**** BEGIN LOGGING AT Wed Jun 01 02:59:56 2011 Jun 01 06:40:10 hi folks Jun 01 06:46:30 I am using a snapshot OE (c.a 1 year old) patched in-house to fit our needs, one of those patches applies to the generate-manifest-2.6.py script (used, if I am correct, to generate the python-2.6-manifest.inc) Jun 01 06:47:31 my question is: when is generate-manifest-2.6.py called ? should it be called manually and its result used to generate a patch for python-2.6-manifest.inc ? Jun 01 06:53:26 Mithfindel: yes manually Jun 01 06:54:46 Mithfindel: and check current generate-manifest-2.6.py because I did few changes to make resulting python-2.6-manifest.inc diff smaller Jun 01 06:57:36 hi all Jun 01 06:58:07 JaMa: thank you for the tip, is your change related to the sorting of package names ? I think I have seen patches in this sense Jun 01 07:08:43 Mithfindel: yes sorting, because without I got different order after each run Jun 01 07:11:25 good morning Jun 01 07:11:56 morning all... Jun 01 07:33:09 JaMa: thanks a lot, nice job :) Jun 01 08:29:40 morning all Jun 01 09:55:05 RP__: "that" checkin is done now Jun 01 09:55:47 following some minor grief with my svn access to ltg. it seems that some unknown agency keeps deleting my authorized_keys file on discovery. :-} Jun 01 10:00:22 probably fusionforge, I suppose. Jun 01 10:04:52 pb_: I'm quite sure it is fusionforge. This beast is quite intrusive... Jun 01 10:05:38 I wish I had some free time to get it into a better shape :-( Jun 01 10:08:08 anyone here have access to delete my oe branches? Jun 01 10:10:33 not right now, but I can do it for you later. Jun 01 10:11:00 pb_: ok thanks, bluelightning/* have been merged, if you could delete those that would be great Jun 01 10:11:15 ok. please send me an email so I don't forget about it. Jun 01 10:13:36 will do Jun 01 10:34:36 pb_: thanks! Jun 01 10:41:36 asterisk can't start since it cant open its log file Jun 01 10:41:45 must fix :) Jun 01 11:20:05 03Paul Eggleton  07master * ra8ab0af776 10bitbake.git/lib/bb/cooker.py: (log message trimmed) Jun 01 11:20:05 bitbake: use layer priority when applying bbappends Jun 01 11:20:05 If the priority of a layer has been specified with BBFILE_PRIORITY_ then Jun 01 11:20:05 use that to sort the list of BBFILES entries, which will affect the order Jun 01 11:20:05 in which .bbappend files are applied. Jun 01 11:20:05 Fixes [YOCTO #1125] Jun 01 11:20:06 Signed-off-by: Paul Eggleton Jun 01 11:39:52 ehmm...would anyone pls try to build mtd-utils on armv6 or 7? Jun 01 11:40:05 I did dare to remove ARM_INSTRUCTION_SET = "arm" (built fine for armv5te) Jun 01 11:40:40 and on strongarm/armv4 too Jun 01 11:41:10 I guess this has been fixed upstream long ago Jun 01 11:41:18 yeah, probably Jun 01 11:41:34 that ARM_INSTRUCTION_SET has been in there forever, as far as I can tell, no indication of what it was actually meant to be fixing Jun 01 11:41:56 if it works on v5 then it should be fine on v6/v7 as well. Jun 01 11:42:11 once I refreshed the recipe peeking at oe-core I noticed they got rid of it Jun 01 11:42:30 right Jun 01 11:50:21 pb_: RP asked me for a patch aimed to oe-core Jun 01 11:50:49 would you port the oe-classic packaging too? Jun 01 11:51:04 dunno, is it better? Jun 01 11:51:08 I haven't compared them. Jun 01 11:51:21 separating jffs2 and ubi mkfs's Jun 01 11:51:41 some hunderd kb Jun 01 11:52:03 and basically you choose one of them Jun 01 11:52:13 in real life Jun 01 11:52:47 right, sounds worthwhile. Jun 01 11:52:52 doing that comparison, a bigger question arises: Jun 01 11:53:00 I wonder how many people actually install mkfs (for either) on the target system, but still. Jun 01 11:53:06 would it be easy/good to have LIC_FILES_CHKSUM in oe-classic? Jun 01 11:53:25 I think it would be fairly easy to add the mechanism, but it would be a lot of work to add the data to all the recipes. Jun 01 11:53:35 no auto? Jun 01 11:53:52 I mean by scripts? Jun 01 11:53:59 you could try, but I think it would be hard Jun 01 11:54:12 that's my first impression :) Jun 01 11:54:22 that's why I asked you :p Jun 01 11:55:59 this is probably the first blocker when importing from oe-dev Jun 01 12:00:50 03Paul Menzel  07master * rb43658b888 10openembedded.git/recipes/lynx/lynx_2.8.7.bb: (log message trimmed) Jun 01 12:00:50 lynx: correct `LICENSE` from `GPL` to `GPLv2` Jun 01 12:00:50 According to `COPYING` [1] the license is GPLv2. Jun 01 12:00:50 This is a fixup for commit 06d31dc0 [2]. Jun 01 12:00:50 commit 06d31dc06beb85a28ca3c2c2c0c2787267a24ff5 Jun 01 12:00:51 Author: Stanislav Brabec Jun 01 12:00:51 Date: Tue May 31 22:28:20 2011 +0000 Jun 01 13:02:20 03Richard Purdie  07master * r6743449610 