**** BEGIN LOGGING AT Thu Mar 19 02:59:57 2009 Mar 19 06:07:35 i am trying to use an external toolchain, i get the message "external toolchain could not be found in /usr/local/angstrom/arm!" Mar 19 06:07:43 has anyone faced this issue before? Mar 19 06:34:30 where do i set the prefix variable? Mar 19 06:34:44 or which environmenta variable controls the ${prefix} in oe Mar 19 08:05:10 morning Mar 19 09:01:14 jo Mar 19 09:09:07 gm Mar 19 09:09:16 hi likewise Mar 19 09:16:00 hi gremlin Mar 19 09:39:58 morning jay7 Mar 19 09:40:04 * Jay7 was forked :) Mar 19 09:40:10 woglinde: hey :) Mar 19 09:56:07 why is external toolchain.bb not working? Mar 19 09:56:34 what exactly is not working? Mar 19 09:57:49 woglinde: The external toolchain could not be found in /usr/local/angstrom/arm! Mar 19 09:58:08 did you install the toolchain there? Mar 19 09:59:38 no i have my toolchain at /opt/. the problem is the ${prefix} variable is awalys at /usr/local/angstrom. how do i control ${prefix} Mar 19 09:59:55 is there an environmental variable to control ${prefix} Mar 19 10:01:01 guess you could link it from ${prefix} to your location in opt Mar 19 10:01:18 not pretty, but quick Mar 19 10:01:58 i dont want to do that. i tried changing external_toolchain.bb and i moved forward a little. i dont think that is the right way to do it Mar 19 10:02:21 has anyone compiled oe with csl? Mar 19 10:02:29 csl is inside oe Mar 19 10:03:06 i dont want to compile csl, i have installed csl on my system at /opt, i want to oe to use it Mar 19 10:03:12 thats what im trying to do Mar 19 10:03:54 /devel/arm/git/openembedded/recipes/gcc/gcc-csl-arm-2008q3.inc Mar 19 10:04:10 morning Mar 19 10:04:13 how is ${prefix} contructed? Mar 19 10:04:13 hi hrw Mar 19 10:04:31 vsriram hm hrw knows it I believe Mar 19 10:04:45 vsriram: it is set to /usr/ in conf/bitbake.conf iirc Mar 19 10:05:01 likewise: yt? Mar 19 10:05:11 hrw he is trying to use external toolchain Mar 19 10:05:44 but the tools are always looked up under /usr/local/angstrom/arm Mar 19 10:05:45 $prefix is not related to toolchain Mar 19 10:06:05 vsriram: add your toolchain to $PATH Mar 19 10:06:31 * mckoan just received a copy od an italian magazine "Elettronica Oggi News" where is mentioned OE :-) Mar 19 10:06:49 woglinde: what's csl anyway? doesn't ring a bell for me :) Mar 19 10:07:08 * mckoan worked to spreading OE and now results start rising Mar 19 10:07:10 gilligan gcc from codesourcery Mar 19 10:07:19 often merged to gcc upstream Mar 19 10:08:21 hrw: what is ${prefix} related to? Mar 19 10:08:48 http://www.ilb2b.it/eo_news/pdf_rivista/EONEWS_509.pdf page 12 Mar 19 10:08:53 CodeSourceryLabs Mar 19 10:08:54 woglinde: oh I see.. didn't know that one Mar 19 10:09:02 and the new not yet published n.510 Mar 19 10:10:35 mckoan: OpenEmbedded is MIT licensed not GPL Mar 19 10:11:33 03Koen Kooi  07org.openembedded.dev * r51b6d90f1d 10openembedded.git/ (4 files in 3 dirs): glib-2.0: add 2.20 Mar 19 10:11:39 03Koen Kooi  07org.openembedded.dev * rf608c18453 10openembedded.git/recipes/gtk+/ (5 files in 2 dirs): gtk+: add 2.16.0 Mar 19 10:12:14 hehe Mar 19 10:12:19 koen is back Mar 19 10:12:31 hrw: yes was my fault Mar 19 10:18:59 I think I could candidate myself as OE PR manager :-D Mar 19 10:23:08 newbie question: when a recipe changes (eg. git pull) do i have to do to do anything extra other than run bitbake again to see the changes in the root_fs? Mar 19 10:23:28 markos_: no Mar 19 10:23:40 i'm asking because mtd-utils binaries (mkfs.jffs2) doesn't appear in the final root_fs Mar 19 10:23:45 mckoan: then start also doing PR on our website ;) Mar 19 10:23:54 markos_: does your image installs it? Mar 19 10:23:56 i've added it to my recipe and it gets built, the binaries are there Mar 19 10:23:56 not sure, there is a line in external_toolchain.bb which says: if [ ! -e ${prefix}/package-status ]; then Mar 19 10:24:26 hrw: i've added IMAGE_INSTALL += "mtd-utils" to my recipe Mar 19 10:24:35 isn't that enough? Mar 19 10:25:17 hrw, give me time I'll do that too ;-) Mar 19 10:25:28 one step at a time Mar 19 10:25:54 markos_: mkfs.jffs2 is in separate package Mar 19 10:27:29 hrw: sorry if I'm appearing to ask stupid questions, i'm totally new to OE, and i can't say i found much documentation about this, do i have to do anything extra then? Mar 19 10:27:48 markos look at the recipe Mar 19 10:27:54 markos_: add mkfs.jffs2 into IMAGE_INSTALL Mar 19 10:28:04 markos_: OE recipe can generate more then 1 package Mar 19 10:28:10 right Mar 19 10:28:16 then you see which pacakges with which files are generated Mar 19 10:29:00 hrw: before doing PR on OE website, I need to be more involved in OE activites like repo puch enabled and access to private ML, but I understand this require time and that you must trust me ;-) Mar 19 10:29:17 ok, i think i got that, mtd-utils produces more than one package and putting mtd-utils in the IMAGE_INSTALL doesn't propagate to the