**** BEGIN LOGGING AT Fri Jun 04 02:59:57 2010 Jun 04 06:10:01 goodmorning Jun 04 06:29:19 gm Jun 04 06:30:03 what is the preferred syntax: SRC_URI_append_xyz = ... or SRC_URI_xyz += .... ? Jun 04 06:46:49 RROR: Error in executing: /home/william/oe/bpod/openembedded/recipes/coreutils/coreutils-native_7.2.bb Jun 04 06:46:49 ERROR: Exception: Message:[Errno 2] No such file or directory Jun 04 06:57:34 when I bitbake linphone , encounter the error, but I don't know how to check it out Jun 04 07:19:27 morning Jun 04 07:19:31 hi Jun 04 07:57:01 eFfeM_work: the former. the latter will not do what you think. Jun 04 08:00:18 morning Jun 04 08:02:36 when working on converting the grep recipes to bbclassextend, i had a problem with the native grep install. executables would show up as grep.grep-native. to solve this, i created a do_install_virtclass-native function. is this the preferred way? Jun 04 08:07:36 10:06 hrw@home:gcc$ git pull Jun 04 08:07:37 Permission denied (publickey). Jun 04 08:07:57 time to catch cbrake or mickeyl or anyone other with proper admin rights... Jun 04 08:14:54 hi, Jun 04 08:15:03 janp: I beleive so, I remember doing that for other update-alternative using recipes Jun 04 08:15:12 mmm my internet connection broke while downloading xfce-something Jun 04 08:15:24 so I've got md5sum and sha256sum mismatches Jun 04 08:15:40 I keep deleting xfce-* tarballs and .md5 in sourcews Jun 04 08:15:50 and the error stays Jun 04 08:15:56 downloading it manually works Jun 04 08:16:16 the md5sum are correct when I downad manually and do md5sum ./tarball Jun 04 08:16:28 move the manual one into sources? Jun 04 08:16:29 what should I do? Jun 04 08:16:31 ok Jun 04 08:16:41 that's what I thought but I was not shure Jun 04 08:16:53 and also I prefered an automatic way Jun 04 08:17:19 thanks a lot Jun 04 08:17:22 OE doesnt cache downloads anywhere other than sources so its a bit odd what you are seeing Jun 04 08:17:29 ok Jun 04 08:17:33 I've packaged staging Jun 04 08:19:17 btw what are the best parameters for an core i7? Jun 04 08:19:24 because there are 2 variables Jun 04 08:19:28 the number of bitbake tasks Jun 04 08:19:35 and the -J N Jun 04 08:19:55 physical cores * 2 + 1 :) Jun 04 08:20:02 Id try 4 for both initially and then fiddle Jun 04 08:20:24 if HT is enabled Jun 04 08:20:35 HT seem enabled because I see 8 cores in htop Jun 04 08:20:50 yeah but HT doesnt assist OE much in my expereince Jun 04 08:20:53 if disabled better is just cores * 2 + 1 but if you have no other tasks Jun 04 08:20:54 ok Jun 04 08:21:25 HT works on the principle two tasks are using different parts of the same physical core Jun 04 08:21:53 that was freebsd suggestion for SMP Jun 04 08:22:24 ok Jun 04 08:23:02 common rule is (num_cores * 2 + 1) to do full load :) Jun 04 08:23:06 but my question was rather Jun 04 08:23:12 there are 2 variables Jun 04 08:23:15 ah Jun 04 08:23:21 I've set the -J to *2 +1 Jun 04 08:23:29 but there is also the bitbake number of tasks Jun 04 08:23:32 this is suggestion for -J Jun 04 08:23:36 indeed Jun 04 08:23:46 I wonder how the 2 variables work together Jun 04 08:23:59 basically if I set -J *2 + 1 which I have Jun 04 08:24:02 -J is only for make Jun 04 08:24:06 afaik Jun 04 08:24:20 what should I set the number of tasks to? Jun 04 08:24:23 I've 4 I think Jun 04 08:24:34 well.. Jun 04 08:25:03 if one task is compiling with -J above it will eat all CPU Jun 04 08:25:07 so Jun 04 08:25:09 indeed Jun 04 08:25:14 but in another hand Jun 04 08:25:23 if there is only one task Jun 04 08:25:30 and that there are some downloads to be done... Jun 04 08:25:42 indeed too :) Jun 04 08:26:19 btw, I like emerge and portmaster for separate downloading queue/process Jun 04 08:27:18 may be 2 tasks and -J NC*2 ? Jun 04 08:27:30 or even 2 tasks and -J NC Jun 04 08:27:40 * XorA wonders if Jay7 has 3 cores :-) Jun 04 08:27:44 ok Jun 04 08:28:21 janp yes Jun 04 08:28:40 XorA: not yet ;) waiting for newest AMD Phenom II X6 Jun 04 08:29:38 hm.. Jun 04 08:30:06 even better - NC tasks, -J NC+1 :) Jun 04 08:30:48 GNUtoo|laptop: the checksums issue you see is in latests oe.dev? Jun 04 08:31:38 GNUtoo|laptop: there was problem not with mismatch, but "missing", because bbclass got name=archive parameter later then recipes using it.. fixed few days agao.. Jun 04 08:33:09 GNUtoo|laptop: I use 2 BB_THREADS and -j5 on my quad phenom II Jun 04 08:34:06 GNUtoo|laptop: it's not optimal when both threads are in do_compile .. but long do_unpack while 2nd thread is in do_compile is not so bad Jun 04 08:35:38 Jay7: heh, I think NC*(NC+1) is probably going to be a bit excessive if NC is much more than 1 Jun 04 08:35:54 I've tried making an Eggdrip Recipe for your bitbake, you can see it here: http://pastebin.com/0JMRigrW Jun 04 08:35:54 But when I execute "bitbake -b openembedded/recipes/and so on" i get theese errors: http://pastebin.com/vGb9uE3d Jun 04 08:35:54 The log.do_install.5731 file contains this: http://pastebin.com/zMDmH8wm Jun 04 08:36:31 pb__: well.. seems you are right Jun 04 08:36:53 compiling is most time consuming part Jun 04 08:37:26 yeah, usually. the trouble is that it varies from one package to another. Jun 04 08:37:40 for some, do_configure is the slowest step, and obviously make -j doesn't help there Jun 04 08:37:46 we can collect some statistics :) Jun 04 08:38:04 I suspect you are probably best off setting BB_N_T to some large number, and P_M to a small number, maybe 2 or even 1. Jun 04 08:38:27 good idea Jun 04 08:38:49 but there can be special tuning for hyperthreading Jun 04 08:39:17 mindThomas kergoth and I told you yesterday what is wrong Jun 04 08:39:22 2x make on one core is ok Jun 04 08:40:01 but 2xMake on 4xCores can eat all disk i/o :) Jun 04 08:41:52 woglinde: the S parameter? Jun 04 08:42:08 I've tried setting that too Jun 04 08:43:39 mindthomas go do to your tmpdir/work/armv6or5or/eggdrop-versionnumber Jun 04 08:43:55 and look what is there Jun 04 08:45:21 pb_ was afk, thanks for your answer Jun 04 08:49:33 which tmp dir are you talking about Jun 04 08:49:44 the linux one - in there, there is no work folder Jun 04 08:50:17 one more Q: glibc recipes (e.g. 10.1) refer to generic-bits_select.h this file only exists in glibc-2.4 subdir still this file is taken, is it ok if a recipe for version X takes a file from the dir from version Y ? Jun 04 08:51:18 mindThomas read local.conf and manual again Jun 04 08:51:40 yes Jun 04 08:51:43 and under that i have Jun 04 08:52:25 in /tmpdir/work/armv4t-angstrom-linux-gnueabi/eggdrop1.6.19-1.0-r0/ i have 3 folders, eggdrop, image and temp Jun 04 08:59:58 try S = ${WORKDIR}/eggdrop Jun 04 09:00:06 just change two things Jun 04 09:00:33 added PR = "" and changed S = "${WORKDIR}/eggdrop1.6.19" which is where it extracts Jun 04 09:00:39 now I get other automake errors Jun 04 09:00:44 mindThomas you found the right dir Jun 04 09:00:47 ups Jun 04 09:01:08 set PR = "r1" Jun 04 09:01:12 http://pastebin.com/Fb18Ba4a Jun 04 09:01:22 look at those errors Jun 04 09:01:23 you said the dir is eggdrop Jun 04 09:01:41 03Roger Monk  07org.openembedded.dev * rf09b40abaf 10openembedded.git/recipes/ti/ti-codec-engine_2.25.03.13.bb: Jun 04 09:01:41 ti-codec-engine: Add v2.25.03.13 Jun 04 09:01:41 Signed-off-by: Roger Monk Jun 04 09:01:41 Signed-off-by: Koen Kooi Jun 04 09:01:48 kergoth, ping.. I have a question about bitbake's Data.setData(): http://paste.debian.net/76085/ Jun 04 09:01:51 03Roger Monk  07org.openembedded.dev * rbef7a76237 10openembedded.git/recipes/ti/ti-xdctools_3.20.00.41.bb: Jun 04 09:01:51 ti-xdctools: Add v.3.20 Jun 04 09:01:51 Signed-off-by: Roger Monk Jun 04 09:01:51 Signed-off-by: Koen Kooi Jun 04 09:02:15 