10bitbake.git/lib/bb/data_smart.py: Jun 01 13:02:20 data_smart: Adding overrides of zero length is pointless so lets not Jun 01 13:02:20 Signed-off-by: Richard Purdie Jun 01 13:09:33 asterisk has a log directory on a volatile fs Jun 01 13:09:51 so it goes away on reboot and ends up not existing Jun 01 13:11:22 Crofton: ah, fun Jun 01 13:11:27 Crofton than put it direcotry creation in a startskript Jun 01 13:11:27 yeah Jun 01 13:11:30 yeah Jun 01 13:11:39 I see not this much problems Jun 01 13:11:50 either initrd Jun 01 13:11:51 I've noticed koen did deprecate volatiles Jun 01 13:12:19 if size is not a problem...fine Jun 01 13:13:45 catullo log # du -sh /var/log/ Jun 01 13:13:47 17M /var/log/ Jun 01 13:13:50 or ups not initrd Jun 01 13:13:53 init.d skript Jun 01 13:14:00 or upstart equivaltent Jun 01 13:14:03 or systemd Jun 01 13:15:14 he marex Jun 01 13:18:51 crofton hm do you need asterisk at work? Jun 01 13:18:55 or private Jun 01 13:19:06 work Jun 01 13:20:06 pirate gsm base station for usr e100 :) Jun 01 13:20:23 I have it running, at least one phone and the echo server in asterisk :) Jun 01 13:20:26 as of last night Jun 01 13:24:23 hehe Jun 01 13:24:45 so you run zecke's, harald's and andera's soft Jun 01 13:24:56 ups andreas's Jun 01 13:24:57 Crofton: heh cool Jun 01 13:25:06 just openbts Jun 01 13:25:46 I'll bring it to fosdem and we can make an annoying app that spams sms to people as they walk by Jun 01 13:25:52 and get thrown in jail :) Jun 01 13:27:27 hi, looks like with qt4-embedded the -plugin-gfx-transformed is ignored and the resulting libraries and executables do not rotate Jun 01 13:28:25 mckoan hm? Jun 01 13:28:26 I'm searching in the net, in qt website, trying to devshell the configure but I don't understand Jun 01 13:28:52 crofton its gsm not cdma? Jun 01 13:28:59 woglinde: qt-4.7.3 Jun 01 13:29:00 yes Jun 01 13:29:36 mckoan okay so at executin time, it dont rotates the screen Jun 01 13:30:10 woglinde: yes, ./qte-example -qws Jun 01 13:30:10 Transformed: driver not found Jun 01 13:30:30 hm? Jun 01 13:30:38 whats the plugin about Jun 01 13:30:44 never heard of it Jun 01 13:31:11 how can I print out the QT_CONFIG_FLAGS ? Jun 01 13:31:40 hm cannt you look at the config logs? Jun 01 13:31:48 woglinde: It allows access to the framebuffer with rotation and things like this. Jun 01 13:32:16 hm wouldnt be directfb plugin better? Jun 01 13:32:45 Crofton: if the laws/procedures they followed to get permission for a private network at ccc are pan-European you'd presumably just be able to do the same thing Jun 01 13:32:52 at FOSDEM I mean Jun 01 13:33:21 "pirate" gsm basestation :) Jun 01 13:33:25 bluelightning at ccc camp there will be again some base stations Jun 01 13:33:41 In Europe it should be possible to get a license for testing purposes which is not that expensive. Jun 01 13:33:46 I run with no antennas Jun 01 13:33:48 florian yes Jun 01 13:33:55 so range is like ten feet :) Jun 01 13:34:02 I know some one who has a license Jun 01 13:34:39 * Crofton knows that regulations exist and is msotly careful not to be a nuisance to real users Jun 01 13:34:41 but we should start now with lte equipment haha Jun 01 13:34:48 yeah Jun 01 13:35:19 woglinde: there is no QT_CONFIG_FLAGS in the logs, strange Jun 01 13:35:22 I jsut think it would be funny to have phones switch to my bts and have it sms them an OE message :) :) :) Jun 01 13:36:15 Crofton: well with the CCC implementation they had to have new SIMs; presumably that's part of the deal Jun 01 13:36:19 crofton where are you located again? Jun 01 13:36:42 yeah, they are being serious Jun 01 13:36:48 Blacksburg, VA Jun 01 13:36:50 * woglinde see cops storming crofton's house Jun 01 13:36:54 heh Jun 01 13:37:05 do FCC inspectors carry firearms? :) Jun 01 13:37:09 anyway Graphics (plugin) ...... transformed qvfb vnc directfb Jun 01 13:37:13 apparently Jun 01 13:37:17 looks like is ok Jun 01 13:37:27 I know US postal inspectors do so it's not hard to imagine Jun 01 13:37:31 a 4.7.3 bug ? Jun 01 13:37:31 mckoan than look if the plugin is installed Jun 01 13:37:43 I think its handeld via packages-dynamic Jun 01 13:38:14 mckoan and please try out directfb Jun 01 13:38:22 and let me know if this works Jun 01 13:39:02 any of you have a qt4e-4.7.3 app and can test export QWS_DISPLAY=Transformed:Rot90 ? Jun 01 13:40:16 woglinde: linuxfb does not support rotation Jun 01 13:40:40 hm I dont talk about linuxfb Jun 01 13:40:47 talking about directfb Jun 01 13:41:39 hm I should build an qemuarm image Jun 01 13:42:54 with qte Jun 01 13:46:29 woglinde: I can't find any directfb option for QWS_DISPLAY Jun 01 13:48:25 # ./qte-example -qws -display directfb Jun 01 