produced packages, right? I have to explicitly enter which ones I want Mar 19 10:29:35 markos right Mar 19 10:29:47 s/puch/push Mar 19 10:30:14 heh.. mtd-utils needs work again Mar 19 10:30:28 hrw hm? Mar 19 10:31:15 hrw whats broken? Mar 19 10:32:52 woglinde: I extracted mkfs.jffs2 over year ago to not get lzo as dependency for mtd-utils package. Mar 19 10:33:04 woglinde: not mtd-utils contains ubi tools and depend on lzo again Mar 19 10:33:58 ok, and sth else, i know now how to add packages to an image, is there a way to remove packages from an image? eg. I'm basing on console-image and it includes stuff like wpa, bluetooth, etc, which I do not need. Mar 19 10:33:58 03Koen Kooi  07org.openembedded.dev * r2612236bf6 10openembedded.git/ (4 files in 2 dirs): abiword: update to 2.6.8 Mar 19 10:34:15 03Koen Kooi  07org.openembedded.dev * rebd561bc14 10openembedded.git/recipes/linux/ (68 files in 11 dirs): linux-omap-pm: add 2.6.28 recipe so git can move to .29rc Mar 19 10:34:29 or do i have to clone console-image and just remove what I don't need and base my image on this one? Mar 19 10:34:53 hrw blame mickeyl Mar 19 10:35:02 he doenst put ubifs into seperate package Mar 19 10:36:10 ? Mar 19 10:36:16 didn't touch ubi for ages Mar 19 10:36:35 markos_: no way to remove unfortunately. We that want that are the minority Mar 19 10:36:45 i have tried to convince folks to do -= Mar 19 10:36:50 but it was shooted down :) Mar 19 10:37:00 so you need to add your own console-image Mar 19 10:37:08 it would be a nice feature Mar 19 10:37:20 not a big problem Mar 19 10:37:22 there's also BAD_RRECOMMENDS, if you're on angstrom Mar 19 10:37:29 with that you can remove some things out of task-base Mar 19 10:37:32 iirc Mar 19 10:37:49 usually wpa and bluetooth should not be included if you don't have it in your machine features though Mar 19 10:37:58 MACHINE_FEATURES Mar 19 10:38:15 can I override that? Mar 19 10:39:16 it's set in your machine Mar 19 10:39:36 well, machine is efika now, it doesn't support them -by default- but one could add wifi/bluetooth usb/pci card Mar 19 10:39:43 ah Mar 19 10:39:45 corner case then Mar 19 10:39:49 check whether they do it with ?= or = Mar 19 10:39:52 markos_: use console-base-image then Mar 19 10:39:54 in the first case you can override Mar 19 10:41:08 uhm, MACHINE_FEATURES = "kernel26 usbhost ext2 alsa" only Mar 19 10:41:30 so I guess i'll try console-base-image instead, perhaps it's more what I want Mar 19 10:42:49 it is Mar 19 10:42:57 I created it specially for such situations Mar 19 10:46:32 ok, final question, how do I tamper with boot configurations in the final image? (i mean without tampering with the rootfs everytime). Is there some kind of 'task/job' that I can run after the install phase to fix some boot parameters (like setup network, hostnames, random root passwords, etc)? Mar 19 10:49:47 I wonder does someone use mtd-utils <1.2.0 Mar 19 10:49:56 as I would like to drop old versions during work on them Mar 19 10:50:05 hrw grep conf Mar 19 10:50:11 and then throw it away Mar 19 10:51:03 slugos uses 1.1.0 Mar 19 10:51:18 hrw hm Mar 19 10:53:25 is there a way to set the size of the final produced image (ext3/jffs2, not the tarball) Mar 19 10:54:20 good morning Mar 19 10:54:27 hi florian Mar 19 10:54:50 markos_: ext2/3 use IMAGE_ROOTFS_SIZE iirc if it is big enough to contain image Mar 19 10:55:31 markos_: jffs2 usually do not need to fill whole mtd partition as it use whole space available by itself. you need to erase flash before writing jffs2 into it Mar 19 10:56:03 hrw: nice to know, thanks Mar 19 11:00:05 where should i specify the PROVIDER, i often get multiple providers warnings Mar 19 11:00:50 vsriram: in distro conf or local conf Mar 19 11:04:25 i dont know. i have PREFERRED_PROVIDER_virtual/${TARGET_PREFIX}binutils = "csl-toolchain_2008q3" in distro.conf Mar 19 11:04:36 but i dont where this checking takes place Mar 19 11:07:34 hrw: for some strange reason, it still takes the external-toolchain.bb and not csl-toolchain_2008q3 even though i have it in my distro.conf Mar 19 11:09:07 vsriram: recipe is csl-toolchain_2008q3.bb? Mar 19 11:09:18 yes Mar 19 11:09:34 vsriram: use PREFERRED_PROVIDER_virtual/${TARGET_PREFIX}binutils = "csl-toolchain" and PREFERRED_VERSION_csl-toolchain = "2008q3" Mar 19 11:17:26 ok. mtd-utils now depends on libc and libgcc Mar 19 11:17:28 hrw: this time i get the error: ERROR: Conflicting PREFERRED_PROVIDER entries were found which resulted in an attempt to select multiple providers Mar 19 11:17:41 hrw: http://pastebin.ca/1365153 Mar 19 11:19:40 vsriram: you lack fre pref-prov entries Mar 19 11:22:54 hrw: i have PREFERRED_PROVIDER_virtual/${TARGET_PREFIX}binutils = "csl-toolchain" in distro.conf file Mar 19 11:23:56 ok, oe boots, works out of the box, amazing work guys :D Mar 19 11:25:04 vsriram: consider defining PREFERRED_PROVIDER_virtual/arm-none-linux-gnueabi-gcc-intermediate consider