm sure that was the intent behind it, but it may well have been done without even profiling it Jun 29 14:45:01 heh Jun 29 14:45:56 heh, turns out that if you forget to put bitbake in your $PATH, the oe-core wrapper script just passes your bitbake args to pseudo instead :-} Jun 29 14:46:10 leads to bizarre errors until you work out what's going wrong. Jun 29 14:46:21 Hi All, I'm getting an odd issue with meta-toolchain anyone else seeing the following? Jun 29 14:46:53 gah, lost the terminal. Basically was unable to find libc >= 2.12 in populate sdk. Jun 29 14:47:10 Ah, rerun finished. Jun 29 14:47:16 Collected errors: | * satisfy_dependencies_for: Cannot satisfy the following dependencies for task-core-standalone-sdk-target: | * libc6 (>= 2.12) * libc6 (>= 2.12) * | * opkg_install_cmd: Cannot install package task-core-standalone-sdk-target. Jun 29 14:47:21 pb_, I tested with my friend on his system and got again same error while cloning openembedded Jun 29 14:47:27 can you just test it please Jun 29 14:47:41 I think there should be error on repository side Jun 29 14:48:02 error is that Jun 29 14:48:03 error: inflate: data stream error (incorrect data check)KiB/s Jun 29 14:48:03 fatal: pack has bad object at offset 718318: inflate returned -3 Jun 29 14:48:03 fatal: index-pack failed Jun 29 14:48:34 doh, bitbake -b broken again Jun 29 14:49:18 as usual on days of the week that end in "y", I suppose Jun 29 14:49:55 :) Jun 29 14:51:10 mgundes: I just did a fresh clone from github and it worked fine Jun 29 14:51:38 hm Jun 29 14:52:09 interesting Jun 29 14:55:28 anyway, thanks pb_, I will try on my home pc again Jun 29 14:56:27 hmm Jun 29 15:00:01 Ah, couple of fixes pushed out by RP in last few mins. Looks relevant, so guess my issue may be fixed. Jun 29 15:02:24 if they went to master, they should have shown up here via CIA, no? or did he only push them to poky? Jun 29 15:03:35 hmm. I guess I may have not pulled for a few mins before posting... Jun 29 15:03:36 Not sure. Jun 29 15:04:33 or longer by the look of it. Jun 29 15:04:48 oops. Jun 29 15:05:24 Fix was actually Koen's at 2.23pm Jun 29 15:05:42 in openembedded-core Jun 29 15:05:46 :) Jun 29 15:14:16 kergoth the server has some problems again Jun 29 15:15:05 which server, melo? Jun 29 15:15:18 yes Jun 29 15:15:24 what's wrong with it? Jun 29 15:15:26 I think so Jun 29 15:15:30 slow Jun 29 15:15:32 15:13:14 up 6 days, 15:17, 4 users, load average: 0.10, 0.07, 0.08 Jun 29 15:15:37 not much evidence of load Jun 29 15:15:43 hm Jun 29 15:15:45 what exactly is slow? Jun 29 15:15:51 webserver Jun 29 15:16:02 and where is cgit webview gone? Jun 29 15:17:25 yeah, apache does seem slow for some reason Jun 29 15:18:06 pb_: I've had a look at how BAD_RECOMMENDS is implemented in oe.dev - it's only applicable to the ipk rootfs. I'm wondering if it would be more acceptable to have any packages in BAD_RECOMMENDS filtered out of each package's metadata as it goes through package.bbclass. Jun 29 15:18:47 mm, dunno Jun 29 15:18:48 that doesn't seem like a good idea, imo. that variable should only affect the filesystem construction, otherwise they won't show up in the feeds Jun 29 15:19:02 at the moment, BAD_RECOMMENDATIONS is per image. your proposal would make it static per DISTRO Jun 29 15:19:28 on the other hand, it is true that with the current system, the bad recommendations will still be honoured if you install the package online on the target, and that might not always be what you want. Jun 29 15:19:40 Ah, that was sort of my plan :) A per-distro way to