mindthomas hm Jun 04 09:02:24 seems some autotools bug Jun 04 09:03:06 woglinde: I am not going to cambridge next week Jun 04 09:03:59 hrw okay Jun 04 09:04:02 thought so Jun 04 09:04:35 will you come to linuxtag? Jun 04 09:04:44 woglinde: I am not part of Linaro/Infrastructure team but Linaro/Foundations Jun 04 09:04:53 hrw ah okay Jun 04 09:05:08 woglinde: thinking of it. need to check exactly which tracks has sense to visit Jun 04 09:05:23 woglinde: as I prefer to not go for whole event due to cash and time Jun 04 09:05:30 hihi I thought you will help out on the booth *g* Jun 04 09:05:38 just a joke Jun 04 09:05:42 hrw: hm Jun 04 09:06:00 hrw: I wish to present kexecboot to Linaro :) Jun 04 09:06:10 can I consult with you? :) Jun 04 09:06:15 jay7 *g* Jun 04 09:06:20 Jay7: mail linaro-dev ML Jun 04 09:06:27 hrw: 10x Jun 04 09:07:22 the extract dir is eggdrop1.6.19 Jun 04 09:07:36 but there is no -r1 or something like that! Jun 04 09:10:28 *sigh* Jun 04 09:10:39 the -r1 is for topdir Jun 04 09:10:47 which is WORKDIR Jun 04 09:10:59 and for the resulting PACKAGE Jun 04 09:11:10 it has nothing to do with the original sourceüpackage Jun 04 09:12:23 but it doesn't matter Jun 04 09:12:57 it gives me same error if I add PR = "r1" Jun 04 09:13:23 PR = "r1" is only for oe sanity Jun 04 09:13:28 sure Jun 04 09:13:32 Jul. 19 - 23, 2010 Linaro Engineering Summit (Prague, CZ) Jun 04 09:13:33 eFfeM_work: yeah, that's basically ok. the alternative would be to have loads of copies of the one file, which is not exactly ideal either. Jun 04 09:13:36 that seems interesting :) Jun 04 09:14:33 but then what is the problem with automake Jun 04 09:14:54 mindthomas maybee its to old Jun 04 09:15:03 never saw the error with version 1.10.2 Jun 04 09:15:21 is that an oe or a package problem? Jun 04 09:15:24 maybee you have to add inherit gettext Jun 04 09:15:39 inherit autotools gettext Jun 04 09:15:55 mindthomas its an autotools problem combined with oe Jun 04 09:15:55 the problem is the option it adds: /usr/bin/autoreconf: unrecognized option '--exclude=autopoint' Jun 04 09:16:35 try what I wrote Jun 04 09:17:19 same error Jun 04 09:18:12 pb_ would have expected the file to be in the files dir in that case Jun 04 09:18:37 pb_ i thought/expected recipes to look into the subdir with the same name and in files/ Jun 04 09:18:50 not in other versions' subdir Jun 04 09:19:02 mindthomas sorry than I am out of knowledge Jun 04 09:20:39 can it be something with the parameters I start with to set up the bitbake enviroment? http://pastebin.com/ZyN13VHz Jun 04 09:22:37 hm looks okay Jun 04 09:24:38 hmm ok :( Jun 04 09:25:04 Would you try making an ipkg (bitbake it) for Eggdrop yourself, on your computer and then see if you can? Jun 04 09:25:42 http://www.eggheads.org/ it's their homepage - I would be so glad if I could get that running on my device, as that would make it a great server! Jun 04 09:26:40 there's irssi, who needs eggdrop ;) Jun 04 09:26:54 eggdrop is a bot Jun 04 09:26:59 irssi is a client Jun 04 09:27:27 yes exactly Jun 04 09:29:19 but would you try that woglinde? Jun 04 09:29:31 and do me that favour? Jun 04 09:30:35 thanks :) Jun 04 09:31:37 later Jun 04 09:32:41 woglinde: I know, but irssi is also scriptable. Jun 04 09:32:55 sure Jun 04 09:33:08 but its not a classic bot as eggdrop Jun 04 09:35:45 eggdrop was great at own time Jun 04 09:36:14 hrw: cbrake is here :) Jun 04 09:38:11 cbrake: hi Jun 04 09:38:31 11:38 hrw@home:openembedded$ git pull Jun 04 09:38:31 Permission denied (publickey). Jun 04 09:38:43 [remote "origin"] url = git@git.openembedded.org:openembedded Jun 04 09:39:00 cbrake: ideas? ssh key got revoked? Jun 04 09:40:51 yeah Jun 04 09:41:03 icedtea-1.8 builds Jun 04 09:46:56 hm uh Jun 04 09:47:16 seems I cannt override autotools.bbclass do_install Jun 04 10:04:59 hm, it seems there is an issue with pinning recipes with BBCLASSEXTEND="naitve": see http://www.pastebin.ca/1877139 Jun 04 10:05:53 I've pinned gnu-config, it tells me the version is not avialable , then it starts to build it? Jun 04 10:06:09 btw the log is from a version which tries to pin both gnu-config and gnu-config-native Jun 04 10:06:20 Hi i wish to rebuild ti-codecs-omap3530-server ti-cmem-module ti-dsplink-module ti-lpm-module ti-sdma-module packages for latest kernel Jun 04 10:06:22 eFfeM_work: the 'files' directory is appropriate if the file applies to all versions, but there are some cases where it applies only to a subset Jun 04 10:06:56 is there any command in bitbake -c rebuild ? for performing this operation Jun 04 10:06:58 ananth_ do bitbake -c clean ti-codecs-omap3530-server ti-cmem-module ti-dsplink-module ti-lpm-module ti-sdma-module Jun 04 10:07:16 pb_ my idea was that files was for multiple versions, i think lots of packages have things in files that apply only to a subset and that are not used by others Jun 04 10:07:39 yes, there are some packages that do that. Jun 04 10:08:05 but, consider the case where version 1.0 and 1.1 require one variant of a file, and version 1.2 and 1.3 require a different variant of the same file. Jun 04 10:08:56 pb_ yeah thought of that, that is kind of a pathetic case in which copying would be needed (imho) Jun 04 10:09:06 Thanks woglinde Jun 04 10:09:29 ananth_ did you build psp-2.6.32 already? Jun 04 10:10:55 No iam working in omap overo board Jun 04 10:11:49 ERROR: Nothing PROVIDES 'ti-cmem-module' and ERROR: Nothing PROVIDES 'ti-sdma-module' Any ideas ? Jun 04 10:12:35 eFfeM_work, ISTR that gnuconfig switched to git, didn't it? http://git.savannah.gnu.org/cgit/config.git Jun 04 10:13:14 I am not sure overo has ti-sdma-module Jun 04 10:13:25 or ti-cmem-module Jun 04 10:14:15 ok Jun 04 10:16:26 anatah try simple bitbake gstreamer-ti Jun 04 10:16:32 which build all i stuff Jun 04 10:16:36 except sgx Jun 04 10:58:49 How to rebuild using bitbake ti-codecs-omap3530-server ti-cmem-module ti-dsplink-module ti-lpm-module ti-sdma-module following modules for 2.6.33 kernel Jun 04 10:59:25 I tried bitbake -c clean and built again Jun 04 10:59:51 but nothing works .... Any idea to build those modules again ? Jun 04 11:03:14 Ananth_: aren't some of them provided by ti-linuxutils recipes? Jun 04 11:04:43 Those information are available in recipes/ti/gstreamer-ti_svn.bb Jun 04 11:05:49 And .bb files are avilable in recipes/ti/ Jun 04 11:06:10 Will bitbake specifiy .bb clean and build will work ? Jun 04 11:06:25 for those modules ? Jun 04 11:15:18 hi zecke Jun 04 11:15:20 florian: any chanche to rescue http://projects.linuxtogo ? Jun 04 11:15:44 ant_work: yes sure, I jus didn't have time for it so far :-( Jun 04 11:15:48 gm Jun 04 11:24:41 hi Jun 04 11:39:48 hrw: I'm not sure what is going on -- I'll check Jun 04 12:03:29 hrw: git clone git@git.openembedded.org:openembedded starts for me Jun 04 12:04:00 14:04 hrw@home:tmp$ git clone git@git.openembedded.org:openembedded Jun 04 12:04:00 Initialized empty Git repository in /tmp/openembedded/.git/ Jun 04 12:04:00 Permission denied (publickey). Jun 04 12:04:31 cbrake: check is my key RSA or not... Jun 04 12:09:12 jo re zecke Jun 04 12:09:42 woglinde: hey... wow... I have had so much private stuff lately... Jun 04 12:11:50 zecke fine Jun 04 12:18:23 one thing which I need to remember when I work on ubuntu packages - edit configure/makefile.in not configure.ac/makefile.am like it was with OpenEmbedded Jun 04 12:18:27 fun Jun 04 12:19:46 hehe... yeah, the patch will be bigger... and it is good when paid by LOC. Jun 04 12:20:19 zecke: no, debian and ubuntu do not run whole autoconf stuff Jun 04 12:21:34 hrw: I know, I remember their patchset's for eabi... needing to patch config.guess in every single