defining PREFERRED_PROVIDER_virtual/arm-none-linux-gnueabi-libc-initial PREFERRED_PROVIDER_virtual/arm-none-linux-gnueabi-libc-for-gcc Mar 19 11:25:15 vsriram: notes from bitbake are worth reading Mar 19 12:01:19 hrw: re: yt? (?) Mar 19 12:06:51 likewise: YouThere Mar 19 12:07:25 likewise: do you have any sheevaplugs left? and can sell them with invoice? Mar 19 12:10:09 hrw: yes, 1, yes Mar 19 12:10:28 hrw: I'm not sure what the tax/import duties will be yet. Mar 19 12:12:20 there are some other then vat in .nl? Mar 19 12:25:57 ~curse git-send-email Mar 19 12:25:58 May the fleas of a thousand camels infest your most sensitive regions, git-send-email ! Mar 19 12:37:12 anyone knows why pulseaudio is stuck at version 0.9.10? the 0.9.11 is set to default_preference = -1.. Mar 19 12:37:42 I do not think it builds Mar 19 12:37:54 I'm interested in changing that Mar 19 12:39:47 03Koen Kooi  07org.openembedded.dev * rb7423a748d 10openembedded.git/ (conf/checksums.ini recipes/gnome/libproxy_0.2.3.bb): libproxy: add 0.2.3 Mar 19 12:39:48 03Koen Kooi  07org.openembedded.dev * r432c38351a 10openembedded.git/ (5 files in 2 dirs): webkit-gtk: bump SRCREV Mar 19 12:39:55 03Koen Kooi  07org.openembedded.dev * r34533ec864 10openembedded.git/recipes/gnome/libsoup-2.4_2.26.0.bb: libsoup: add 2.26.0 Mar 19 13:08:54 03Michael 'Mickey' Lauer  07org.openembedded.dev * rd2b2368415 10openembedded.git/conf/bitbake.conf: bitbake.conf: weak assignment for DEPLOY_DIR Mar 19 13:08:55 03Michael 'Mickey' Lauer  07org.openembedded.dev * ra0169ba065 10openembedded.git/conf/distro/include/fso-autorev.inc: fso-autorev: add libeflvala Mar 19 13:08:56 03Michael 'Mickey' Lauer  07org.openembedded.dev * rb884b02725 10openembedded.git/conf/distro/ (include/sane-toolchain.inc minimal.conf): Mar 19 13:08:56 sane-toolchain.inc: move non-weak assignments for CACHE and DEPLOY_DIR to minimal.conf; make them weak Mar 19 13:08:58 You can override DEPLOY_DIR now in your local.conf. Mar 19 13:19:08 hrw: SlugOS uses the old one because of some sort of problem with the fetcher for 1.2 -- i do not not recall the details, but I was trying to get a release done at the time so rather than fix it, I just set the PREFERRED_VERSION. I can take another look at it. Mar 19 13:19:20 ok Mar 19 13:21:54 what are the sign off words again? Mar 19 13:22:01 basically I agree, but can't test Mar 19 13:51:55 does oe support arm? Mar 19 13:52:22 03Angus Ainslie  07fso/milestone5.5 * r48cb8c21a7 10openembedded.git/classes/openmoko2.bbclass: Allow openmoko2 to succeed on unsplitable sections Mar 19 13:52:30 03Angus Ainslie  07fso/milestone5.5 * r6290b1b7fc 10openembedded.git/packages/efl1/edje-sdk_svn.bb: Add enia library paths Mar 19 14:01:25 error404notfound: yes Mar 19 14:15:44 re Mar 19 14:17:02 hrw: wb Mar 19 14:23:02 which lightweight browser would you recommend for inclusion in OE for an embedded device with display output? probably with desktop env, just a fullscreen browser would be fine. i might use firefox with predefined settings to go fullscreen, but if there is a lighter solution that would be even better Mar 19 14:23:22 midori gpe-minibrowser(2) Mar 19 14:24:47 thanks Mar 19 14:26:25 hm, that's more targetted to pda-like displays, my display would be ~1024x600 or more Mar 19 14:27:34 does anyone know if someone is putting any effort into creating Qt embedded 4.5 recipes? Mar 19 14:27:45 midori looks a better choice :) Mar 19 14:28:35 jimsheldon: Qt 4.5 dropped embedded part iirc Mar 19 14:29:30 hrw: ah I must have missed that, thanks Mar 19 14:30:29 jimsheldon: we are trying to add Qt-4.5 (x11) recipe Mar 19 14:30:50 hrw: ht/e is still supported Mar 19 14:30:58 s/ht/qt Mar 19 14:31:02 qt/e has to die Mar 19 14:31:03 mckoan: oh cool Mar 19 14:31:06 definitelly Mar 19 14:31:20 * hrw -> other tasks now Mar 19 14:44:27 03Khem Raj  07hrw/poky-merge/add-rpm-packages-support * r9e83f3f1b9 10openembedded.git/: Merge branch 'org.openembedded.dev' of git@git.openembedded.net:openembedded into org.openembedded.dev Mar 19 14:44:27 prepare for CIA-2 flood Mar 19 14:44:30 03Robert Schuster  07hrw/poky-merge/add-rpm-packages-support * r994c99d511 10openembedded.git/conf/checksums.ini: checksums.ini: Added cmake 2.6.3 entry. Mar 19 14:44:31 03Khem Raj  07hrw/poky-merge/add-rpm-packages-support * r3992e7dd43 10openembedded.git/: Merge branch 'org.openembedded.dev' of git@git.openembedded.net:openembedded into org.openembedded.dev Mar 19 14:44:32 03Robert Schuster  07hrw/poky-merge/add-rpm-packages-support * rb9fece3136 10openembedded.git/packages/cmake/cmake-native_2.6.3.bb: cmake-native 2.6.3: New recipe. Mar 19 14:44:36 03Khem Raj  07hrw/poky-merge/add-rpm-packages-support * r31abc4f834 10openembedded.git/classes/autotools.bbclass: (log message trimmed) Mar 19 14:44:36 autotools.bbclass: Add new pattern to match -L in dependency_libs Mar 19 14:44:37 * Current pattern match does one think wrong consider Mar 19 14:44:39 something