13:48:25 directfb: driver not found Jun 01 13:48:56 opkg list-local | grep qt Jun 01 13:49:00 but in qt-embedded.inc are set Jun 01 13:49:23 again its set handeld via packages-dynmaic Jun 01 13:49:38 so you might need to install the package Jun 01 13:49:44 same for transfrom Jun 01 13:50:07 mckoan sometimes I think you didnt understand the oe stuff Jun 01 13:50:51 woglinde: somentimes it's true, it's too much stuff :-D Jun 01 13:51:46 but packages-dynamic is long used and known feature Jun 01 13:51:52 to split up stuff Jun 01 13:52:29 woglinde: http://pastebin.com/uW2UgqzV Jun 01 13:53:26 do you see tranfsfrom or directfb qt4e plugin? Jun 01 13:53:29 I am not Jun 01 13:53:36 so please install it Jun 01 13:53:38 so may be a do_install problem Jun 01 13:53:57 opkg update Jun 01 13:54:13 opkg list | grep directfb or grep transfrom Jun 01 13:54:39 woglinde: wouldn't be better to add them in do_install? Jun 01 13:55:01 *sigh* Jun 01 13:55:16 ok no problem Jun 01 13:55:20 pleas please learn about PACKAGES_DYNAMIC Jun 01 13:55:28 and why it is used in qt-case Jun 01 13:55:36 woglinde: I guess Jun 01 13:56:02 mckoan is trying to boot a full-image, w/out package management Jun 01 13:56:17 than he need a proper image Jun 01 13:56:37 with the transfrom or directfb qt plugin installed Jun 01 13:56:59 ant_work: yes Jun 01 14:05:22  Jun 01 14:06:21 mckoan yes? Jun 01 14:06:44 woglinde: sorry but is very hard to guess how new variables work without any doc ;-) I'll try to understand Jun 01 14:07:08 woglinde: however thanks for the hint :-D Jun 01 14:07:20 khem: ping Jun 01 14:07:22 morning all Jun 01 14:07:50 hi kergoth Jun 01 14:07:57 hey pb_ Jun 01 14:08:32 I'm trying to comment #PACKAGES_DYNAMIC = "${QT_BASE_NAME}-plugin-* ${QT_BASE_NAME}-translation-* ${QT_BASE_NAME}-fonts-*" Jun 01 14:09:49 mckoan why comment? Jun 01 14:10:00 just go to your deploy dir Jun 01 14:10:11 search the ipk's Jun 01 14:10:18 and add them to your image install Jun 01 14:10:30 *sigh* Jun 01 14:10:49 woglinde: but I want to generate them in the target image without opkg Jun 01 14:11:05 guess what your image_install do? Jun 01 14:11:21 or what is it about? Jun 01 14:11:55 have look at qt4-base-image.bb Jun 01 14:12:27 qt4e-base-image.bb Jun 01 14:12:28 sorry Jun 01 14:13:54 mckoan btw. what is the image you choosed? Jun 01 14:16:00 mckoan: so you're just trying to get some extra packages installed in the image? that doesn't have anything to do with PACKAGES_DYNAMIC, you just need to add them to IMAGE_INSTALL or the equivalent in your image recipe. Jun 01 14:16:23 commenting out PACKAGES_DYNAMIC is unlikely to achieve anything except more NoSuchProvider exceptions Jun 01 14:18:52 * mckoan is drowning in the code Jun 01 14:19:22 woglinde: I'm reading qt4-base-image.bb but where is task-qt4e-qwt ? Jun 01 14:19:33 proably qt4e-demos-image should be fixed to include all plugins Jun 01 14:19:33 find didn't help Jun 01 14:19:50 mckoan than its broken and should be fixed Jun 01 14:22:39 mckoan: which bit is it that you are having trouble with? Jun 01 14:23:08 the image construction code is really not all that complicated to drive, you just give it a list of packages and it installs them for you. Jun 01 14:23:42 pb_ problem is that he didnt understand whyt PACKAGES_DYNAMIC does Jun 01 14:23:55 and how to get packages into an image Jun 01 14:24:07 well, those are two almost-completely separate things Jun 01 14:24:15 and that I am really wonder Jun 01 14:24:24 because mckoan is a long time oe user Jun 01 14:24:42 hmmm -- https://gitorious.org/~esben/oe-lite/esbens-core/blobs/parser-rewrite/lib/oelite/arch.py - I wonder if we should look into something like this for oe-core. the structure for the cpu data seems nice, and the use of config.sub/config.guess for build arch/os determination is as well Jun 01 14:24:47 PACKAGES_DYNAMIC is, admittedly, poorly documented but I don't think it is relevant to this case. Jun 01 14:25:04 http://www.lmgtfy.com/?q=what+does+packages_dynamic+do%3F will probably find you as good an explanation as any. Jun 01 14:25:13 pb they are related if you cannt start your qte app in your image and it says plugin not found Jun 01 14:25:56 that's nothing to do with PACKAGES_DYNAMIC, it just means that you don't have the right package mentioned in the image installation list. Jun 01 14:26:26 if you dont know that it is splitted up into serval packages Jun 01 14:26:33 you might search in wrong ends Jun 01 14:27:05 so technical they might not be related Jun 01 14:27:53 kergoth: yeah, the usage of config.guess/config.sub is kind of nice. Jun 01 14:28:00 I'm not sure I am so keen on the monstrous cpu table though Jun 01 14:28:25 