package. Jun 04 12:21:48 brb Jun 04 12:22:22 hi Jun 04 12:24:54 kergoth, ..fyi: i've now sent the above Data.setData() suggestion to the list as [PATCH 6/8] cache: use a set() for __depends Jun 04 12:26:12 kergoth, it does not speedup anything measurable, but IMHO it's more intuitive to use a set there Jun 04 12:27:59 hrw????? Jun 04 12:28:08 hrw wrong Jun 04 12:28:16 hrw depends on the package-maintainer Jun 04 12:28:27 woglinde: yes, I know and yes, my patches touch both files Jun 04 12:28:38 ugly Jun 04 12:28:50 and you can always make it the right way Jun 04 12:29:03 I hacked enough on .deb's Jun 04 12:29:04 I want to minimum size of elightment image Jun 04 12:29:15 anand make your own image Jun 04 12:30:12 can i comment unneeded application in recipes/images/omap3-desktop-image.bb Jun 04 12:30:18 will this work ? Jun 04 12:30:24 anand sure Jun 04 12:30:53 you can comment on the task.bb too Jun 04 12:31:14 where this task.bb be available Jun 04 12:31:16 ? Jun 04 12:31:53 recipes/tasks Jun 04 12:44:38 03Dmitry Eremin-Solenikov  07org.openembedded.dev * rc65ee13298 10openembedded.git/recipes/pcmciautils/pcmciautils.inc: Jun 04 12:44:38 pcmciautils: fix build to use flex from sysroots, not from host Jun 04 12:44:38 Signed-off-by: Dmitry Eremin-Solenikov Jun 04 12:44:49 03Vladimir Zapolskiy  07org.openembedded.dev * r52b03b11c0 10openembedded.git/recipes/xcb/ (libxcb/configure_git.patch libxcb_git.bb): Jun 04 12:44:49 libxcb: fix paths for installed python modules Jun 04 12:44:49 This fixes a compilation problem for recent libxcb versions, providing Jun 04 12:44:49 correct paths to the installed python modules. Jun 04 12:44:50 Signed-off-by: Vladimir Zapolskiy Jun 04 12:49:07 what's the state of the bitbake 1.10 branch (there were no commits for 7 weeks)? Is it superseeded by master? I am missing the '-i' option there (which can be fixed by uncommenting 2 lines) Jun 04 12:50:57 03Dmitry Eremin-Solenikov  07org.openembedded.dev * r3db17146cb 10openembedded.git/classes/icecc.bbclass: (log message trimmed) Jun 04 12:50:57 icecc.bbclass: fix irt changes in toolchains installation Jun 04 12:50:57 The installation of toolchains in /cross/ has changed a bit, resulting Jun 04 12:50:57 in failures of icecc.bbclass to find necessary files (libstdc++.so, g++, etc.) Jun 04 12:50:57 Fix that by pointing the class file to libstdc++ in staging and corrent name Jun 04 12:53:33 Dmitry Eremin-Solenikov are you here? Jun 04 12:58:33 03Phil Blundell  07org.openembedded.dev * r04d50d5a04 10openembedded.git/recipes/images/gtk-demo-image.bb: gtk-demo-image: new Jun 04 12:58:44 03Phil Blundell  07org.openembedded.dev * rc235146da8 10openembedded.git/recipes/linux-atm/linux-atm_2.5.1.bb: linux-atm: update to 2.5.1 Jun 04 12:58:54 uh uh commits from phil Jun 04 12:59:40 pb_: how to geta @gnu.org address? how major must a contribution be? Jun 04 13:01:00 zecke you are asking it every year Jun 04 13:02:15 yeah Jun 04 13:02:22 zecek libc arm thread stuff Jun 04 13:02:25 i don't have secondary storage. :) Jun 04 13:02:52 so the page with the information just gets thrown away. :) Jun 04 13:03:26 03Phil Blundell  07org.openembedded.dev * r1d84ab01cc 10openembedded.git/recipes/thttpd/ (files/htpasswd_getline.diff thttpd_2.25b.bb): thttpd: fix a compilation failure Jun 04 13:03:59 zecke: I don't really remember. I've had it for about fifteen years and I don't quite recall how it came about in the first place. Jun 04 13:05:08 I have some mickey mouse patches in glibc, gdb and GNU Smalltalk now. :) Jun 04 13:08:24 very good :-) Jun 04 13:09:15 zecke make your own project and get it hosted and approved by gnu Jun 04 13:09:21 03Phil Blundell  07org.openembedded.dev * r632493caa1 10openembedded.git/recipes/gtkextra/ (gtk+extra-2.1.2/auto-lossage.patch gtk+extra_2.1.2.bb): gtk+extra: update to 2.1.2, apply lame patch for autotools issuette Jun 04 13:09:22 this may give you a gnu adress Jun 04 13:09:37 pb fixing bugtracker stuff? Jun 04 13:09:43 GNU zecke Jun 04 13:10:25 woglinde: no, just clearing out some old patches that I had in another tree Jun 04 13:11:46 ah Jun 04 13:17:55 zecke, :) Jun 04 13:18:25 jo stefan Jun 04 13:18:38 stefan I found a micro-sd in the bug 1 Jun 04 13:18:48 but didnt flashed it yet Jun 04 13:18:54 woglinde: hi, good! Jun 04 13:18:57 woglinde: ok Jun 04 13:19:22 iced-tea 1.8 builds now too Jun 04 13:19:38 woglinde: At some point I would like to coordinate with you and Robert if we have areas where we conflict or need to take care of each other Jun 04 13:19:46 woglinde: ah, cool Jun 04 13:20:02 woglinde: btw, what are the tasks/images you are testing with Jun 04 13:20:05 stefan sure Jun 04 13:20:10 woglinde: Should use the same here :) Jun 04 13:20:27 I will commit some stuff on monday Jun 04 13:20:40 woglinde: ok, great Jun 04 13:20:42 jalimo-trunk Jun 04 13:20:48 for icedtea Jun 04 13:20:54 but we will integrate it into oe Jun 04 13:21:05 because of the staging and patchstage changes Jun 04 13:21:24 we need to branch it for stabel oe and dev oe Jun 04 13:22:14 woglinde: I think I will start doing build tests when it hits any branch in OE. Jun 04 13:22:30 hm Jun 04 13:22:44 zecke, perhaps you could apply my bb typo fixes (and approve the rest that you like) :) Jun 04 13:23:10 woglinde: Or do you prefer another way? Jun 04 13:23:33 woglinde: I still need to understand your plan before I can see where to hook in :D Jun 04 13:25:30 blindvt: most of it looks fine, I don't know setdefault and need to learn about this first for the patch Jun 04 13:25:49 setdefault? Jun 04 13:25:55 blindvt: I will try to do it tomorrow, it is already 21:25 here in Taipei Jun 04 13:26:03 woglinde: some method on the dict or such. :) Jun 04 13:26:15 hi zecke Jun 04 13:26:33 woglinde: so far I knew two idioms for stuff inside dicts... just add it and handle exceptions when it was not there Jun 04 13:26:50 hm defaultdict? Jun 04 13:26:51 woglinde: or check if it is not there, then create an empty entry, then append... apparently there is a better way Jun 04 13:27:14 zecke yes defaultdict Jun 04 13:27:38 http://docs.python.org/release/2.5.2/lib/defaultdict-examples.html Jun 04 13:28:02 zecke, k, TIA. dict().setdefault("foo",[1,2,3]) returns dict()["foo"] or enters [1,2,3] if there is no dict()["foo"]. dict().setdefault(1) defaults to None, i.e. if dict()[1]: return dict()[1] else dict()[1]=None Jun 04 13:29:36 zecke, but i don't speak python fluently -- in contrast, i'm pretty python illiterate, just scribble what i need -- , so better refer to the docs ;) Jun 04 13:31:52 zecke, put short, that setdefault hunk is just what _i_ think is more python'ish for exactly the same thing that was there before. The former code just looks ugly, consider it a (n unrelated) style-cleanup only Jun 04 13:35:11 blindvt: well, it is nice to learn new things. Jun 04 13:35:26 blindvt: max is a builtin? do you have any number for this change by itself? Jun 04 13:37:21 zecke, the change was not measurable but is easier to read and feels more appropriate. Style-thing again, and probably a tad faster, too Jun 04 13:38:12 blindvt: well, peak memory will be likely worse due the extra list allocation Jun 04 13:43:44 zecke, given that this spot deals with dependencies and the list of dependencies is usually relatively small (just a couple of dozen), i doubt that it costs noteworthy amounts of memory, but i'm of course fine with dropping that hunk. Jun 04 13:45:10 blindvt: well, I only have one standard for code. :) Jun 04 13:45:54 blindvt: you could use the func= argument of max to use the first item of the pair... :) Jun 04 13:45:55 zecke, well, i do have one standard per language, but yea :) Jun 04 13:46:31 