veto recommended packages. Jun 29 15:20:24 If you want to do it statically in the distro then you can just clobber RRECOMMENDS from a .bbappend or your distro.conf Jun 29 15:20:57 As it is, in oe-core there's no scope to leave out RRECOMMENDed packages from a layer. Which is exactly the sort of thing one would want to do to slim down an embedded rootfs. Jun 29 15:21:07 oe wiki is way slow these days. Melo does not seem overloaded Jun 29 15:21:21 is distro.conf parsed *after* recipes? Jun 29 15:21:33 cbrake: yeah, we were just talking about that. I couldn't quite see what was going on: no big listen backlog in apache, and no load, it just doesn't seem to be serving requests very quickly for some reason Jun 29 15:21:37 celston: no, but you can use overrides Jun 29 15:21:54 celston: au contraire, there are multiple ways you can do that in oe-core today. Jun 29 15:22:17 if you never want to even build the bad packages, you can mask them out or force them to skip at the parse stage Jun 29 15:22:18 is that why BAD_RECOMMENDATIONS is not in oe-core? Jun 29 15:22:43 if you want to build them but disregard the recommendations then you can set an override on the recommending package, or provide a python method to mangle the metadata in some custom way Jun 29 15:22:57 Can't skip at parse stage when you want to run the recipe, but not generate a specific package of that recipe, right? Jun 29 15:23:11 no, I think the main reason BAD_RECOMMENDATIONS isn't in oe-core is just because oe-core was forked from oe before it was added Jun 29 15:23:47 celston: so just override PACKAGES instead of skipping, if you want to do that Jun 29 15:24:15 celston: right. you can h4x0r PACKAGES, though you need to be a bit careful with that because you might just end up with the FILES for the package you deleted landing in another one. Jun 29 15:25:11 I don't think there's any convenient way to say "do population for this package as normal but throw away the results". That might be a useful thing to add. Jun 29 15:25:13 so something like: PACKAGES = ${@oe_filter_out('list of packages to veto', '${PACKAGES}', d)} ? Jun 29 15:25:46 probably := not =, but something along those rough lines Jun 29 15:25:55 modulo what I said above about the FILES Jun 29 15:26:12 I don't thing screwing with what packages get emitted is generally the best plan. Generally speaking we emit everything at all times and then just do different things with the results. there are certainly corner cases, but that's the general policy. if you're concerned about bad recommendations only being implemented by rootfs_ipk, then implement it in the others Jun 29 15:26:23 (though i really think we need to refactor the rootfs construction, we have a lot of duplication right now :\) Jun 29 15:26:50 yeah, I was thinking about that the other day. the trouble is that although there's a lot of conceptual duplication, most of the actual code is not common. Jun 29 15:27:04 So preferred would be use BAD_RECOMMENDATIONS, and implement in rpm and deb? Jun 29 15:27:12 true. would need a lot of hooks Jun 29 15:27:31 celston: that depends on your needs. do you also need them to not be pulled in when installed from a feed? Jun 29 15:27:41 if you don't care about that case, then i'd say just implement in the other rootfs packages Jun 29 15:28:36 Not using feeds at the moment, nor likely to. Embedded platform. Jun 29 15:28:38 kergoth: for example, in the case of BAD_RECOMMENDATIONS, rootfs_ipk implements it by just preloading the opkg status file with entries for the packages you