fair enough. just a different way of structuring the info that's in our siteinfo.bbclass, afaict Jun 01 14:28:57 yeah, it seems to be a sort of mash-up of siteinfo and the conf/include/machine/* fragments Jun 01 14:29:19 now, admittedly, siteinfo and conf/include/machine/* do both suck, but it isn't totally clear to me that this is so much of an improvement Jun 01 14:31:48 also, the presence of stuff like: Jun 01 14:31:49 if guess[1] == 'pc': Jun 01 14:31:49 guess[1] = 'unknown' Jun 01 14:32:02 makes me wonder whether this mechanism is quite as robust as we might hope it to be Jun 01 14:36:56 heh, indeed Jun 01 14:38:19 in principle I think it is better to have the machine/cpu info split up, rather than in one big array. makes it easier for people to add new machines in branches/overlays without creating a merge nightmare every time. Jun 01 14:39:05 possibly the idea of having a declarative syntax in a machine config file, and then autogenerating some of the site bits from that, would be a good one though. Jun 01 14:39:30 that recent powerpc -ldl thing does illustrate that the curent site files are out of control and something needs to be done to get them back in their box. Jun 01 14:43:07 on another topic, does anybody have a good mechanism for doing nfsroot-based development with oe{-core}? something like an nfs server that shared the pseudo state is what we need, I guess. Jun 01 14:48:09 pb_: there is a helper script from an earlier version of poky that I was touching up, however it has too much machine-specific stuff in it which I'm guessing is why it was removed from poky Jun 01 14:48:53 all it does though is take some of the pain out of setting up a temporary export of the rootfs though really Jun 01 14:49:03 re Jun 01 14:49:05 damn Jun 01 14:49:08 dma stuff Jun 01 14:49:18 here at the host computer Jun 01 14:49:31 yeah, that doesn't sound like quite what I want Jun 01 14:49:33 under to much io load kernel gives up Jun 01 14:49:41 or the bus Jun 01 14:53:09 hm, bitbake's diagnostics seem to have gotten rather worse since 1.8 or so Jun 01 14:53:26 pb? Jun 01 14:53:30 bye kama Jun 01 14:53:33 ups jama Jun 01 14:53:41 if you do "bitbake -b file-that-doesn't-exist" then you get half a dozen lines of python exception spew culminating in "MultipleMatches" Jun 01 14:54:01 whereas I'm fairly sure you used to get a relatively sensible "no such file" kind of thing. Jun 01 14:54:11 * woglinde looks at kergoth Jun 01 14:54:32 pb_: something for zenlinux to add to his error rework list maybe? Jun 01 14:54:40 and I also noticed this morning that if you try to build a recipe which skipped itself then you get a crypic message about NoneType being not iterable Jun 01 14:54:48 definitely worth filing a bug Jun 01 14:55:01 which, I guess, isn't all that cryptic once you figure out what it means, but it gave me pause for thought the first time. :-} Jun 01 14:55:14 RP made a recent change that likely changed taht behavior Jun 01 14:55:26 best ask him Jun 01 14:55:33 ah right. did he make it better or worse? :-) Jun 01 14:56:04 kergoth btw. on what do you work this times? Jun 01 14:56:05 apparently -b was supposed to list the available recipes? I had no idea that was even an option. i guess he 'fixed' that capability and broke something more useful Jun 01 14:56:23 oh, I had no idea that option existed either Jun 01 14:56:32 * woglinde saw so few commits from kergoth Jun 01 14:56:34 theres a thread on one of the lists Jun 01 14:56:54 woglinde: mostly focused on internal stuff at the moment, though i still have some things i'm trying to get merged into oe-core Jun 01 14:57:04 (e.g. the devshell rework) Jun 01 14:57:31 bluelighting: oh right, yeah. I'll reply to his note on the list. Jun 01 14:57:58 not doing much major work outside of internal bits right now, hopefully i can do more in a month or so Jun 01 14:58:12 have a pandaboard now, just need to pick up a power supply and i can actually run something out of oe on it Jun 01 14:58:13 heh Jun 01 15:01:34 kergoth okay ;) Jun 01 15:01:43 yeah panda is nice Jun 01 15:01:58 but still the worse kernel/driver situation Jun 01 15:03:00 asterisk - 1.4.39.2-r0.9 Jun 01 15:03:15 ANy thoughts why there is a .9 on the end of the PR? Jun 01 15:03:50 1.4? Jun 01 15:03:58 thats old Jun 01 15:03:58 DISTRO_PR? Jun 01 15:04:11 or INC_PR, or... Jun 01 15:04:13 ah, I'll look at that Jun 01 15:04:17 not INC_PR Jun 01 15:04:26 I'm familiar with that Jun 01 15:04:48 yeah, must be distro wide Jun 01 15:04:49 thanks Jun 01 15:05:01 some one should update our recipes Jun 01 15:05:19 woglinde, I'll look at that after I get the basic image done Jun 01 15:06:01 crofton I wonder anyway if this