yeah... applying "will this slow down page loading" does not really apply to anything but render engines. :) Jun 04 13:46:40 i think i have more than one standard too Jun 04 13:47:00 zecke, ah, that would be even better, sure. But IIRC the mtime is the second item of the pair, isn't it. The first one is the filename Jun 04 13:47:19 blindvt: do you happen to have a git tree for your changes? Jun 04 13:47:28 Do someone have any glue about this build failure (it is when we build gs) http://paste.debian.net/76115/ ? Jun 04 13:47:31 blindvt: max... no idea, i think you can select which item is supposed to be compared. Jun 04 13:47:59 zecke, only a local one, sorry. I think /me as read-only user cannot have a branch in the repo, can i? Jun 04 13:48:29 blindvt: you could send me your key... but that is for the next time Jun 04 13:49:47 zecke, a personal branch would have been handy for several parts, but i don't know how you do them Jun 04 13:50:47 does do_install_append_virtclass-native() works? Jun 04 13:52:23 blindvt: you can send me your ssh key, I will give you access to bitbake.. then you should commit to a branch of yours... alternatively use gitorious, repo.or.cz Jun 04 13:52:36 blindvt: I have pushed the typo/spelling fixes now Jun 04 13:53:02 blindvt: the os.stat -> os.path.isfile change is cosmetic or a perf change? Jun 04 13:53:13 zecke, a perf change Jun 04 13:53:38 blindvt: due avoiding exceptions or because python caches? Jun 04 13:53:50 zecke, 4, 7, 8 make a noteworthy difference Jun 04 13:54:25 blindvt: for 5 i would like you to look into not creating a tmp list. :) Jun 04 13:54:27 s/4// Jun 04 13:55:40 blindvt: for the setdefault I need to remember what our minimum python version is... i think it is 2.5 Jun 04 13:55:51 and now my wife calls... I will continue tomorrow morning Jun 04 13:55:58 4 was trying to cut down on stat(). I fear i've deleted my comparison straces of 4 earlier today Jun 04 13:56:03 in any case, thanks a lot for taking your time and doing these changes! Jun 04 13:56:43 OT: did you ever stumble across a LRU cache in stock python Jun 04 13:57:01 zecke, setdefault is in 2.5 (and already used in the same file, either setData() or getData, i.e. the other) Jun 04 13:57:42 zecke, re LRU cache, not that i'm aware of, no Jun 04 14:00:13 03Antonio Ospite  07org.openembedded.dev * r66aa4ed941 10openembedded.git/files/device_table-minimal.txt: Jun 04 14:00:13 device_table-minimal.txt: fix input device nodes path Jun 04 14:00:13 Input event device nodes are usually under /dev/input. Jun 04 14:00:13 Signed-off-by: Antonio Ospite Jun 04 14:00:13 Acked-by: Koen Kooi Jun 04 14:00:19 03Antonio Ospite  07org.openembedded.dev * rb4cae15439 10openembedded.git/files/device_table-ezx.txt: Jun 04 14:00:19 device_table-ezx.txt: add needed device nodes Jun 04 14:00:19 * With all the mux nodes we can start playing again with gps on A780 Jun 04 14:00:19 * More fb nodes are needed for pxa fb overlay access Jun 04 14:00:20 * Audio devices are needed too in the case of static /dev Jun 04 14:00:37 03Ilya Petrov  07org.openembedded.dev * r2db8511584 10openembedded.git/conf/machine/include/motorola-ezx-base.inc: Jun 04 14:00:37 motorola-ezx-base.inc: Add support for ubifs Jun 04 14:00:38 Signed-off-by: Ilya Petrov Jun 04 14:00:38 Signed-off-by: Antonio Ospite Jun 04 14:00:39 Acked-by: Koen Kooi Jun 04 14:00:39 03Antonio Ospite  07org.openembedded.dev * rf6a2eb34e8 10openembedded.git/recipes/xserver-kdrive-common/xserver-kdrive-common/Xserver: Jun 04 14:00:43 xserver-kdrive-common/Xserver: fix EZX settings Jun 04 14:00:43 Support all EZX machines, and set up the mouse as a tslib device. Jun 04 14:00:43 Signed-off-by: Antonio Ospite Jun 04 14:00:43 Acked-by: Koen Kooi Jun 04 14:00:44 03Antonio Ospite  07org.openembedded.dev * rb1854052c2 10openembedded.git/recipes/tslib/tslib/a780/tslib.sh: Jun 04 14:01:23 * add ezx-gen-blob to EXTRA_IMAGEDEPENDS Jun 04 14:01:24 * use MODUTILS=26 Jun 04 14:01:24 (30 lines omitted) Jun 04 14:05:53 blindvt: we have no usernames. Jun 04 14:06:00 hm master bitbake is slower than 1.8.18 Jun 04 14:06:03 zecke, ok then Jun 04 14:06:58 zecke, perhaps you could use a deque as LRU cache -- FIFO? dunno Jun 04 14:08:04 woglinde: how so? everyone i've talked to has been amazed by the improvement in speed, particularly the runqueue generation Jun 04 14:08:17 * kergoth isn't particularly awake yet, ugh, woke up with a headache, hate those days Jun 04 14:08:22 maybee I have false target Jun 04 14:08:35 hi kergoth btw, Jun 04 14:09:07 or shoud I test .10 trunk? Jun 04 14:09:10 * kergoth 's inbox is full of threads he needs to read, falling behind Jun 04 14:09:38 zecke, but i vaguely remember that there was a LRU in python, but i don't remember if it was shipped per default or self-written. been years Jun 04 14:10:41 woglinde, for which working-set is master slower than the old release? Jun 04 14:10:46 kergoth: could you reply to the other part of the mail as well? Jun 04 14:12:00 there's nothing to be said. The way you said things pissed me off. I don't expect you intended that, and I left before I got more pissed off or said anything regrettable. Jun 04 14:23:45 morning kergoth Jun 04 14:24:32 blindvt: thought about using a defaultdict for depends_cache rather than setdefault calls? all the patches look like definite improvements, nicely done. In a couple cases it uses slightly more memory, but likely nothing noticeable, and its worth it to make the code more pythonic Jun 04 14:26:23 * kergoth grumbles and goes to get tylenol and caffeine Jun 04 14:29:52 sigh Jun 04 14:30:04 still the BBCLASSEXTEND bug Jun 04 14:31:35 Unknown Event: Jun 04 14:31:39 hm intressting Jun 04 14:34:37 woglinde: did you manage to make python emit pointer errors like C :-D Jun 04 14:36:13 I only checkout master Jun 04 14:36:31 and run bitbake packagewhichdontexist Jun 04 14:49:55 03Koen Kooi  07org.openembedded.dev * rc3d41d503c 10openembedded.git/recipes/u-boot/ (2 files in 2 dirs): u-boot git: update beagleboard patches Jun 04 14:49:56 03Koen Kooi  07org.openembedded.dev * r1b3d1b48d2 10openembedded.git/recipes/gstd/gstd_git.bb: gstd: also bump PR Jun 04 14:52:29 hi Jun 04 14:52:35 I'm getting: ERROR: Exception: Message:[Errno 2] No such file or directory Jun 04 14:53:04 in do_populate_sysroot of coreutils-native and I don't get any further Jun 04 14:53:21 I can't figure out what file is it it's trying to access Jun 04 14:53:33 but my guess is - chrpath; would that be plausible? Jun 04 14:54:14 I have chrpath, but only in $HOME/bin, which is in $PATH Jun 04 14:54:53 would it be possible bitbake/OE changes PATH? Jun 04 14:55:11 the chrpath sanity check passes, btw. Jun 04 14:55:27 bitbake -e | grep PATH= Jun 04 14:55:33 oh, thanks Jun 04 14:55:50 -e dumps the metadata. global if you don't specify a recipe, specific if you specify a recipe Jun 04 14:55:57 always good for verifying things are the way you think they are :) Jun 04 14:58:34 whoa, it produces quite some output Jun 04 15:04:28 there's a lot of metadata :) Jun 04 15:05:31 anyway, it was chrpath, if I hardcode the parh into relocatable.bbclass, then it builds Jun 04 15:05:34 * zub investigates further Jun 04 15:08:10 odd. oe doesn't kill the path, it just adds to it Jun 04 15:09:08 hmm Jun 04 15:09:33 may be I ran into same issue and ended up clearing it myself in little env setup script Jun 04 15:11:07 * kergoth really likes using bitbake master with bblayers, adios setup.sh Jun 04 15:11:33 heh I use bblayers too Jun 04 15:12:01 <3 Jun 04 15:13:01 my setup.sh is mainly DISTRO= ad MACHINE= Jun 04 15:13:42 ahh, right Jun 04 15:14:52 http://www.vim.org/scripts/script.php?script_id=2441 Jun 04 15:15:03 