like this Mar 19 14:44:41 dependency_libs=' -L/home/kraj/work/oe/build/eglibc/omap5912osk/work/x86_64-linux/ncurses-native-5.4-r17/ncurses-5.4/lib' Mar 19 14:44:44 this pattern when matched gets converted to Mar 19 14:44:52 dependency_libs=' -L/home/kraj/work/oe/build/eglibc/omap5912osk/staging/x86_64-linux/usr/lib/lib' Mar 19 14:44:54 03Frans Meulenbroeks  07hrw/poky-merge/add-rpm-packages-support * r337be70ba5 10openembedded.git/: Merge branch 'org.openembedded.dev' of git@git.openembedded.net:openembedded into org.openembedded.dev Mar 19 14:44:58 03Frans Meulenbroeks  07hrw/poky-merge/add-rpm-packages-support * r5ad2cf5ec3 10openembedded.git/packages/proftpd/files/make.patch: proftpd: added missing make.patch Mar 19 14:45:01 03Koen Kooi  07hrw/poky-merge/add-rpm-packages-support * re49b459065 10openembedded.git/conf/checksums.ini: checksums.ini: merge checksums from mamona Mar 19 14:45:03 shit.. I forgot about that branch ;( Mar 19 14:45:04 03Khem Raj  07hrw/poky-merge/add-rpm-packages-support * r151270f47b 10openembedded.git/: Merge branch 'org.openembedded.dev' of git@git.openembedded.net:openembedded into org.openembedded.dev Mar 19 14:45:13 03Stefan Schmidt  07hrw/poky-merge/add-rpm-packages-support * r1704bef15d 10openembedded.git/conf/checksums.ini: checksums.ini: Add sums for connman packages. (Changed location) Mar 19 14:45:16 03Frans Meulenbroeks  07hrw/poky-merge/add-rpm-packages-support * r971a08f557 10openembedded.git/packages/proftpd/proftpd_1.3.2.bb: proftpd: fixed paths Mar 19 14:45:21 03Frans Meulenbroeks  07hrw/poky-merge/add-rpm-packages-support * r37dbb4f357 10openembedded.git/: Merge branch 'org.openembedded.dev' of git@git.openembedded.net:openembedded into org.openembedded.dev Mar 19 14:45:25 03Stefan Schmidt  07hrw/poky-merge/add-rpm-packages-support * r84eecfff21 10openembedded.git/conf/distro/include/preferred-om-2008-versions.inc: preferred-om-2008-versions.inc: Update more versions to available ones Mar 19 14:45:29 03Stefan Schmidt  07hrw/poky-merge/add-rpm-packages-support * rd704e9b027 10openembedded.git/packages/connman/connman_0.12.bb: connman: Add latest release Mar 19 14:45:36 03Roger Monk  07hrw/poky-merge/add-rpm-packages-support * r45362bc05d 10openembedded.git/packages/dsplink/ti-codec-combos_3.16.bb: Mar 19 14:45:39 ti-codec-combos: Combo Build Cleanup and additional html docs in package Mar 19 14:45:45 * Updated recipe to correct build warnings where combos weren't properly cleaned before building, and hence gave confusing warnings. Mar 19 14:45:48 * Removed some unused paths from XDCPATH Mar 19 14:45:50 * Added top level html doc to install target and qualiTI reports Mar 19 14:45:52 03Khem Raj  07hrw/poky-merge/add-rpm-packages-support * r0c7cbe92da 10openembedded.git/: Merge branch 'org.openembedded.dev' of git@git.openembedded.net:openembedded into org.openembedded.dev Mar 19 14:45:55 03Khem Raj  07hrw/poky-merge/add-rpm-packages-support * rcf14de067d 10openembedded.git/: Merge branch 'org.openembedded.dev' of git@git.openembedded.net:openembedded into org.openembedded.dev Mar 19 14:46:02 03Koen Kooi  07hrw/poky-merge/add-rpm-packages-support * r5422d5ba54 10openembedded.git/conf/machine/dm355-leopard.conf: dm355-leopard: add board file for the dm355 leopard board Mar 19 14:46:11 03Jader H. Silva  07hrw/poky-merge/add-rpm-packages-support * rbf59c40c3e 10openembedded.git/classes/kernel.bbclass: (log message trimmed) Mar 19 14:46:14 Kernel bbclass "too many open files" bugfix Mar 19 14:46:16 tempfile.mkstemp() is used to create a temporary file. This function Mar 19 14:46:18 return a tuple with an OS file descriptor and a filename. Filename is Mar 19 14:46:18 can i have more than one required images? Mar 19 14:46:20 stored in "tmpfile" but descriptor is not stored anywhere, but it is Mar 19 14:46:22 still open because it's only an integer to python so it is not closed at Mar 19 14:46:24 the end of the function. Mar 19 14:46:26 03Michael 'Mickey' Lauer  07hrw/poky-merge/add-rpm-packages-support * r8c1892e90a 10openembedded.git/packages/vala/vala-native_git.bb: Mar 19 14:46:31 (117 lines omitted) Mar 19 14:46:33 03Mike Westerhof  07hrw/poky-merge/add-rpm-packages-support * r8745858efa 10openembedded.git/packages/thttpd/thttpd_2.25b.bb: thttpd: disable PARALLEL_MAKE due to build failures Mar 19 14:46:38 03Richard Purdie  07org.openembedded.dev * rccf1cf94c2 10openembedded.git/recipes/python/ (6 files): Mar 19 14:46:41 Add python-urlgrabber, python-iniparser, yum-metadata-parser and native variants (from Poky) Mar 19 14:46:45 OE changes: Mar 19 14:46:47 python-iniparse-native: use distutils_stage_all Mar 19 14:46:49 python-urlgrabber-native: use distutils_stage_all Mar 19 14:46:50 ok, now proper meat goes Mar 19 14:46:51 yum-metadata-parser-native: use distutils_stage_all Mar 19 14:46:53 03Michael 'Mickey' Lauer  07hrw/poky-merge/add-rpm-packages-support * r18e898f914 10openembedded.git/conf/distro/minimal.conf: minimal: ship debug apps, if requested Mar 19 14:52:05 morning Mar 19 15:00:13 gm Mar 19 15:07:15 morning Mar 19 15:21:38 good morning Mar 19 15:22:08 does anyone has tried packaged-staging out there? Mar 19 15:22:32 of course Mar 19 15:23:00 angstrom uses it, minimal uses it, kaeilos uses it, openmoko uses it... Mar 19 15:23:04 basically every current OpenEmbedded distro Mar 19 15:23:10 kergoth: so after a packaged-staging, what do I need? only the pstage dir? Mar 19 15:23:21 can I remove the tmp dir? Mar 19 15:23:25 if you want to retain the packages across wipes of tmpdir, just move DEPLOY_DIR_PSTAGE out of tmp Mar 19 15:23:35 DEPLOY_DIR_PSTAGE = "${TOPDIR}/pstage" or somethign Mar 19 15:23:47 kergoth: pstage is outside tmp now Mar 19 15:23:56 I once built Poky with p-s and each recipe from scratch Mar 19 15:23:57 then go ahead and wipe it, thats all you need Mar 19 15:24:08 took some time but I found many missing deps that way Mar 19 15:24:12 but in the pstage dir I don't see ipk in it ?! Mar 19 15:24:22 mpoullet: they're in the pkgs dir in pstage. Mar 19 15:24:24 look closer. Mar 19 15:24:30 find is your firend Mar 19 15:25:06 at the moment I've only a pstaging_lists there, but the build is running right now... Mar 19 15:25:20 :-) Mar 19 15:25:23 well, packages are emitted as each package is built Mar 19 15:25:29 technically all you need is the pkgs dir, not even the full pstage dir Mar 19 15:26:01 kergoth: thanks for the infos Mar 19 15:26:18 * mpoullet waits for the pkgs dir to pop up Mar 19 15:31:28 anyone looked at making devshell smarter about pstage use? Mar 19 15:31:43 if the package was built out of a pstage package, then the ${S} dir is empty, and devshell is pretty useless :) Mar 19 15:34:24 a lot of packages seem to have been built yet but no ipk in the pstage dir... Mar 19 15:34:47 only the stamp-cache-staging file goes bigger and bigger... Mar 19 15:38:54 is there something like DEPLOY_DIR_PSTAGE for the tmp/download dir ? Mar 19 15:39:35 DL_DIR you mean? Mar 19 15:40:18 hrw: I think so, yes :)! Mar 19 15:41:11 what do you mean? Mar 19 15:41:20 DL_DIR can be moved out of tmpdir just fine, feel free Mar 19 15:42:04 yes because I'm tired to mv it out of tmp, then del tmp, then mv it back in there... Mar 19 15:42:10 ? Mar 19 15:42:24 just put DL_DIR = "/path/to/wherever/you/want/your/sources" in your local.conf Mar 19 15:42:28 not exactly difficult Mar 19 15:42:45 has anyone messed with the dolt libtool stuff in OpenEmbedded? Mar 19 15:42:49 hey hrw, what's the java stuff you're doing for? android by chance? Mar 19 15:43:09 is bitbake in git yet? Mar 19 15:43:26 Tartarus: no, it is Bug Labs and their stuff Mar 19 15:43:39 kergoth: Poky has dolt in libtool Mar 19 15:43:49 hrw: ah, cool Mar 19 15:43:55 * kergoth grabs a poky tree to poke around Mar 19 15:44:06 k Mar 19 15:44:20 Tartarus: rpm stuff is merged - do a test Mar 19 15:46:01 anyone happen to recall why we set installed="no" in our .la files in staging? Mar 19 15:49:22 hm Mar 19 15:49:29 hrw, k Mar 19 15:53:24 kergoth: not really, there's was a huge discussion one or two years ago about that Mar 19 15:53:36 and the outcome was that it was better to leave em uninstalled Mar 19 15:53:36 hmm, k Mar 19 15:53:39 but i can't recall the details Mar 19 15:53:44 the archives will tell ;) Mar 19 15:54:10 given the prefix is staging for native packages, they really are installed there. i could see the argument for non-native, since their prefix isn't overridden, but native/cross, not so much Mar 19 15:55:46 i even don't know whether this discussion was before the new staging layout or after Mar 19 15:55:57 feel free to reopen it Mar 19 15:56:00 i'm sure someone will remember Mar 19 15:56:59 k Mar 19 16:03:11 the discussion was certainly on the mailing list Mar 19 16:03:14 as I remeber it Mar 19 16:03:22 but it was well before the new staging layout Mar 19 16:03:28 so is long lost in my mind Mar 19 16:03:30 might be worth revisiting then Mar 19 16:03:37 also, why does gmane's search suck so much? Mar 19 16:04:01 no idea, its annoying Mar 19 16:04:33 it's an odd oversight, considering how good their browse interface is Mar 19 16:04:37 (though not without its quirks) Mar 19 16:04:52 hrw, same old import sqlite error for me Mar 19 16:04:58 i'll try a totally clean build now Mar 19 16:05:41 Tartarus: I got images under ubuntu 7.10 Mar 19 16:06:48 ok Mar 19 16:06:53 i'll hope nuking the tmp dir helps then Mar 19 16:27:49 03Koen Kooi  07org.openembedded.dev * r332db9ccd2 10openembedded.git/recipes/linux/linux-omap-2.6.28/add-resizer-driver.patch: linux-omap 2.6.28: update resizer patch Mar 19 17:33:15 kergoth: here is the discussion about .la stuff for installed=yes/no Mar 19 17:33:17 http://thread.gmane.org/gmane.comp.handhelds.openembedded/7769/focus=8020 Mar 19 17:33:30 oh, thanks. you didn't need to take the time to find it, i would've found it eventually :) Mar 19 17:33:58 I was looking for my