don't want. It's only about a half-dozen lines of code and, except for the loop that iterates over the packages, it's all ipkg specific Jun 29 15:29:18 ah, good point Jun 29 15:29:24 hmm Jun 29 15:29:52 it's annoying when we keep hitting these cases where something isn't implemented in the others, but maybe that should just be handled in the commit review process, don't add a feature to just one Jun 29 15:29:55 but this isn't the first time Jun 29 15:30:08 PACKAGE_INSTALL_ATTEMPTONLY started ipk specific too Jun 29 15:30:08 This is why I started to look further back up the process, to see if there was a common point to implement it (although I'm stuggling against my lack of knowledge :-) ) Jun 29 15:30:30 yeah. on the other hand, seems a bit harsh to require everyone who wants to add something to rootfs_ipk to set themselves up to test all the others as well. Jun 29 15:30:57 true Jun 29 15:31:11 I have always taken the view that the crazed rpm users can add these features for themselves :-} Jun 29 15:31:18 To me, the way to get this feature into all packages is to have the bad recommendations list modify the package metadata emitted. Jun 29 15:31:25 it'd be nice to have a central place where you add hooks, and if the others don't support it, they don't have those hooks, and it could explicitly warn if you try to opt-in and it isn't implemented by your packaging of choice Jun 29 15:31:41 celston: yeah. well, as we discussed before, that would be a slightly different feature. Jun 29 15:31:47 not necessarily a bad one, but not the same thing. Jun 29 15:32:01 and, if you want to do that, you can do it already without any core hacking. Jun 29 15:32:53 ah ha! no wonder i've been being bitten by this. turns out its not always safe to assume provided zlib-native, even if the dev files are installed. something must require a specific version. end up with python-native without zlib support, for some reason Jun 29 15:33:01 * kergoth removes from his automagic assumptions code Jun 29 15:33:34 wow Jun 29 15:33:41 I guess you could have a preinstall packages hook which did whatever each packaging system required to make it *think* the package was installed (a la ipk) Jun 29 15:33:42 zlib hasn't changed since, what, 1957 or something? Jun 29 15:33:47 ? Jun 29 15:33:53 last year Jun 29 15:34:08 huh, wonder why its doing that. i really can't be bothered to find out at this point Jun 29 15:34:14 yeah, weird Jun 29 15:35:11 heh, screwingw ith assume provided rather reminds me of ccache Jun 29 15:35:20 may well work great most of hte time, but once in a whiel you get bitten in the ass Jun 29 15:38:41 * kergoth still really favors the whole sane chroot approach rather than the everything-and-its-brother-is-a-native-recipe Jun 29 15:39:21 Hmm. finaly get back to the sdk populate and now it gives: Jun 29 15:39:23 | * satisfy_dependencies_for: Cannot satisfy the following dependencies for task-core-standalone-sdk-target: | * eglibc * eglibc-dev * | * opkg_install_cmd: Cannot install package task-core-standalone-sdk-target. Jun 29 15:39:30 something feels rather circular to me... Jun 29 15:45:14 In meta-toolchain which var I can use to build the path to the bindir of host binaries? Jun 29 15:45:48 bindir_native I think Jun 29 15:46:06 fray: this is nativesdk related Jun 29 15:46:18 bindir_nativesdk then Jun 29 15:46:25 I used "${SDKPATHNATIVE}${bindir_nativesdk}" Jun 29 15:46:41 git it shows i686-i586 on the path Jun 29 15:46:43 I'm just looking at meta/conf/bitbake.conf Jun 29 15:47:11 Is there any mirroring on the