maybe useful for our Python people Jun 04 15:15:25 nice Jun 04 15:15:43 i've been using a compiler plugin that runs a python compile to check syntax, or a pylint one Jun 04 15:15:47 some parts from https://dev.launchpad.net/UltimateVimPythonSetup maybe too Jun 04 15:15:54 (the pylint one actually runs every time you save the file, which is kind of cool) Jun 04 15:15:59 but that pyflakes one looks nice Jun 04 15:16:20 * kergoth grabs Jun 04 15:35:49 if I wrap the chrpath binary by a shell script that logs the PATH, I see: PATH='/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin' Jun 04 15:36:10 which is considerably shorther than the PATH bitbake was started with Jun 04 15:39:52 I just synced my OE repo with upstream and I see an error in opkg-nogpg-nocurl_svn saying "Unknown fetch error: too many values to unpack" Jun 04 15:39:56 Has anyone seen this? Jun 04 15:45:33 what's up woglined? Jun 04 15:45:36 *woglinde Jun 04 15:45:52 what does: ERROR: 'STAMP' while parsing ... mean? Jun 04 15:46:37 huh, I just executed bitbake in wrong dir :) Jun 04 15:58:03 zub: might want to set TOPDIR in local.conf pointing to the full path, rahter than letting it be set to the dir you run bitbake in Jun 04 15:58:11 end up with fewer rogue 'tmp' dirs that way too ;) Jun 04 15:58:36 as for the chrpath path - adding CHRPATH_BIN = "/path/to/chrpath" into local.conf works around the issue too Jun 04 15:58:47 I'm unable to find where/how is the PATH changed... Jun 04 15:58:55 weird Jun 04 15:59:19 zub: that's bitbake... it clears the complete environment Jun 04 15:59:28 kergoth: when TOPDIR is not set, then it defaults to .? Jun 04 15:59:34 no, it clears everything but specific variables Jun 04 15:59:38 of which PATH is one Jun 04 15:59:41 zub: yep Jun 04 16:00:16 well, the PATH must have survived to the sanity check, as that passed; only when (later) chrpath is called from relocatable.bbclass it fails Jun 04 16:03:06 zub: yes, this environment cleanup is unpredictable :( That's why I applied http://ensc.de/bitbake/0002-env.patch and call bitbake through a wrapper script which cleans up the env Jun 04 16:05:09 with that patch the sanity check would fail right away I guess; that sounds like better behaviour than the mysterios breakage Jun 04 16:05:56 blah, I saw + where there was - Jun 04 16:05:57 :) Jun 04 16:06:01 kergoth: Would you do me a favour, try to make a recipe for Eggdrop and bitbake on your computer and see if you can get it working, cause I can't? http://www.eggheads.org/ Jun 04 16:06:25 ensc|w: when is serve called? Jun 04 16:09:16 zub: I do not know this exactly... It is called atleast for the fetch methods (which is the reason why I use the patch) Jun 04 16:10:06 ok, I'll stick to explicit chrpath path, but keep wary of path issues Jun 04 16:10:44 btw. is bitbake 1.10 or even git master prefered over 1.8? Jun 04 16:11:07 wiki recommends 1.8, SHR uses 1.10 Jun 04 16:13:25 1.10 isn't technically released yet, but it should be, it should be at that level Jun 04 16:13:36 i'd suggest everyone use it, if for no other reason than to find any remaining bugs :) Jun 04 16:15:42 hm, I'm seeing quite often fatal: The remote end hung up unexpectedly, on different boxes, with different servers Jun 04 16:16:05 I mean: when do git clone/pull Jun 04 16:16:32 kergoth: I know a bug: '-i' is not supported Jun 04 16:16:44 what do you mean? Jun 04 16:16:55 bitbake -i does not work with bitbake 1.10 Jun 04 16:17:07 but that's trivial to fix by uncommenting two lines Jun 04 16:17:07 "does not work" isn't a useful bug report :) Jun 04 16:17:18 I bumbed into it too Jun 04 16:17:44 kergoth: but the best description for the current behavior Jun 04 16:17:47 $ bitbake -i Jun 04 16:17:48 ERROR: Nothing to do. Use 'bitbake world' to build everything, or run 'bitbake --help' for usage information. Jun 04 16:20:26 zub: https://www.cvg.de/people/ensc/0013-interactive.patch Jun 04 16:20:43 Is it possible to control IO ports with Apache and a PHP script on an ARM Device? Jun 04 16:20:53 ensc|w: thx Jun 04 16:25:40 it then dies on a simple parse for me: http://linux.fjfi.cvut.cz/~zub/bitbake-i-parse.log Jun 04 16:27:34 ok... I used only 'getvar' ;) Jun 04 16:33:54 argh Jun 04 16:34:11 monkeypatching bb.data.getVar is proving difficult, because bitbake is swallowing the fucking exception somewhere Jun 04 16:34:14 * kergoth grumbles Jun 04 16:34:21 hm yes -i is broken on master Jun 04 16:39:42 hi woglinde, you've tried making a recipe? Jun 04 16:40:13 mindthomas sorry not yet Jun 04 16:40:22 hmm ok then :( Jun 04 16:40:25 but you maybe will :) Jun 04 16:40:56 But then can you tell me if it's possible to control IO ports with Apache and a PHP script on an ARM Device? Jun 04 16:41:33 ????? Jun 04 16:41:42 apache php Jun 04 16:41:45 sorry Jun 04 16:42:02 I like lighttp and rails Jun 04 16:43:00 03Henning Heinold  07org.openembedded.dev * r793e0d81c9 10openembedded.git/recipes/jikes/jikes_1.22.bb: jikes: change RDEPENDS so that jikes-native don't depends on target-classpath Jun 04 16:43:04 in the Angstrom image i'm using currently Apache is installed Jun 04 16:43:33 and I know PHP very well, so it would be great if I could control it with PHP Jun 04 16:43:58 are there some kind of Shell command then, to read or write to IO's? Because PHP can execute shell commands too Jun 04 16:47:27 mindThomas: I suggest you write simple c program for that and execute it from php Jun 04 16:47:43 sys.exec Jun 04 16:47:45 great Jun 04 16:48:47 If I write a C program, I have to crosscompile it using bitbake and my own recipe, just like the Hello World example - right? Jun 04 16:49:20 if you want that yes Jun 04 16:49:23 mindThomas: yes, or you could use a cross-toolchain Jun 04 16:49:34 bestway would be to use autotools Jun 04 16:50:35 well, if it's a really simple thing, a makefile would do Jun 04 16:51:33 zub but you a 50/50 chance that it will not crosscompile right Jun 04 16:51:40 woglinde: why? Jun 04 16:51:43 woo, my monkeypatched getvar appears to work, i think i can use this to audit the variable references for tasks Jun 04 16:52:01 monkeypatch? Jun 04 16:52:14 http://en.wikipedia.org/wiki/Monkey_patch Jun 04 17:17:01 args Jun 04 17:17:04 args Jun 04 17:17:04 args Jun 04 17:23:04 hmm, i think exceptions in events aren't being handled properly at all Jun 04 17:23:18 event handlers, that is Jun 04 17:46:29 Doesn't Apache have PHP as standard? Jun 04 18:41:26 woo Jun 04 18:41:29 this works Jun 04 18:42:29 gogogo Jun 04 18:43:12 the downside to this method is it only captures the vars referenced in the codepaths that are actually run Jun 04 18:43:17 so its not full proof Jun 04 18:43:24 but its at least a basic sanity check against my signature code Jun 04 18:45:55 hmmmm Jun 04 18:48:37 kergoth: got an interesting bug with amend class, if you are interested :) Jun 04 18:49:49 sure, what's the issue? Jun 04 18:50:02 (there are some known limitations, hopefully this isn't one of those :) Jun 04 18:51:07 kergoth: just have a simple amend.inc with PR_append = "-amend"... works fine when it sits in recipes/busybox/ and when it sits in recipes/busybox/busybox-1.13.2/ it expands 2 times and PR becomes -r29.1-amend-amend :) Jun 04 18:51:27 ieeks conditionrace with coreutils Jun 04 18:51:28 denix: is it in both locations? Jun 04 18:51:35 i think the current version parses all available amend.inc's Jun 04 18:51:47 kergoth: nope, just one Jun 04 18:52:00 /devel/arm/oetmp-ang/sysroots/i686-linux/usr/bin/cp: Das Programm kann nicht ausgeführt oder verändert werden (busy) Jun 04 18:52:02 well, the code does exactly the same thing in both cases Jun 04 18:52:05 engli Jun 04 18:52:06 sh Jun 04 18:52:15 programm cannt be executed or changed Jun 04 18:52:15 woglinde: someting trying to use cp while coreutils-native is staging, probably Jun 04 18:52:30 which seems odd, since its in the default deps Jun 04 18:52:38 re Jun 04 18:52:43 I was having this problem and had to _move_ amend.inc from version specific FILESPATH to generic... Jun 04 18:52:45 denix: well, the amend.inc already runs after the recipe does Jun 04 18:52:49 so why not use +=? Jun 04 18:52:59 i'd be curious to see if that behaves differently Jun 04 18:53:09 let me try that Jun 04 18:54:46 it could be something to do with how the finalize() bits are handled, though i don't see what offhand Jun 04 19:02:31 kergoth: hmm, += adds a space and screws things up big time... :) Jun 04 19:05:59 kergoth flaly net? Jun 04 19:06:17 trying to get the wifi AP to suck less Jun 04 19:06:18 heh Jun 04 19:06:31 ah Jun 04 19:07:54 denix: dunno if you got my message, but .