own sake Mar 19 17:34:12 ah :) Mar 19 17:56:19 hrw|gone: heh, i'm thinking having libtool automatically use dolt across the board probably isn't the best way to go. there are packages that need the real libtool Mar 19 17:56:20 hrm Mar 19 18:13:02 03Theodore A. Roth  07org.openembedded.dev * r7c62ab8358 10openembedded.git/conf/local.conf.sample: local.conf.sample: Replace packages/ with recipes/. Mar 19 18:13:44 awww.. CROSS COMPILE Badness: /usr/include in INCLUDEPATH: /usr/include/directfb.. any ideas? Mar 19 18:14:48 toggles_w: hmmm this could be due to missing include paths. Check the build logs Mar 19 18:15:44 khem: i think it's because the makefile includes "-I/usr/include/directfb" instead of the x-compile headers but im not sure how to fix it Mar 19 18:21:01 toggles_w: makefile assumes native compilation you need to add ${STAGING_INCDIR} Mar 19 18:21:12 toggles_w: something like CFLAGS += " -I${STAGING_INCDIR}" Mar 19 18:21:17 in recipe Mar 19 18:21:38 khem: thanks mate, I will try it Mar 19 18:25:25 still there.. -fvisibility=hidden -I/usr/include/directfb .. hmm Mar 19 18:26:05 Do you get a new path added too Mar 19 18:26:47 actually its missing files in your staging which it ends up pulling from /usr/include/directfb Mar 19 18:27:16 you should remove this -I/usr/include/directfb from makefile and see what all files does it want Mar 19 18:27:53 khem: this is going to involve a patch isnt it Mar 19 18:28:19 yeah or you can also hack it inside your build Mar 19 18:28:30 hmm.. Mar 19 18:28:34 khem: thanks Mar 19 18:28:59 the run.* scripts in the build tree will help you set up the env Mar 19 18:39:46 khem: is there any way to use LIBRARY_PATH for cross compiler? Mar 19 18:40:15 denix0: for linking ? Mar 19 18:40:20 yep Mar 19 18:41:05 need to specify extra "-L" via the env variable... Mar 19 18:41:14 it does not use it Mar 19 18:41:20 yes thats why Mar 19 18:41:21 but LIBRARY_PATH only works for native Mar 19 18:41:51 is there any other env var? Mar 19 18:41:58 thats why LIBRARY_PATH is avoided in cross why would one want to use ti Mar 19 18:42:36 I think adding -L to LDFLAGS would be simpler Mar 19 18:42:42 simple - SDK Mar 19 18:43:35 LDFLAGS usually overwritten by Makefiles Mar 19 18:44:14 thats too bad but you can confgure with the LDFLAGS you want Mar 19 18:44:26 so the expectation is to use SDK and its setup script to crosscompile out of the tree w/o changing Makefiles Mar 19 18:44:34 specify LD_FLAGS = "mine" ./configure Mar 19 18:44:43 please update this: thx Mar 19 18:44:44 NOTE: The MD5Sums did not match. Wanted: '0499a958a561781b125c740f18061ea8' and Got: 'a6dd36a02fdd009bb4b9e290ebe82e99' Mar 19 18:44:47 NOTE: Task failed: Checksum of 'ftp://ftp.gnu.org/gnu/m4/m4-1.4.12.tar.gz' failed Mar 19 18:45:05 hmmm Mar 19 18:45:39 khem: there are hundreds of manual ways - the idea is to do it automatically when someone uses the SDK Mar 19 18:46:39 denix0: LIBRARY_PATH is meant for native IMO define a new var like SDK_LIBRARY_PATH Mar 19 18:46:52 and beat gcc/linker to accept it would work Mar 19 18:47:19 khem: :) ah, ok, hack the gcc/ld... Mar 19 18:47:45 I was hoping there is way around it w/o touching compiler/linker Mar 19 18:47:52 CruX|: when a checksum changes, the file changed out from under us. it's a good idea to determine exactly what changed, like if it was compromised. blindly updating to the new checksum is not good policy Mar 19 18:49:07 it would be nice to be able to source /usr/local/angstrom/arm/environment-setup and then just simply call make... Mar 19 18:53:40 kergoth: m4-native_1.4.12.bb was added this monday and the tar file has not changed since then Mar 19 18:53:54 so I wonder why this mismatch Mar 19 18:54:11 thats odd Mar 19 18:55:37 +SRC_URI = "${GNU_MIRROR}/m4/m4-${PV}.tar.gz" Mar 19 18:55:41 this could be Mar 19 18:57:39 * kergoth thinks about ways to control side effects in OpenEmbedded tasks & builds Mar 19 18:58:56 03Michael 'Mickey' Lauer  07org.openembedded.dev * r47bb0f5ca3 10openembedded.git/recipes/linux/linux-msm7xxxx_git.bb: linux-msm7xxx: bump htcraphael SRCREV to a version that boots Mar 19 19:14:30 I am trying to get smarter at it. so how to I discover what command to run to update the Packages.gz from my ipk repository. I know it is done using bitbake and I think the method that gets called is called package_update_index_ipk but how do I get is called? Mar 19 19:15:02 I asked this same question about two days ago on irc :( Mar 19 19:16:44 What are you trying to do, in the end? Mar 19 19:16:49 Have a feed your machine can use? Mar 19 19:17:18 If so, look in contrib/angstrom/build-feeds.sh Mar 19 19:18:31 the answer is "bitbake package-index" , but still Mar 19 19:19:16 Tartarus: at first just to get a up-to-date index but on the long run understand how to help myself with such problems :p Mar 19 19:19:41 keesj: ipkg-make-index --help Mar 19 19:21:30 hrw|gone, still got that same problem on 8.10 Mar 19 