openembedded-core repo? I could have sworn that pulling a new copy just sent me slightly back in time... Jun 29 15:47:19 jic23: github Jun 29 15:47:47 fray: I tried this: if [ -e ${SDK_OUTPUT}/${SDKPATHNATIVE}${bindir_nativesdk}/cmake ]; then Jun 29 15:47:50 this fails Jun 29 15:47:57 ya, I'm using github since melo had problems Jun 29 15:48:05 jic23: not transparent mirroring, if that's what you mean Jun 29 15:48:20 fray: this has been expanded to if [ -e /home/otavio/hacking/el/tmp-eglibc-eglibc/work/i586-oe-linux/meta-toolchain-ostt-osnt-1.0-r0/sdk/image//usr/local/oecore-i686-i586/sysroots/i686-oesdk-linux/usr/bin/cmake ]; then Jun 29 15:48:37 maybe I'm just imagining things... Jun 29 15:48:47 are you pulling from github, or from melo? Jun 29 15:48:48 I don't know.. sorry Jun 29 15:48:49 fray: the oecore-i686-i586 is wrong Jun 29 15:49:09 * fray pulls from github.. I suggest others do that as well Jun 29 15:49:17 damn, think i just got bit by the whole perl-native thing. does oe-core not have the perl-native fixes from oe.dev yet? Jun 29 15:49:17 hrm Jun 29 15:49:29 which fixes are those? Jun 29 15:49:37 kergoth Tartarus checked and thought they were all there Jun 29 15:49:39 the bits that made it an early dependency and all Jun 29 15:49:42 hmm, k Jun 29 15:50:22 Er, no? Jun 29 15:50:27 humm Jun 29 15:50:37 I pointed out a few more /usr/bin/env perl ones that were needed and koen had requested while saul was out Jun 29 15:50:53 And I'm not 100% convinced there aren't horrible problems with what we're doing with perl off on its own subdir Jun 29 15:51:04 But I haven't had the time to come up with the breaking cases yet Jun 29 15:52:12 Not to worry on mysterious git fun. I am just going mad. Found the commit I thought had disappeared. Jun 29 15:52:13 i wonder how hard it'd be to create a minimal chroot that has *every* -native dep we need, minus the bits we need patched like autoconf-native. might be an interesting experiment Jun 29 15:52:36 Still can't build eglibc based toolchain though. Jun 29 15:52:40 | checking for XML::Parser... configure: error: XML::Parser perl module is required for intltool Jun 29 15:52:43 * kergoth mutters Jun 29 15:59:03 hmm.. should meta-toolchain be trying to do eglibc-dbg? all the rest of eglibc just got renamed to libc6-? Jun 29 16:19:19 did I miss any overlays when searching for realtek/ralink firmware files, or are they just missing from meta-* so far? Jun 29 16:22:43 Anyone have any thoughts on toolchain missing eglibc dependency issues? Jun 29 16:24:06 is there an easy way to figure out where the dependencies are actually coming from? Jun 29 16:32:57 got to run so have posted query on mailing list... Jun 29 18:49:21 RP__: around ? Jun 29 18:50:05 RP__: I am seeing that python is holding 30% of ram through out the build from 2G ram I have Jun 29 18:50:19 this is the one process that is running bitbake Jun 29 18:50:52 30% ram is a lot Jun 29 18:53:13 and sometimes I see another instances of python which are shortlived but through out their life they also hold additional 30% of ram Jun 29 18:53:38 it is that bitbake is putting the datastore in ram all the time Jun 29 18:54:03 khem: the children aren't holding 30% of ram. its likely that its all shared with the main process Jun 29 18:54:09 try using smem instead of ps/top Jun 29 18:54:20 http://www.selenic.com/smem/ Jun 29 18:55:18 kergoth: ok Jun 29 18:55:21 insalling Jun 29 18:59:12 kergoth: ok so why is 30% Jun 29 18:59:16 do u know Jun 29 18:59:30 since I think its a constant steal though out the build Jun 29 