= won't append a space, but is still immediate like += Jun 04 19:09:22 kergoth: the build is not finished yet, but from log messages I see that .= also expands twice, leading to -amend-amend suffix Jun 04 19:10:38 duplication in your FILESPATH, perhaps? Jun 04 19:10:44 i don't think amend.bbclass checks Jun 04 19:10:55 i.e. BPN vs PN Jun 04 19:13:14 kergoth: hmm, let me check FILESPATH... Jun 04 19:14:02 * kergoth suspects thats it Jun 04 19:14:18 probably just need to avoid parsing the same file twice, in amend.bbclass Jun 04 19:14:23 alternatively, filter out dupes in FILESPATH Jun 04 19:22:30 kergoth: FILESPATHPKG=${PF}:${P}:${PN}:${BP}:${BPN}:files:. Jun 04 19:22:46 right Jun 04 19:22:52 kergoth: and ${P} is the same as ${BP}, which is busybox-1.13.2 Jun 04 19:22:54 and if its not native or cross, B and BP will be the same Jun 04 19:22:56 right, exactly Jun 04 19:23:04 s/B and BP/P and BP/ Jun 04 19:30:30 hmm, i should backport the commit that makes it so LAYERDIR doesn't have to be immediately expanded, into 1.10 Jun 04 19:30:36 right now i can't use the same layer.conf for 1.10 and master Jun 04 19:31:14 ugh, just hit a circular python module reference with bitbake somehow Jun 04 19:31:18 :( Jun 04 19:37:01 kergoth: hmm, this problem with amend and PR is relatively new, as I don't remember seeing it before... Jun 04 19:38:13 kergoth: trying to see if any of the recent changes caused it... conf/bitbake.conf wasn't updated for couple weeks though Jun 04 19:40:47 denix: amend.bbclass was changed to load all amend.inc's in the filespath, not just the first Jun 04 19:41:55 ah, I see Jun 04 19:42:08 the issue was Jun 04 19:42:15 if there was an amend.inc in a collection the user didn't create Jun 04 19:42:22 and they created one of their own in a higher priority overlay Jun 04 19:42:25 quit Jun 04 19:42:27 opps Jun 04 19:42:31 it would blow away the changes, since it loads theirs instead Jun 04 19:42:37 they'd have to copy the bits from the other amend.inc into theirs Jun 04 19:42:41 thats why i changed it Jun 04 19:42:45 more intuitive this way, i think Jun 04 19:42:47 hi autoconf-2.65-r9.0 says "autom4te: need GNU m4 1.4 or later: /usr/bin/m4" should I bump m4? I bitbaken an eee710 xfce46-image, a bug sato2-image (not yet in oe) and now while bitbaking a bug illume-image and it fails...strange that m4 ...ah m5 not m4-native...I'll check Jun 04 19:43:13 also /usr/bin/m4 seems suspect Jun 04 19:43:16 I'll look Jun 04 19:43:21 gnuto m4 should be fixed Jun 04 19:43:25 it does make sense, just isn't applicable for me yet... :) Jun 04 19:43:30 * kergoth nods Jun 04 19:43:34 well, fixing this is pretty trivial Jun 04 19:43:42 just add a 'seen' set() to amend.bbclass Jun 04 19:43:48 don't parse if its already been done Jun 04 19:44:27 <_chase_> kergoth: So if two amend.inc files define a do_compile the one from the higher priority overlay will be overwritten by the next one that gets parsed? Jun 04 19:44:36 yes Jun 04 19:44:39 hey denix whats up with Linaro :) Jun 04 19:44:45 is TI full on to it Jun 04 19:44:47 that isn't the case for a _prepend/_append, because those don't override one another, its cumulative Jun 04 19:45:02 ok Jun 04 19:45:09 <_chase_> OK. So still need to be somewhat careful. Good to know about the _prepend/_append though. Jun 04 19:45:14 kergoth: sounds like a plan, let me try it Jun 04 19:45:18 <_chase_> thanks for this info. Jun 04 19:45:24 you always need to be careful when it comes to order of operations in the metadata Jun 04 19:45:25 I'll look as soon as I've some time Jun 04 19:45:29 and overrides Jun 04 19:45:41 we really need to add the ability to trace back and see *how* a variable got set the way it is Jun 04 19:45:51 set by this line in this file, or these lines in these files, or so Jun 04 19:45:52 khem: from what I heard, yes :) Jun 04 19:46:43 denix: are they needing ARM experts :) Jun 04 19:46:59 or is all stuff to be done in UK Jun 04 19:47:07 hm seems montavista lost some customers Jun 04 19:47:18 woglinde: ? Jun 04 19:47:22 khem you should ask hrw Jun 04 19:47:34 khem: not really sure Jun 04 19:47:36 kergoth linaro Jun 04 19:48:00 * kergoth hasn't had a chance to read about that yet, way too behind on news & mail Jun 04 19:48:02 and I didnt heard or read about montavista in last month Jun 04 19:48:10 's Jun 04 19:48:10 woglinde: I wonder if it turns out to be a fork of kernel or consolidation of all arm madnedd Jun 04 19:48:13 madness Jun 04 19:48:23 argh Jun 04 19:48:30 Message: 'module' object has no attribute 'localpath' Jun 04 19:48:36 that error is the bane of my existence Jun 04 19:48:54 hrms Jun 04 19:49:02 ~lart jamvm Jun 04 19:49:02 * ibot forces jamvm to use Outlook Express Jun 04 19:49:28 any one else has opinions on the patches for gettext upgrade I posted Jun 04 19:50:03 ~curse beta Jun 04 19:50:42 oh no drogba is out for football Jun 04 19:50:52 championchip Jun 04 19:51:03 woglinde: I am all geared up for world cup Jun 04 19:51:28 khem today car footbal championship is running Jun 04 19:51:31 with stefan raab Jun 04 19:51:43 woglinde: oh TV total Jun 04 19:51:48 * GNUtoo|laptop is fighting against plone4 beta bugs....I need to help someone...so not a lot of time for embedded theses days Jun 04 19:51:50 khem yes Jun 04 19:51:50 I remember him Jun 04 19:52:07 Anyone who can tell me how to get PHP on an ARM device Jun 04 19:52:15 If I already got Apache running, what should I do? Jun 04 19:52:21 bitbake php Jun 04 19:52:23 woglinde: if I choose proxy-libintl can it substitute for gettext fully Jun 04 19:52:25 or so Jun 04 19:52:31 no Jun 04 19:52:38 khem only libintl Jun 04 19:52:48 woglinde: ah ok Jun 04 19:52:56 hm Jun 04 19:53:03 woglinde: and then i get an ipkg which I just can install on my device? Jun 04 19:53:08 khem mom Jun 04 19:53:29 khem for target device the deps will not install any gettext stuff Jun 04 19:53:47 only libintl Jun 04 19:54:44 woglinde: USE_NLS does it mean more that libintl Jun 04 19:55:23 lol Jun 04 19:55:29 italy lost against mexico Jun 04 19:55:46 khem only .mo files Jun 04 19:55:59 woglinde: ok Jun 04 19:56:09 woglinde: I am trying to sort out uclibc builds Jun 04 19:56:14 which will all have there one package Jun 04 19:56:29 hm Jun 04 19:56:29 and at one point of time we could get gpe images without getttext if I not mistaken Jun 04 19:56:40 hm sure Jun 04 19:56:51 right now its pulling in gettext in some cases Jun 04 19:56:53 use proxy-libintl Jun 04 19:56:59 yeah I have it Jun 04 19:57:13 but I think I have some recipes asking for gettext Jun 04 19:57:21 I will sort it out Jun 04 19:57:51 bitbake -g Jun 04 19:57:56 use master Jun 04 19:58:05 the new third dep file Jun 04 19:58:07 is really good Jun 04 19:58:18 to find out who is pulling it in Jun 04 19:59:01 btw bitbake