version will work with bts stuff Jun 01 15:06:10 it does Jun 01 15:06:22 at least the echo test works Jun 01 15:09:21 hm dont you need to patch something? Jun 01 15:09:30 I didn't Jun 01 15:09:33 or works the modul compile out of tree? Jun 01 15:09:38 yes Jun 01 15:09:43 what? Jun 01 15:09:44 or is it just included Jun 01 15:10:01 I build an image with asterisk installed and openbts works well wnough to call the echo test Jun 01 15:10:30 hm intressting Jun 01 15:10:44 I am not an expert :) Jun 01 15:10:52 Is this the right place to ask a bb question? Jun 01 15:11:09 depends on the question, but you can try Jun 01 15:11:17 this is, famously, not a distro or machine support channel Jun 01 15:11:56 I want to assign a variable in a recipe, it originates in the environment, but I want to do some processing on it before assignment. How would I do that? Jun 01 15:12:08 I need to remove a prefix from an env variable before using it in the recipe. Jun 01 15:12:45 celston that smells like a false workaround Jun 01 15:13:33 Context: users on our network will typically have CVSROOT set, but it contains the :pserver: prefix, which I *don't* want in the SRC_URI cvs string. Jun 01 15:14:07 you can do something like: Jun 01 15:14:14 cvs? Jun 01 15:14:23 f = ${@transform(os.getenv("CVSROOT"))} Jun 01 15:14:26 poor users Jun 01 15:14:40 but I think you might have to add CVSROOT to BB_ENV_EXTRAWHITE to stop it getting stripped at an earlier stage Jun 01 15:15:33 all that said though, I do sort of wonder why you can't just put the whole string in the SRC_URI. is your CVSROOT so dynamic that it needs to come from the environment? Jun 01 15:15:33 Funnily enough, it seems to make it through to the recipe already, I'm not sure how or why. Jun 01 15:15:44 I know, CVS is horrific. Jun 01 15:16:26 pb_, not dynamic but it has the CVS username in, which may be different than the local username (it is for me) Jun 01 15:16:47 ah, right, fair enough Jun 01 15:17:21 anyway, yeah, if CVSROOT is making it as far as the bitbake environment then I think what I wrote ought to work. Jun 01 15:27:25 pb_ - thanks that's got me going in the right direction I think. I'm surprised that it's as complicated as it is though, it's something which would be trivial in shell/make Jun 01 15:28:30 it's no less trivial in python -- its just that you're more used to shell/make ;) Jun 01 15:28:59 * pb_ stabs bash Jun 01 15:29:04 something weird going on with my command history Jun 01 15:29:46 My python is reasonable, but it's certainly longer than: Jun 01 15:30:03 VAR = ${shell echo ${VAR1} | sed 'blah'} Jun 01 15:30:20 s/{/(/g Jun 01 15:30:38 hm. not much longer. Jun 01 15:30:43 VAR1 = "${@os.environ.get('VAR1').replace('blah', '')}" Jun 01 15:30:57 if you wanted to bypass the built in env->metadata bits to avoid having to introduce a temporary variable Jun 01 15:30:59 var = ${@re.sub(.., .., os.getenv(".."))} Jun 01 15:31:05 pb_: :) Jun 01 15:32:07 ah, but if os.getenv returns False you get an error, if it's not set in Make it just carries on (with horrific results no doubt :-) ) Jun 01 15:32:16 celston: os.environ.get('VAR1', '') Jun 01 15:32:21 now it gives you an empty string if its not set Jun 01 15:33:19 I should point out that I'm not arguing that Make is better, I know it's horrible and it's just my unfamiliarity with bb that's the cause of my problems. Jun 01 15:33:34 much of bitbake's behavior is influenced by make Jun 01 15:33:41 But it's a good way for me to explain what I'm trying to achieve Jun 01 15:33:43 lazy evaluation, := for immediate, ?= for 'if unset', etc Jun 01 15:33:58 Rather, it's not horrible itself, but it doesn't scale. Jun 01 15:34:04 s/evaluation/expansion/ would be clearer Jun 01 15:34:31 Hands up if you've ever created a monster make system that ever scared you :) Jun 01 15:34:37 Hands up if you've ever created a monster make system that scared you :) Jun 01 15:35:28 too many times Jun 01 15:35:30 I did the first buildroot that supported make menuconfig, with a perl script to autogenerate the make rules from the dependencies in the kconfig files, that was pretty disgusting Jun 01 15:35:39 this is what led to oe being started, really Jun 01 15:38:16 gosh, copying the contents of DL_DIR is a bit tedious. Jun 01 15:38:24 * pb_ watches screenfuls of eglibc svn file scroll past Jun 01 15:38:36 heh Jun 01 15:39:01 heh Jun 01 15:39:14 i'd rm -rf ${DL_DIR}/*/ first ;) Jun 01 15:39:25 yeah Jun 01 15:39:30 course, it'd take a while to repopulate if you need it, but still Jun 01 15:39:38 maybe I'll just give up on copying it, I guess the other machine can just download for itself Jun 01 15:39:52 * pb_ moving his builds to a bigger