18:59:43 thats like 600M of RAM Jun 29 19:00:10 hard to say. i did some work to improve the up front ram usage, but haven't looked at it since implementing parallel parsing Jun 29 19:00:35 you'd have to drop into a pdb shell at various points and inspect the heap with a python memory profiling tool (there are a few of them) Jun 29 19:00:47 thats how i cut the up front ram usage by 11% quite a while ago Jun 29 19:01:22 I am almost done with the sdk Jun 29 19:01:25 it can create scaling problems Jun 29 19:01:31 for bitbake Jun 29 19:01:39 Now, my problem is with the building and variables to use Jun 29 19:02:14 if I look at the ipk contents, as follow, it shows i686-i686 as local dir Jun 29 19:02:15 (devel)~/hacking/el% dpkg -c tmp/deploy/ipk/i686-nativesdk/cmake-nativesdk_2.8.3-r1.0_i686-nativesdk.ipk| grep bin/cmake Jun 29 19:02:18 -rwxr-xr-x root/root 4834884 2011-06-29 02:10 ./usr/local/oecore-i686-i686/sysroots/i686-oesdk-linux/usr/bin/cmake Jun 29 19:02:35 but when I build the sdk it is installed in i686-i586 local dir Jun 29 19:02:52 did u set SDKMACHINE Jun 29 19:03:00 * otavio checks Jun 29 19:03:07 khem: first place i'd say to look would be to make sure that the recipe info objects from the up front parsing get ditched after they're added to the cachedata. the parsed data from all recipes can take up a fair bit of ram Jun 29 19:03:23 kergoth: SDKMACHINE = "i686" Jun 29 19:03:24 pretty sure they do, but its one thing to check Jun 29 19:03:33 otavio: you missed Jun 29 19:03:35 :) Jun 29 19:04:13 damn, the git server is slow at fetching lately Jun 29 19:04:26 kergoth use github Jun 29 19:04:33 it seems to be planned :) Jun 29 19:04:53 so people dont pull from it directly Jun 29 19:08:08 :-) Jun 29 19:09:37 hmm Jun 29 19:09:44 i could use insteadofs to avoid having to change my .git/config's Jun 29 19:09:53 automatically redirect to github for pulls and upstream for pushes Jun 29 19:09:54 * kergoth tries that Jun 29 19:10:23 khem: any clue? Jun 29 19:10:42 I keep getting a QA issue No GNU_HASH in the elf binary for a custom package. All the custom package is doing is pulling some html webpages from an SVN repo and putting them into a package (bb). How could that fail an elf binary check? Jun 29 19:12:18 ahh, it turns out there is a pre-compiled java applet in there. I guess that is what is failing it. How can I get around that since the applet is suppose to be pre-compiled? Jun 29 19:14:18 found it: INSANE_SKIP_objc = "True" Jun 29 19:16:24 khem: https://github.com/kergoth/homefiles/commit/d502fe7 Jun 29 19:16:25 heh Jun 29 19:16:30 seems to work :) Jun 29 19:16:44 now i just git clone oe:foo and it pulls from github, and git push sends it to git.oe.org Jun 29 19:16:55 kergoth, thanks.... Jun 29 19:17:02 ka6sox-work: :) Jun 29 19:17:08 nope INSANE_SKIP_objc = "True" didn't stop the QA ERROR. Jun 29 19:20:59 wonder if the insteadof/pushinsteadof bits should go on the wiki or the mailing list or both, so people know how to pull from github and push to upstream easily. anyone that knows about insteadOf/pushInsteadOf won't have a problem, but not everyone knows about them Jun 29 19:21:05 * kergoth makes *heavy* use of them Jun 29 19:21:10 git clone poky:poky-contrib, etc Jun 29 19:21:22 just about every git server i find ends up with an alias like this :) Jun 29 19:22:55 actually INSANE_SKIP_${PN} = "True" did work, I just had a typo. Jun 29 20:50:59 pb_: ping Jun 29 21:13:43 kergoth: I do git remote add rw git@git.openemebedde.org:openemebedded Jun 29 21:13:55 then git push rw master:master