master has -i broken Jun 04 19:59:53 gnutoo its known Jun 04 20:01:48 woglinde: who are you rooting for after germany Jun 04 20:02:01 I am rooting for spain and argentina Jun 04 20:02:14 hm we will see Jun 04 20:02:23 I like the netherlands Jun 04 20:02:32 but in the big tourniers they play crap Jun 04 20:02:35 robben ? Jun 04 20:02:46 synders Jun 04 20:02:48 hmmm parsing is fast Jun 04 20:02:50 ups snyders Jun 04 20:05:14 khem http://de.wikipedia.org/wiki/Niederl%C3%A4ndische_Fu%C3%9Fballnationalmannschaft#Saison_2009.2F10 Jun 04 20:06:06 Sneijder Jun 04 20:07:03 hm without van breukelin Jun 04 20:07:12 as goalie Jun 04 20:07:23 van Bommel hmmm unlucky guy I saw his lose the final to Inter Jun 04 20:07:37 robben too Jun 04 20:07:50 what happened to van der saar Jun 04 20:07:59 ups I meant van der saar Jun 04 20:08:02 sorry Jun 04 20:08:06 to old Jun 04 20:08:10 heh Jun 04 20:08:10 he is nearly 40 Jun 04 20:08:11 ok Jun 04 20:08:23 he still plays in the league though Jun 04 20:08:43 yeah in england I guess Jun 04 20:08:45 lets google Jun 04 20:09:16 manu Jun 04 20:09:18 till 2011 Jun 04 20:10:35 he played 13 years in the national team Jun 04 20:10:42 http://de.wikipedia.org/wiki/Edwin_van_der_Sar Jun 04 20:11:37 yeah he is good golie Jun 04 20:11:49 but italian one is the one with iron hands Jun 04 20:14:44 woglinde: glib-2.0 needs gettext Jun 04 20:16:03 back for oe works Jun 04 20:16:41 khem hm not really Jun 04 20:16:48 I think its some koen stuff Jun 04 20:17:11 try to unser USE_NLS for glib Jun 04 20:17:17 it should work Jun 04 20:17:39 but never tested it Jun 04 20:17:45 woglinde, you told me to fix m4,but shouldn't I fix autoconf instead because of that message: "autom4te: need GNU m4 1.4 or later: /usr/bin/m4" it picks the host's m4 Jun 04 20:17:48 and m4 is staged Jun 04 20:17:56 a which on devshell gives: Jun 04 20:17:59 gnutoo right Jun 04 20:18:11 /home/gnutoo/embedded/oe/oetmps/bug/sysroots/x86_64-linux/usr/bin/m4 Jun 04 20:18:23 woglinde: yep Jun 04 20:18:24 dont know why its using host m4 Jun 04 20:18:28 just commented it out Jun 04 20:18:36 rebuilding again with -g Jun 04 20:18:54 any opinions on the 4 bitbake.conf changes i have in patchwork at the moment? minor stuff Jun 04 20:19:26 I'll push a new scummvm btw but not today Jun 04 20:19:31 kergoth: I think they looked ok from surface Jun 04 20:19:40 also I have converted to INC_PR Jun 04 20:19:42 kergoth: although I dont know the impact Jun 04 20:19:51 this may have Jun 04 20:20:01 btw now that I have a core i7 maybe I could test openoffice recipe Jun 04 20:20:16 kergoth: I especially liked the EXTENDPE :) Jun 04 20:20:29 gee, patchwork is so slow... Jun 04 20:20:51 http://patchwork.openembedded.org/patch/2111/ Jun 04 20:21:03 I wonder if it has any problems with cross packages Jun 04 20:21:32 DJWillis, can I bump scumvm soon(not today)? any objections Jun 04 20:21:37 hi btw Jun 04 20:21:40 is there a way to put a carriage return seperator in my local.conf file for PREMIRRORS? It's breaking when angstrom-mirrors appends it's own mirrors Jun 04 20:21:55 kergoth: all 4 look sane to me :) Jun 04 20:21:59 rphillips: why not a space? Jun 04 20:22:09 khem: i don't see how. BUILD_ARCH and BUILD_OS never change. Jun 04 20:22:16 GNUtoo|laptop: it's not my recipe or I would have bumped it a while ago, seems like a great idea, 1.1.1 is pretty solid. Jun 04 20:22:20 in native it just sets TARGET to match, cross does similarly Jun 04 20:22:21 denix: it isn't splitting on a space... bitbake is splitting on a \n Jun 04 20:22:38 DJWillis, ok but you've got commit access no? Jun 04 20:22:43 * kergoth thinks Jun 04 20:22:45 who's recipe is it then? Jun 04 20:23:04 ok so no gettext finally in native-sdk-image Jun 04 20:23:10 lets build Jun 04 20:23:42 03Denys Dmytriyenko  07org.openembedded.dev * r4d642fd5d3 10openembedded.git/classes/amend.bbclass: Jun 04 20:23:42 amend.bbclass: don't parse same amend.inc if FILESPATH references it multiple times Jun 04 20:23:42 Signed-off-by: Denys Dmytriyenko Jun 04 20:23:42 Acked-by: Chris Larson Jun 04 20:23:59 may be I will get stuck in libiconv which sort of needs gettext Jun 04 20:24:01 lets see Jun 04 20:24:06 GNUtoo|laptop: no I don't have commit, no idea who it is, I fixed up the Angstrom tweaks at the ScummVM end (I have commit there ;-)). I have a 1.1.1 recipe on my PC but it is building plugins (so more engines run on low end devices) but the recipe could use a cleanup Jun 04 20:24:37 denix: bitbake-1.8.18/lib/bb/fetch/wget.py line 108 Jun 04 20:24:58 DJWillis, yes specially for INC_PR Jun 04 20:25:03 I've a recipe too locally Jun 04 20:25:13 but I bet your recipe would be better than mine Jun 04 20:25:17 but mine has INC_PR Jun 04 20:25:21 or maybe Jun 04 20:25:23 I push Jun 04 20:25:27 and you correct on top? Jun 04 20:25:43 let me pastebin a diff Jun 04 20:25:55 rphillips: can you do MIRRORS_append? Jun 04 20:26:18 or _prepend Jun 04 20:26:30 GNUtoo|laptop: push, I am busy with some real life stuff and a few blog posts I got bullied into writing about setting up OpenPandora stuff anyway ;) Jun 04 20:26:37 denix: testing that... Jun 04 20:26:50 lol ok Jun 04 20:27:15 where is your recipe so I rebase your changes on top of mine and also credit you in the commit message Jun 04 20:27:33 * Tartarus does some FILESPATH -> FILESPATHPKG changing and bitbake -e'ing Jun 04 20:28:29 (I'm interested in your ./configure options) Jun 04 20:29:07 GNUtoo|laptop: it does not have all the ARM codepaths in yet, i'll send it over as soon as I get home. Jun 04 20:29:31 ok thanks Jun 04 20:29:57 btw how much a pandora costs? Jun 04 20:30:30 GNUtoo|laptop: no idea, I only have an old devboard, £200+ I think. Jun 04 20:30:50 ok Jun 04 20:30:54 I'll convert that Jun 04 20:34:29 denix: no joy... still the same error: too many values to unpack. Jun 04 20:38:17 I wonder something.... Jun 04 20:38:37 autoconf need some autotools for generating his ./configure Jun 04 20:38:42 mmm Jun 04 20:39:26 mmm devshell gives Jun 04 20:39:27 checking for GNU M4 that supports accurate traces... /home/gnutoo/embedded/oe/oetmps/bug/sysroots/x86_64-linux/usr/bin/m4 Jun 04 20:39:32 I'll try to clean and rebuild Jun 04 20:39:42 maybe some issues with multiple threads Jun 04 20:39:47 of bitbake Jun 04 20:40:31 that's the equivalent of the speed ticket Jun 04 20:41:51 mmm still /usr/bin/m4 Jun 04 20:42:02 ahhh Jun 04 20:42:05 packaged staging Jun 04 20:42:21 how do I prevent it from re-extracting the exact same tmpdir? Jun 04 20:42:41 I had the issue while patching scummvm for preventing stripping Jun 04 20:42:52 that is to say I create a patch Jun 04 20:43:00 and then it stays in the dir Jun 04 20:43:07 bitbake -f Jun 04 20:43:12 ok thanks a lot Jun 04 20:43:29 -f clean? Jun 04 20:43:44 no Jun 04 20:43:49 first clean Jun 04 20:43:56 ok I get it Jun 04 20:43:58 than bitbake -c compile -f foo Jun 04 20:44:18 ok thanks a lot Jun 04 20:48:38 I found out do_install of udev/attr breaks when the group of the user running bitbake contains a space. That's most probably an upstream issue. But - how non-standard is it for the froup name to contain space? Is it even considered legal? Jun 04 20:48:54 either way, on that box the name of the group is "Domain Users" :-/ Jun 04 20:49:20 and I see: ../include/install-sh ... -g Domain Users ... Jun 04 20:49:50 this obviously breaks: cp: cannot stat `Users': No such file or directory Jun 04 20:52:12 GNUtoo|laptop: http://pastebin.com/VuxgtRJy is the 1.1.1 no strip patch Jun 04 20:52:17 ieeehks groupnames with space Jun 04 20:52:21 ok thanks Jun 04 20:52:24 I've one too Jun 04 20:52:28 :) Jun 04 20:52:35 woglinde: I'm just a poor user there :) Jun 04 20:52:37 