computer Jun 01 15:39:55 pb_: do you know if there was some fix to EGLIBC recently? In my older build some ld output landed in a linker script... Jun 01 15:40:09 zecke: oh, yeah, I have a patch for that lying around in my tree Jun 01 15:40:26 somewhere Jun 01 15:41:19 actually, looks like I might have deleted the repo with the patch in by mistake. but the basic idea was just to remove the "2>&1" from the linker invocations that create the .lds files in eglibc. Jun 01 15:41:25 there are two places that happens Jun 01 15:43:21 kergoth python complains that re is not defined, how would I import that in a single-line python variable expansion? Jun 01 15:43:36 or does that mean I need to def a fn? Jun 01 15:44:03 03Stanislav Brabec  07master * r60dce4afbc 10openembedded.git/recipes/lynx/lynx_2.8.7.bb: Jun 01 15:44:03 lynx: Fixed download URL of the latest release. Jun 01 15:44:03 Signed-off-by: Stanislav Brabec Jun 01 15:44:25 celston: you could def a fn, or you could do OE_IMPORTS += "re" Jun 01 15:44:52 (there's a a snippet of python that runs at the end of the recipe parsing which imports all the modules in OE_IMPORTS into the context used to execute python snippets Jun 01 15:44:53 ) Jun 01 15:45:04 sweet, thanks. Jun 01 15:45:06 ideally bitbake would handle that directly Jun 01 15:45:10 np Jun 01 15:45:23 pb_: ah, did you understand the root cause as well? e.g. it only happens when a certain lib is staged/in sysroot? Jun 01 15:45:43 zecke: it seemed to be related to my host zlib in some way, though I didn't really debug it any further than that. Jun 01 15:46:04 might be some race condition, and/or missing dependency, and/or gypsy curse effect Jun 01 15:46:22 it seems to have stopped happening for me at the moment, anyway, though I think the 2>&1 in there is still wrong. Jun 01 15:53:36 kergoth adding to OE_IMPORTS didn't work for me, but def'ing a function which did the import did. Is it a question of timing, i.e.: if I append in my recipe does the module get appended to the list before the list is imported? I'm pretty unclear on bitbake parsing versus execution order. Jun 01 15:55:08 thats why i think we need to get it into bitbake. because its done at the end of recipe parsing, there's a danger that your code could be accessed before then. and of course, you can't use :=, otherwise it evaluates immediately, long before the code runs that imports Jun 01 15:55:20 heh Jun 01 16:02:33 thank you guys, have a nice rest of the day Jun 01 16:04:41 yeah Jun 01 16:04:51 or find a better language for writing those expressions in :-} Jun 01 16:05:19 it does suck a bit that the python expression syntax is so limited. Jun 01 16:05:41 I still think lua would be ideal for these sort of oneliners Jun 01 16:05:49 * kergoth shrugs Jun 01 16:06:22 yikes, we need to audit our dependency graphs, i see Jun 01 16:06:43 yeah Jun 01 16:07:03 micro-base-image is 938 tasks, and about the first 700 of those are -native packages of one kind or another. Jun 01 16:07:15 damn Jun 01 16:07:20 right now it's at task 877 of 938 and it's just building gcc-cross Jun 01 16:07:27 that's sad Jun 01 16:07:51 it is rather Jun 01 16:08:31 now, I can see the benefit to having oe build stuff for itself: on the other side of the coin I was annoyed when I found that I had to install chrpath and help2man by hand on my new build host, and I spent a moment or two thinking "why couldn't they just be -native recipes?" Jun 01 16:08:52 but it does seem that a load of the stuff we're building at the moment is just useless detritus and doesn't actually contribute anything to the build outcome Jun 01 16:09:43 hm, what's the difference between libgcc and gcc-runtime? Jun 01 16:11:01 ah, here we go, the first actual target package commences Jun 01 16:11:07 NOTE: Running task 895 of 938 (ID: 183, /home/pb/oe/oe-core/meta/recipes-core/sysvinit/sysvinit_2.88dsf.bb, do_configure) Jun 01 16:11:09 heh Jun 01 16:26:46 I'm using btrfs. Is there any way to force all bitbake copy operations to always "cp -reflink=auto"? Jun 01 17:19:53 is there a way to get the "perl-native multilib patch for ubuntu 11.04" put into the 2011.03-maintenance branch? Jun 01 17:21:28 sure Jun 01 17:21:39 make a reuqest on the oe-dev ml Jun 01 17:22:34 woglinde: can I do that even if i had nothing to do with the patch? Jun 01 17:23:31 you should test if it works Jun 01 17:23:41 and make a line Tested-by Jun 01 17:24:08 and signed-off if you have something to alter Jun 01 17:24:22 because I dont think it works out of the box Jun 01 17:25:33 i cherry -picked 86fde1fe0281cc8c9d2bf1012d8eff5e24a69e55 and I was then able to build perl-native on my ubuntu 11.04 x86_64 machine, is that a good