Jun 29 21:14:06 if you use teh pushInsteadOf bits, you don't need to mess with that Jun 29 21:14:11 it'll automatically translate it Jun 29 21:17:07 hmmm Jun 29 21:19:36 I like it this way since I know what I am doing Jun 29 21:19:53 otherwise sometimes things get pushed out inadvertently Jun 29 21:21:16 generally, i don't push until i'm ready for it to go upstream. if i want it on github, i push to my github remote, rather than the main one. but that makes sense, i see your point, as an extra sanity check, forces you to be more explicit Jun 29 21:21:56 its a nice distraction and channels the mind a bit Jun 29 21:22:27 I am trying to bisect oe-core for a problem mips/uclibc Jun 29 21:22:31 and with layers Jun 29 21:22:34 guess what Jun 29 21:22:36 its hell Jun 29 21:22:41 hi khem Jun 29 21:23:04 so I have to keep going back on meta-oe etc. Jun 29 21:23:13 since the changes are dependendent Jun 29 21:23:32 * ant__ is battling with setup-scripts Jun 29 21:23:44 ant__: whats wrong there Jun 29 21:23:54 seems I had bad git checkouts Jun 29 21:25:05 so, I have an idea about how to reimplement the git fetcher. see https://gist.github.com/1054808 - and think of the ./git-cache as being DL_DIR. Jun 29 21:25:07 thoughts? Jun 29 21:25:12 * kergoth goes to take the dog for a walk Jun 29 21:27:24 khem: again, following your steps doing "git clone git://git.angstrom-distribution.org/setup-scripts" Jun 29 21:28:08 ant__: ok and then you need to checkout oe-core branch Jun 29 21:28:09 same error as before git clone git://git.angstrom-distribution.org/setup-scripts Jun 29 21:28:26 gen-site-config: command not found Jun 29 21:28:49 $ git clone git://git.angstrom-distribution.org/setup-scripts Jun 29 21:28:49 Cloning into setup-scripts... Jun 29 21:28:49 remote: Counting objects: 402, done. Jun 29 21:28:49 remote: Compressing objects: 100% (292/292), done. Jun 29 21:28:49 remote: Total 402 (delta 153), reused 231 (delta 74) Jun 29 21:28:51 Receiving objects: 100% (402/402), 55.32 KiB, done. Jun 29 21:28:54 Resolving deltas: 100% (153/153), done. Jun 29 21:29:19 something is wrong with your box Jun 29 21:29:29 andrea@mizar /oe/setup-scripts/sources/openembedded-core/scripts $ ls *g Jun 29 21:29:29 gen-site-config Jun 29 21:30:03 what is the error Jun 29 21:30:08 | /oe/setup-scripts/build/tmp-angstrom_2010_x-eglibc/work/i686-linux/zlib-native-1.2.5-r0/temp/run.do_siteconfig_gencache.12938: line 92: gen-site-config: command not found Jun 29 21:30:24 oh Jun 29 21:30:31 I thought you have cloning errors Jun 29 21:30:37 me too :/ Jun 29 21:31:07 are you on oe-core branch of setup-scripts Jun 29 21:31:07 let's check PATH's Jun 29 21:32:00 http://paste.debian.net/121454/ Jun 29 21:33:08 did you source ~/.oe/environment-oecore Jun 29 21:33:14 yes Jun 29 21:34:51 what were you building when this happens Jun 29 21:35:07 this is my tee Jun 29 21:35:08 http://paste.debian.net/121455/ Jun 29 21:35:32 METADATA_BRANCH = "" Jun 29 21:35:38 yep Jun 29 21:36:03 go to setup-scripts dir Jun 29 21:36:08 and do ./oebb.sh update Jun 29 21:36:18 I see there are other angstrom-setup-scrupts in gitorius. older. Jun 29 21:36:47 gitorious is not to be used now Jun 29 21:37:05 ok Jun 29 21:37:17 done Jun 29 21:37:33 retry Jun 29 21:37:43 cleaning tmp Jun 29 21:37:48 no Jun 29 21:37:53 argh Jun 29 21:38:00 too late ;) Jun 29 21:38:36 I dont know if folks have used gentoo so much with oe-core Jun 29 21:38:44 so u might end up with some odd balls Jun 29 21:39:06 that would be strange, being oe-dev builds fine Jun 29 21:39:34 they have diverged