but I was interested in the recipe Jun 04 20:52:43 specially the configure options Jun 04 20:52:56 I'll force configure Jun 04 20:53:41 still configure:2401: result: /usr/bin/m4 Jun 04 20:53:57 oe_runconf works well in devshell Jun 04 20:53:59 mmm Jun 04 20:54:29 GNUtoo|laptop: ahh, I had the patch on my work laptop, hmmm, let me see what I can get from SSH to my home box, got a mail I can send the files over to? I still need to strip the plugins in my recipe. Jun 04 20:54:42 GNUtoo A no-log.org Jun 04 20:55:43 I've done Jun 04 20:55:46 touch GNUtoo Jun 04 20:55:50 let's see if it persist Jun 04 20:56:52 no GNUtoo file but still /usr/bin/m4 Jun 04 20:56:55 strange strange Jun 04 20:57:18 GNUtoo|laptop: sent over. Jun 04 20:57:39 ok thanks a lot Jun 04 20:59:16 mmm Jun 04 20:59:21 export ac_cv_path_M4="${ac_cv_path_M4=${bindir}/m4}" Jun 04 20:59:32 I'll look Jun 04 21:01:58 what's ac_cv_path? Jun 04 21:02:08 it's not in the tutorial Jun 04 21:02:24 path for finding m4 macros? Jun 04 21:02:33 that in this case will look for m4 Jun 04 21:02:34 ? Jun 04 21:09:10 if I comment that export ac_cv_path_M4="${ac_cv_path_M4=${bindir}/m4}" it works fine Jun 04 21:12:13 thanks a lot DJWillis Jun 04 21:46:22 ah ok Jun 04 21:46:23 http://en.wikipedia.org/wiki/File:Screenshot-Ubuntu-Software-Centre.png Jun 04 21:46:33 maybe it was in a state where it didn't displayed them Jun 04 21:46:41 or maybe I only looked on the left Jun 04 21:46:44 oops Jun 04 21:46:53 sorry Jun 04 21:47:06 I was talking about software center on #trisquel Jun 04 22:12:58 hmmm all these site cached variables probably should be moved into recipes Jun 04 22:14:02 I think many times they might be injecting wrong values into autoconf results Jun 04 22:15:03 we could use qemu to do the runtime tests :) Jun 04 22:36:14 khem: not all are recipe specific. many are common. recipes that use glib may need glib results, whereas glib itself doesn't Jun 04 22:39:39 yeah Jun 04 22:40:18 I think one time I will use qemu- to interface autoconf with sysroot Jun 04 22:40:22 and see how it goes Jun 04 22:41:14 that'd be cool Jun 04 23:04:37 kergoth: I have flex-native in ASSUME_PROVIDED but its adamant that it wants it from /scratch/oe/sysroots/i686-linux/usr/bin/flex Jun 04 23:04:46 why it does not look into /usr/bin Jun 04 23:06:05 ha EXTRA_OEMAKE = "-e 'STRIP=echo' 'LIB_OBJS=-lc -lsysfs' LEX='${STAGING_BINDIR_NATIVE}/flex'" Jun 04 23:06:12 thats why Jun 04 23:06:41 recipes/pcmciautils/pcmciautils.inc Jun 04 23:07:07 * khem does a git praise Jun 04 23:07:52 c65ee13298e8c22274b5bcabe06f2116b8b7ebf2 Jun 04 23:08:35 pcmciautils: fix build to use flex from sysroots, not from host Jun 04 23:08:40 why ? Jun 04 23:11:19 heh Jun 04 23:11:28 if it has to be a full path, which i doubt, /usr/bin/env is our friend Jun 04 23:17:40 I dont like this hard dependency on host file being picked from sysroot Jun 04 23:17:46 agreed. Jun 04 23:18:05 I need to ask the author why he thinks so before I revert this part Jun 04 23:18:14 it was committed today Jun 04 23:28:50 kergoth: your recent patch, have you tried it on openmoke ? Jun 04 23:28:58 openmoko Jun 04 23:29:17 or was it an overright that you are fixing Jun 04 23:31:27 s/overright/oversight/ Jun 04 23:35:44 will double check sanity with -e Jun 04 23:36:01 * kergoth suspects those were just copy/paste errors Jun 04 23:36:22 * kergoth 'll reply to the thread after he verifies Jun 04 23:37:25 * kergoth hugs git-wtf Jun 04 23:40:01 * kergoth tests bernhard's patch series Jun 04 23:55:05 okay, i'm sick of 'new RPATH' spam Jun 04 23:55:58 heh Jun 04 23:56:01 bb.debug time? Jun 04 23:56:21 * Tartarus pushes all the rest FILESPATH -> FILESPATHPKG changes Jun 04 23:56:27 03Tom Rini  07org.openembedded.dev * ra67d228220 10openembedded.git/recipes/freetype/ (5 files): Jun 04 23:56:27 freetype-native: Drop unnecessary FILESPATH Jun 04 23:56:27 Signed-off-by: Tom Rini Jun 04 23:56:28 03Tom Rini  07org.openembedded.dev * re97414ea1f 10openembedded.git/recipes/genext2fs/genext2fs_1.3+cvs.bb: Jun 04 23:56:28 genext2fs (1.3+cvs): Switch to FILESPATHPKG Jun 04 23:56:34 03Tom Rini  07org.openembedded.dev * re81448c639 10openembedded.git/recipes/emacs/ (emacs-x11_22.3.bb emacs-x11_23.1.bb): Jun 04 23:56:34 emacs-x11: Switch to FILESPATHPKG Jun 04 23:56:35 Signed-off-by: Tom Rini Jun 04 23:56:35 03Tom Rini  07org.openembedded.dev * rc80700c6bf 10openembedded.git/recipes/gtk+/gdk-pixbuf_2.10.14.bb: Jun 04 23:56:36 gdk-pixbuf (2.10.14): Switch to FILESPATHPKG Jun 04 23:56:36 Signed-off-by: Tom Rini Jun 04 23:56:37 03Tom Rini  07org.openembedded.dev * r82c83e6563 10openembedded.git/recipes/busybox/busybox-static_1.11.3.bb: Jun 04 23:56:39 busybox-static: Switch to FILESPATHPKG Jun 04 23:56:39 Signed-off-by: Tom Rini Jun 04 23:56:39 03Tom Rini  07org.openembedded.dev * ree30da53f5 10openembedded.git/recipes/bzip2/bzip2-full-native_1.0.5.bb: Jun 04 23:56:47 Signed-off-by: Tom Rini Jun 04 23:56:47 03Tom Rini  07org.openembedded.dev * rb132d5d42d 10openembedded.git/recipes/mesa/ (mesa-dri_6.5.2.bb mesa-dri_7.0.3.bb): Jun 04 23:56:48 mesa-dri (6.5.2 / 7.0.3): Switch to FILESPATHPKG Jun 04 23:56:48 Signed-off-by: Tom Rini Jun 04 23:56:49 03Tom Rini  07org.openembedded.dev * r1d1cdd05ff 10openembedded.git/recipes/linux/ (7 files): Jun 04 23:56:49 linux-handhelds: Switch to FILESPATHPKG Jun 04 23:56:50 Signed-off-by: Tom Rini Jun 04 23:57:00 03Tom Rini  07org.openembedded.dev * r8326b50257 10openembedded.git/recipes/xorg-xserver/xserver-kdrive-imageon_1.2.0.bb: Jun 04 23:57:00 xserver-kdrive-imageon: Switch to FILESPATHPKG Jun 04 23:57:00 Signed-off-by: Tom Rini Jun 04 23:57:03 03Tom Rini  07org.openembedded.dev * r73fec08e31 10openembedded.git/recipes/xorg-xserver/xserver-kdrive-nomadik_1.3.0.0.bb: Jun 04 23:57:03 xserver-kdrive-nomadik: Switch to FILESPATHPKG Jun 04 23:57:03 Signed-off-by: Tom Rini Jun 04 23:57:03 03Tom Rini  07org.openembedded.dev * re8f2c5e1e0 10openembedded.git/recipes/xorg-xserver/xserver-kdrive-glamo_1.3.0.0+git.bb: Jun 04 23:57:03 xserver-kdrive-glamo: Switch to FILESPATHPKG Jun 04 23:57:15 snes9x-sdl-qpe: Switch to FILESPATHPKG Jun 04 23:57:15 Signed-off-by: Tom Rini Jun 04 23:57:15 03Tom Rini  07org.openembedded.dev * reaf6296d71 10openembedded.git/recipes/python/python-setuptools-native_0.6c9.bb: Jun 04 23:57:15 python-setuptools-native: Drop unnecessary FILESPATH Jun 04 23:57:15 Signed-off-by: Tom Rini Jun 04 23:57:29 03Tom Rini  07org.openembedded.dev * r9ed234c270 10openembedded.git/recipes/netkit-rpc/netkit-rpc-native_0.17.bb: Jun 04 23:57:29 netkit-rpc-native: Drop unnecessary FILESPATH Jun 04 23:57:29 Signed-off-by: Tom Rini Jun 04 23:57:29 03Tom Rini  07org.openembedded.dev * r0997ea8069 10openembedded.git/recipes/xorg-xserver/xserver-kdrive_git.bb: Jun 04 23:57:30 xserver-kdrive (git): Switch to FILESPATHPKG Jun 05 00:04:36 Also, posting a patch to rootfs_ipk.bbclass to the ML, comments / acks please Jun 05 00:04:43 * Tartarus hopes tbird doesn't goof it up Jun 05 02:44:30 I was told, back in the "oebuild" days, you guys had a really nifty way of getting bash to call functions in python. Anyone know where I can find more info about this? Maybe an old repo, or OE code history? Jun 05 02:56:40 Nothing like that exists today. Jun 05 02:57:54 you can use a python snippet to execute python code and put the results in the emitted shell, via ${@} Jun 05 02:57:56 thats about it Jun 05 02:58:02 it's likely we'll implement helper commands that interact with the backend via xml/rpc at some point, to let the shell tasks send output to the user, since the frontend was split from the backend in master, i expect Jun 05 02:58:23 OK **** ENDING LOGGING AT Sat Jun 05 02:59:57 2010