enough test? Jun 01 17:25:42 sure Jun 01 17:26:00 write this in the cover letter Jun 01 17:26:09 attache the patch Jun 01 17:26:15 add line Tested-by Jun 01 17:26:21 or similar Jun 01 17:26:32 dont know the excat syntax at the moment Jun 01 17:26:47 okay Jun 01 17:26:50 cherry-pick Jun 01 17:26:53 format-patch Jun 01 17:27:03 with cover-letter Jun 01 17:27:16 and Tested-BY Jun 01 17:27:25 that should be enough Jun 01 17:28:05 ok, thanks Jun 01 17:29:30 and than hope the maintaince branch maitainer applying it Jun 01 17:29:50 if they do update the status of you patch in patchwork Jun 01 17:34:51 hope? I'm not that unresponsive ;) Jun 01 17:35:42 *g* Jun 01 17:35:58 gchiii or you convice tartarus to do it now Jun 01 17:36:24 The rules are the rules Jun 01 17:36:31 post, 24h waiting period, then I apply Jun 01 17:36:54 ah okay Jun 01 17:36:55 sorry Jun 01 17:37:29 than convice tartarus to send the cherry-pick request Jun 01 17:37:32 *g* Jun 01 17:37:40 ok, I wasn't sure what the rules said about that Jun 01 17:38:27 tartarus, would you like to send the cherry-pick request? Jun 01 17:43:57 gchiii: No, I want someone else to submit it and I'll apply it after the review period Jun 01 17:44:01 or rather Jun 01 17:44:12 I want someone else to cherry pick, test, and post the cherry-pick + tested-by patch Jun 01 17:48:57 gchiii sorry, now you must do the hardwork by yourself Jun 01 17:50:33 oh well, that copy and paste work is tricky, but I will try. Jun 01 17:57:25 copy and paste? Jun 01 17:57:29 no no Jun 01 17:57:33 do cherry-pick Jun 01 17:57:39 and than format-patch Jun 01 17:57:53 configure format-patch and git sendmail right Jun 01 18:13:35 woglinde: I will set up my git sendmail stuff now Jun 01 18:21:22 kergoth: hello Jun 01 18:21:26 I am here now Jun 01 18:22:41 hi khem Jun 01 18:31:31 hi all Jun 01 18:33:42 woglinde: hello Jun 01 20:50:18 Hi everybody, unfortunately I do not know openembedded much but I am insisting on learning this time :) Jun 01 20:50:40 I have an image bb and I can creare rootfs, kernel images and bootloader Jun 01 20:51:04 I need to create a firmware from these by our own tool Jun 01 20:51:33 I read manual about IMAGE_POSTPROCESS_COMMAND Jun 01 20:52:18 I think I can use this environment for creating firmware by assigning my command to this variable Jun 01 20:52:43 but I know there is a better way Jun 01 20:53:00 I did some search and found some bb files about firmware creation Jun 01 20:53:23 such as firmware-ti-wl1251.bb Jun 01 20:53:25 what exactly do you mean by "a firmware"? Jun 01 20:53:35 no, that recipe packages up the firmware for a certain wireless chipset Jun 01 20:53:40 it has nothing to do with deploying an image Jun 01 20:53:48 uboot+kernel+rootfs > a single binary for update issues Jun 01 20:55:33 I want to find a way which will be logically suitable to openembedded Jun 01 20:55:48 do you have any suggestion about this? Jun 01 20:56:37 I just want to combining my built images Jun 01 20:56:46 with my tool Jun 01 20:57:02 ? Jun 01 20:57:17 I have a tool, fwgenerator Jun 01 20:57:40 this application add headers to beginning of kerne, rootfs and bootloader Jun 01 20:57:47 which is needed for which device? Jun 01 20:58:12 I am working on arm based board Jun 01 20:58:48 it is designed by picochip and they provide us their drivers and applications Jun 01 20:59:04 we compile all thing by openembedded and we get kernel, rootfs, bootloader Jun 01 20:59:12 sure otherwise it might have been an gpl violation Jun 01 20:59:23 after all, we use our fwgenerator tool and give these binaries to this tool Jun 01 20:59:33 they dont use u-boot? Jun 01 20:59:34 this tool creates a firmware for board upgrade issues Jun 01 20:59:39 yes, we use u-boot Jun 01 20:59:45 ah Jun 01 20:59:57 I want to do this firmware generation operation automatically in openembedded Jun 01 21:00:04 okay than you maybee need a class Jun 01 21:00:28 look at the way uboot mkimage prepares the zImage Jun 01 21:00:44 that should be easy to adapt Jun 01 21:00:54 or you might want to it at image creation Jun 01 21:01:02 than look how images are generated Jun 01 21:01:10 hm I have an image bb Jun 01 21:01:22 like pc7302-dev-image.bb Jun 01 21:01:38 there is image posthook Jun 01 21:01:39 I call this bb and it creates my kernel rootfs uboot binaries Jun 01 21:02:04 I need to call a tool to firmware creation Jun 01 21:02:21 maybee it's helping otherwise you need a new hook which kicks in after fs's are packed Jun 01 21:02:42 do you mean IMAGE_POSTPROCESS_COMMAND variable? Jun 01 21:02:46 yes Jun 01 21:03:03 but kernel and rootfs are both images Jun 01 21:03:22 so it would be run after both image creation Jun 01 21:03:25 