quite Jun 29 21:39:37 ok, almost there... (zlib-native) Jun 29 21:39:46 hmm so fast Jun 29 21:40:28 oh you did not delete sstate I guess Jun 29 21:40:33 which was smart Jun 29 21:40:40 knowingly or unknowingly Jun 29 21:40:45 no, I cleaned all Jun 29 21:40:59 did u rm -rf build/* ? Jun 29 21:41:16 yes, all inside build Jun 29 21:41:24 k Jun 29 21:41:27 same failure Jun 29 21:41:43 gen-site-config: command not found Jun 29 21:43:08 hmm Jun 29 21:43:11 ok Jun 29 21:43:17 I never saw this error Jun 29 21:43:24 open new shell Jun 29 21:43:34 and then source the ~/.oe/environment-oecore Jun 29 21:43:50 and cd into setup-scripts and redo bitbake console-image Jun 29 21:44:23 a, I was invoking bitbake from within home Jun 29 21:44:44 those scripts put cruft in /home Jun 29 21:44:53 no Jun 29 21:44:56 so I did stay there Jun 29 21:44:58 you should be in topdir Jun 29 21:45:16 yes, they do. I have oe on another partition... Jun 29 21:45:20 and infact you can be anywhere below topdir Jun 29 21:45:22 but not above it Jun 29 21:45:42 then why put the env-setup script there?? Jun 29 21:45:53 they are in ~/.oe Jun 29 21:46:16 ah, so it's me, having /oe on different mount Jun 29 21:46:24 if you have improvements by all means send the patches Jun 29 21:46:41 let me see Jun 29 21:47:49 ok, now the sanity checker worked Jun 29 21:47:51 http://paste.debian.net/121456/ Jun 29 21:48:20 ok now you stand better chance Jun 29 21:48:31 emerge mercurial first Jun 29 21:51:05 yes, doing that Jun 29 21:54:39 now terminal Jun 29 21:55:35 ok, parsing Jun 29 21:55:43 the key is to invoke bitbake from within topdir Jun 29 21:55:54 yes, I was used to do this Jun 29 21:55:54 I will add that as footnote to my blog Jun 29 21:56:05 even, always from inside BUILDDIR Jun 29 21:56:14 thats not needed Jun 29 21:56:19 with current setup Jun 29 21:56:35 it will find the builddir Jun 29 21:57:00 this mechanism was working long ago min oe-dev too, then somehow broke Jun 29 21:57:03 iirc Jun 29 21:57:32 ok, passed Jun 29 21:57:38 I'll clean all once more Jun 29 21:59:21 you dont have to Jun 29 21:59:21 khem: fyi http://paste.debian.net/121365/ Jun 29 21:59:55 ^^ this was on another Gentoo box lacking chrpath and quilt Jun 29 22:00:27 anyway you now know the right way Jun 29 22:00:34 same error: bitbake called from outside Jun 29 22:01:11 check_sanity_eventhandler' failed Jun 29 22:01:25 so no descriptive message Jun 29 22:01:55 yes bitbake devs know that bb does not do good job of error reporting Jun 29 22:02:27 ant__: so are you set now ? Jun 29 22:02:54 Yes, seems so, thx. I've launched a fresh build right now. Jun 29 22:03:05 k Jun 29 22:03:15 let me know if you need help Jun 29 22:03:25 thx again Jun 29 22:03:37 np Jun 29 22:04:03 ah, should I install pseudo now? Jun 29 22:04:26 Pseudo is not present but is required, building this first before the main build Jun 29 22:04:26 Parsing recipes. Jun 29 22:04:44 seems is building pseudo-native Jun 29 22:06:06 yes Jun 29 22:06:10 let it go Jun 29 22:06:57 last thing...to avoid intermittent races on Gentoo, I did ASSUME_PROVIDE linux-libc-headers-native Jun 29 22:07:06 with the old oe-dev Jun 29 22:07:39 let's hope the isue disappeared Jun 29 22:08:46 * ant__ is typing in the dark, hiding from nasty tiger mosquitos Jun 29 22:10:59 khem: any clue about the nativesdk issue? Jun 29 22:11:22 ant__: heh Jun 29 22:11:34 ant__: mosquitos; death to them heh Jun 29 22:12:26 ant__ heh I thought we had the monopoly on mosquitos in Minnesota.. Jun 29 22:13:25 http://www.mosquitaire.com/ Jun 29 22:13:33 