right? Jun 01 21:03:36 there will be double run? Jun 01 21:03:41 am I right? Jun 01 21:03:47 ? Jun 01 21:04:12 but maybee IMAGE_POSTPROCESS_COMMAND is to early Jun 01 21:04:15 you mean I can add IMAGE_POSTPROCESS_COMMAND into my pc7302-dev-image.bb Jun 01 21:04:27 than you have to look at how the various fs-images are created Jun 01 21:04:38 and kick in at this level Jun 01 21:04:54 hmm Jun 01 21:05:06 but shouldnt be that hard at all Jun 01 21:05:09 half day Jun 01 21:05:11 max Jun 01 21:05:24 actually I thought I need to create a new bb like a wrapper Jun 01 21:05:53 which will include my original bb(pc7302-dev-image.bb) and I will add some command Jun 01 21:06:12 can I create a bb which calling or including another bb? Jun 01 21:06:50 as I said try IMAGE_POSTPROCESS_COMMAND first Jun 01 21:07:02 or look at the fs-create process Jun 01 21:07:13 hm ok I will try Jun 01 21:07:19 thats the cleanest way Jun 01 21:07:34 you maybee need a .bb for your firmwareimagegeneratot Jun 01 21:07:37 tor Jun 01 21:08:24 you mean I can call pc7302-dev-image.bb and then after success build I can call fwgenerator.bb manually? Jun 01 21:08:35 no!!!!!!!!!!!!!!!!!!!!!!! Jun 01 21:09:12 you need maybee fwgenerator.bb to generate the fwgenerator what ever lang it is and install it into sysroot/yourhostarm Jun 01 21:09:27 args Jun 01 21:09:29 hostarch Jun 01 21:09:36 so it can be called Jun 01 21:09:39 from path Jun 01 21:10:06 I have allready compiled it by linked statically Jun 01 21:10:47 you right, it will be good for compiling it with openembedded but it is not urgent now Jun 01 21:10:49 so write an bb others can compile it too Jun 01 21:11:04 sure you can make a fake fw.bb for now Jun 01 21:11:06 yes, you are right, I will write a bb for it too Jun 01 21:11:14 and set ASSUME_PROVIDED in local.conf Jun 01 21:12:15 I checked manual but couldnt find ASSUME_PROVIDED Jun 01 21:12:25 why is it for? Jun 01 21:14:15 re Jun 01 21:14:17 sorry Jun 01 21:14:27 < kays_> yes, you are right, I will write a bb for it too Jun 01 21:14:32 was the last Jun 01 21:14:40 in case you wrote something more Jun 01 21:14:45 oh ok Jun 01 21:14:55 and set ASSUME_PROVIDED in local.conf Jun 01 21:14:55 I checked manual but couldnt find ASSUME_PROVIDED Jun 01 21:15:03 why is it for? Jun 01 21:15:28 kays_ you cant set some native package here to use from host system Jun 01 21:15:32 qemu for instance Jun 01 21:15:41 so qemu will be not build inside oe Jun 01 21:15:48 qemu-native Jun 01 21:15:51 I meant Jun 01 21:15:52 ;) Jun 01 21:16:20 hmm Jun 01 21:24:52 It is a simple question, if anybody answer I will appreciate Jun 01 21:25:04 actually I know TOPDIR is set to where I run bitbake Jun 01 21:25:26 but even I searched too much I couldnt where it is set Jun 01 21:25:37 does anybody know whete it is set? Jun 01 21:25:54 I checked bitbake.conf, local,conf, auto.conf etc. Jun 01 21:26:04 I never see anywhere :) Jun 01 21:26:42 hi kgilmer Jun 01 21:27:05 good morning woglinde_ Jun 01 21:28:27 woglinde_, can you tell me where TOPDIR environment is set please, I am sure you know :) Jun 01 21:28:40 hm sorry Jun 01 21:28:45 why? Jun 01 21:28:49 hi ant Jun 01 21:28:59 you dodnt need to fiddle with TOPDIR Jun 01 21:29:16 when I run bitbake it create tmp directory up to TOPDIR Jun 01 21:29:39 than set the TMPDIR variable in local.conf Jun 01 21:30:04 TMPDIR = /devel/arm/oetmp-ang Jun 01 21:30:26 you are right, I thougt there is a configuration file where it is set Jun 01 21:30:46 ok thanks, but it is a mistory, I couldnt find setting place of TOPDIR :) Jun 01 21:30:58 its maybee in bitbake self Jun 01 21:31:09 check the python files Jun 01 21:32:16 you are right again :) Jun 01 21:32:33 but again dont fiddle mit TOPDIR Jun 01 21:32:37 it sets by os.getcwd() funtion Jun 01 21:32:50 if TOPDIR is not set Jun 01 21:32:56 ok, I will add to local.conf Jun 01 21:33:00 thank you Jun 01 21:52:21 03Andrea Adami  07org.openembedded.dev * rfbbe2f7dbb 10openembedded.git/recipes/kexecboot/ (kexecboot-klibc_git.bb kexecboot_git.bb): Jun 01 21:52:21 kexecboot: move to git rev 698cf7185e013e873aa7df9388a31d857727d408 Jun 01 21:52:21 * follow upstream developement Jun 01 21:52:21 * new: Enable menu item selection by keys 0-9 Jun 01 21:52:21 * debug for 'rotated qvga' case in progress Jun 01 21:52:22 Signed-off-by: Andrea Adami Jun 01 21:56:44 have a nice day, bye Jun 01 22:09:25 03Andrea Adami  07org.openembedded.dev * r5c8f30f1cb 10openembedded.git/recipes/kexecboot/kexecboot.inc: Jun 01 22:09:25 kexecboot: update list of configure args in kexecboot.inc Jun 01 22:09:25 Signed-off-by: Andrea Adami Jun 02 00:30:01 ericben: ping **** ENDING LOGGING AT Thu Jun 02 02:59:56 2011