german engineering Jun 29 22:14:01 expensive but working rather well (w/out CO2) Jun 29 22:14:43 hehe that wouldn't work here.. ;) Jun 29 22:15:02 problem is shere quantity of the little bastards Jun 29 22:15:41 yes, those are for gardens Jun 29 22:16:28 go to google maps.. put in "Webster MN".. go to the sat image.. you'll quickly seen why it won't work well here.. ;) Jun 29 22:16:42 cheapest: http://www.youtube.com/watch?v=aDHf2qnsNYY&feature=player_embedded Jun 29 22:16:57 3 drops of diesel Jun 29 22:17:55 ya.. people used to do that around here.. throw a small amount of diesel into the drainage ditches.. but if you are caught doing that these days its a huge fine Jun 29 22:18:21 people here experimented with vaseline lube ;) Jun 29 22:18:38 ewwwwww. Jun 29 22:18:49 * fray notes even though he lives in the rural area, he does his best to dump out any standing water.. Jun 29 22:19:04 of course the 15 acres of marsh land that I have, doesn't exactly help Jun 29 22:19:29 but diesel is what the old farmers did Jun 29 22:20:30 google for 'mosquito magnet' independence or patriot Jun 29 22:20:35 (normally this time of year the marsh next to me has dried out.. but not this year.. it still has standing water in most of it.. Jun 29 22:20:46 that the CO2 -- propane version? Jun 29 22:20:55 is american stuff ;) Jun 29 22:21:17 yup.. thats the one.. Jun 29 22:21:32 I know people who have them and say they are really good.. and don't burn through much propane.. Jun 29 22:21:53 the cost though has turned me off on them.. cheeper to buy DEET spray... Jun 29 22:21:57 it starts to be an issue here around Mediterranean see Jun 29 22:22:18 those beasts came from Asia Jun 29 22:22:24 where abouts are you? Jun 29 22:22:50 Northern Italy / Verona (Garda Lake) Jun 29 22:23:29 you'll see the mosquitos on Google Maps ;) Jun 29 22:23:55 ahh ok.. I spent 7 weeks in Turin (and Veneria specifically) a few years ago.. didn't really notice the mosquitos while I was there.. Jun 29 22:24:09 people says those have been 'imported' together with used tires Jun 29 22:24:17 a few years ago Jun 29 22:24:25 wouldn't surprise me Jun 29 22:24:33 with a ship Jun 29 22:24:43 probably in Venice Jun 29 22:24:59 maybe from China, they say... Jun 29 22:25:23 who knows... Jun 29 22:26:05 big concern here for mosquitos is West Nile virus.. affects live stock, specifically horses and birds Jun 29 22:26:17 and people Jun 29 22:26:43 ya.. we don't get a lot of cases though of people getting sick.. but lots of horses... Jun 29 22:26:53 ah Jun 29 22:27:12 blue jays, black birds and horses seem to be the primary causalty of west nile here.. Jun 29 22:27:43 khem: warnings about missing license for libgcc Jun 29 22:30:10 and gcc-runtime Jun 29 22:51:10 khem: toolchain + natives built correctly on Gentoo, now the restshould be a cup of tea Jun 29 22:52:10 no races until now Jun 29 23:19:17 fray: in San Jose the virus is found in people too :) Jun 29 23:19:37 I keep reading cases in Newspaper every now and then Jun 29 23:20:46 is it actually affecting them (beyond flu-like symptoms?) Jun 29 23:21:08 around here for the first couple years lots of folks got really sick.. but that, for the most part, has stopped.. Jun 29 23:52:56 good night Jun 30 01:39:01 03Tom Rini  07master * r0bc40262c9 10openembedded.git/recipes/opkg-utils/opkg-utils_svn.bb: Jun 30 01:39:01 opkg-utils: Add LICENSE field Jun 30 01:39:01 Quick code check and COPYING file and it's GPLv2 (or later) Jun 30 01:39:01 Signed-off-by: Tom Rini **** ENDING LOGGING AT Thu Jun 30 02:59:57 2011