19:21:56 Can you give me the package list from your working 7.10 setup? Mar 19 19:23:21 03Marcin Juszkiewicz  07org.openembedded.dev * raed057419c 10openembedded.git/recipes/rpm/ (10 files in 2 dirs): (log message trimmed) Mar 19 19:23:21 rpm/rpm-native: Add recipe for 4.4.2.3 (from Poky) Mar 19 19:23:21 Squashed set of changes from Poky: Mar 19 19:23:21 svn r5069 - rpm/rpm-native: Add recipe for 4.4.2.3 Mar 19 19:23:21 svn r5085 - rpm: Patch out some Makefile strangeness causing build failures Mar 19 19:23:23 svn r5093 - rpm: Enable python extensions and improve cross compiling patches Mar 19 19:23:27 svn r5103 - rpm: Tweak patches to remove popt problem Mar 19 19:23:31 03Richard Purdie  07org.openembedded.dev * rcabcbfe980 10openembedded.git/recipes/beecrypt/ (beecrypt_3.1.0.bb files/x64fix.patch): Mar 19 19:23:34 beecrypt: get patch from Poky to fix building on 64bit machines Mar 19 19:23:36 git-svn-id: https://svn.o-hand.com/repos/poky@5068 311d38ba-8fff-0310-9ca6-ca027cbcb966 Mar 19 19:23:37 Tartarus: it was same as yours + few "not related to build" packages Mar 19 19:23:38 03Richard Purdie  07org.openembedded.dev * r7fe23916b9 10openembedded.git/recipes/python/ (6 files): Mar 19 19:23:43 Add python-urlgrabber, python-iniparser, yum-metadata-parser and native variants (from Poky) Mar 19 19:23:45 OE changes: Mar 19 19:23:47 python-iniparse-native: use distutils_stage_all Mar 19 19:23:49 python-urlgrabber-native: use distutils_stage_all Mar 19 19:23:51 yum-metadata-parser-native: use distutils_stage_all Mar 19 19:23:53 03Richard Purdie  07org.openembedded.dev * r5089082042 10openembedded.git/classes/rpm_core.bbclass: Drop pointless rpm_core class. (from Poky) Mar 19 19:24:00 03Richard Purdie  07org.openembedded.dev * r7732204f1e 10openembedded.git/ (11 files in 4 dirs): (log message trimmed) Mar 19 19:24:03 Add yum and createrepo (from Poky) Mar 19 19:24:05 Squashed set of changes from Poky: Mar 19 19:24:07 svn r5098 - Add yum-native and createrepo Mar 19 19:24:09 svn r5108 - createrepo-native: Add fix for empty data fields when creating repositories Mar 19 19:24:15 svn r5109 - yum-native: Add patch to fix install paths Mar 19 19:24:17 svn r5110 - yum-native: Add some hacks to make yum work better in the cross build case Mar 19 19:24:19 03Richard Purdie  07org.openembedded.dev * r3166def128 10openembedded.git/classes/rootfs_rpm.bbclass: (log message trimmed) Mar 19 19:24:22 rootfs_rpm.bbclass: added from Poky Mar 19 19:24:24 Squashed changes from Poky: Mar 19 19:24:26 svn r5101 - Add rootfs_rpm.bbclass, a start at image generation from rpms. Its Mar 19 19:24:32 not expected to be entirely functional yet. Mar 19 19:24:34 svn r5112 - rootfs_rpm.bbclass: Remove the feed directory before packaging the image Mar 19 19:24:36 svn r5134 - rootfs_rpm.bbclass: Add basic postinstall handling during image creation Mar 19 19:24:38 03Richard Purdie  07org.openembedded.dev * r1af5030de0 10openembedded.git/classes/package.bbclass: Mar 19 19:24:41 package.bbclass: Write out extra useful data into the pkgdata files (from Poky) Mar 19 19:24:47 git-svn-id: https://svn.o-hand.com/repos/poky/trunk@5388 311d38ba-8fff-0310-9ca6-ca027cbcb966 Mar 19 19:24:49 03Marcin Juszkiewicz  07org.openembedded.dev * r6f854d71c3 10openembedded.git/recipes/gtk+/gtk+_2.14.2.bb: Mar 19 19:24:54 gtk+: make 2.14.2 default version instead of 2.10.14 Mar 19 19:24:56 Ångström uses this version and many components of metadata depends Mar 19 19:25:00 on at least 2.12. Mar 19 19:25:02 (21 lines omitted) Mar 19 19:31:50 re Mar 19 19:32:28 I've a problem with packaged-staging build, it almost works, but I get an error near the end of the second build, any idead? http://pastebin.com/d651207a3 Mar 19 20:12:41 it seems that my problem is related to the opkg-native package Mar 19 20:13:09 it looks at opkg-cl and doesn't find it Mar 19 20:14:01 but if I copy it first and then run bitbake, it seems that it removes it and then looks at it, a little bit weird... does anyone see this before? Mar 19 20:14:07 i ran into that once, it tried to use the staging package of opkg-cl, and opkg-cl wasn't installd yet, so it couldn't e used to install itself :) Mar 19 20:15:07 kergoth: huuummm, is there a known fix ? :-/ Mar 19 20:15:24 not that i know of. remove the opkg-cl package to force it to be rebuilt? :) Mar 19 20:15:30 i dunno, haven't had time to investigate Mar 19 20:15:39 its just on my very long list of annoyances i've hit in OpenEmbedded Mar 19 20:16:43 kergoth: ok, thanks, that's sad, I wanted to demonstrate packaged-staging at work tomorrow, but I think I can't... Mar 19 20:16:46 hm, after the packages -> recipes move I am left with 8 .bb files in my packages/perl dir that I am pretty sure I have checked in, but they do not appear in my recipes/perl dir Mar 19 20:16:56 how come/what to do? Mar 19 20:17:17 well, if you added them, so they weren't upstream whent he move happened, of course they wouldnt move :) Mar 19 20:17:26 directories dont move, directories dont even exist as entities in the repository Mar 19 20:17:27 _files_ move Mar 19 20:17:32 do a git mv of those into the new area Mar 19 20:17:51 then recommit or what ? Mar 19 20:18:19 i thought they were there but cia.vc is not responding and my git knowledge is too limited to get that from the log Mar 19 20:18:21 unless you plan on leaving it in an uncommitted pile in your local repository Mar 19 20:18:25 ah Mar 19 20:18:26 * kergoth shrugs Mar 19 20:18:42 i thought i committed so want to avoid creating a problem, that is why i ask Mar 19 20:18:45 03Frans Meulenbroeks  07org.openembedded.dev * r547b54b6b8 10openembedded.git/packages/perl/libxml-writer-perl_0.606.bb: libxml-writer-perl: fixed HOMEPAGE and DEPENDS Mar 19 20:19:01 hmmm Mar 19 20:19:20 apparently they were committed but not pushed, at least that one Mar 19 20:19:34 did i now mess up ? Mar 19 20:19:55 if it wasn't pushed, you can always modify the commit to put them in the other location Mar 19 20:20:10 or commit the git mv of them, then merge the commits into one with rebase -i Mar 19 20:20:13 or.. :) Mar 19 20:20:37 at least teh one cia2 gave above is worng as it is still in recipes Mar 19 20:21:18 s/recipes/packages/ Mar 19 20:23:24 kergoth: I've removed opkg-cl from the pstage area and it seems to work... Mar 19 20:23:32 03Frans Meulenbroeks  07org.openembedded.dev * r1883d47878 10openembedded.git/ (conf/checksums.ini packages/perl/libxml-writer-perl_0.606.bb): libxml-writer-perl_0.606: created: Perl module for writing XML documents Mar 19 20:25:20 but now I've still an error in the last step do_rootfs touch: cannot touch `/home/mpoullet/oe/tmp/deploy/glibc/ipk/Packages': No such file or directory http://pastebin.com/d403a954c Mar 19 20:25:36 * mpoullet1 trying to add the missing dirs per hand Mar 19 20:26:36 03Frans Meulenbroeks  07org.openembedded.dev * recf17eb568 10openembedded.git/ (conf/checksums.ini packages/perl/libxml-twig-perl_3.33.bb): libxml-twig-perl: created: Perl module for processing huge XML documents in tree mode Mar 19 20:26:50 hm, guess they were not pused and now showing up, these are commits from last sunday :-( Mar 19 20:27:35 there will be some more coming; i'll git mv them to fix it Mar 19 20:31:31 pushed a commit to move those files from pacakges to recipes, cia is slow but it is on its way, hope i didn't create a stale packages dir Mar 19 20:34:24 03Frans Meulenbroeks  07org.openembedded.dev * rf1a4ac9227 10openembedded.git/ (conf/checksums.ini packages/perl/libtermreadkey-perl_2.30.bb): libtermreadkey-perl: created: perl terminal key read functions Mar 19 20:38:31 eFfeM: good evening. we had to fix proftpd to build the _makenames tools for the host instead of for the target. did you have this problem? Mar 19 20:38:43 03Frans Meulenbroeks  07org.openembedded.dev * rcefe445cd5 10openembedded.git/ (2 files in 2 dirs): libfile-slurp-perl created: single call read & write file routines Mar 19 20:39:16 likewise: i'm surprised, I've build the package and installed it on my beagle and tried it Mar 19 20:39:34 maybe i did not test well enough :-/ Mar 19 20:39:37 could you try "file /tmp/work/../proftpd/lib/libcap/_makenames" Mar 19 20:39:47 will do Mar 19 20:41:35 interesting, i found the dir, but there is no exe or obj file (this is with r2) Mar 19 20:42:34 ah, i think i know what is going wrong, one sec Mar 19 20:43:20 eFfeM: we have a fix for it, let me show what we have Mar 19 20:43:34 eFfeM: I just cannot understand how it could work for you Mar 19 20:43:59 * mpoullet1 has still no luck with packaged-staging Mar 19 20:44:05 likewise it worked for me because I have libcap and libcap2 that's why it works; the internal libcap is only needed if there is no libcap dir I guess Mar 19 20:44:27 guess there should be a DEPENDS = "libcap2" or so Mar 19 20:44:32 opkg -test upgrade Mar 19 20:44:43 OT : what's the sheeva plug new community site ? Mar 19 20:44:48 eFfeM: no that's not what I mean; during the build it will *execute* on the build host lib/libcap/_makenames. Mar 19 20:45:08 eFfeM: local to the build dir (it executes ./_makenames to be exact). Mar 19 20:45:32 all the packages are there in the pstage dir but opkg-cl doesn't find them when it does the rootfs stage: http://pastebin.com/dca1ce89 Mar 19 20:45:36 eFfeM: could you show me "file /tmp/work/../proftpd-2-bla/lib/libcap/_makenames" Mar 19 20:45:42 yes, but I guess it only does a make in that dir if libcap is not there Mar 19 20:45:53 likewise there is no exe in that dir, not even object files Mar 19 20:46:02 eFfeM: ah ok, yes, I understand now. Mar 19 20:46:05 can you do a ldd -v /usr/sbin/proftpd | grep libcap on the target Mar 19 20:46:15 you might need an opkg install ldd upfront Mar 19 20:46:27 eFfeM: I didn't deploy yet. Mar 19 20:46:30 i get: libcap.so.2 => /usr/lib/libcap.so.2 (0x40025000) Mar 19 20:46:30 /usr/lib/libcap.so.2: Mar 19 20:46:46 eFfeM: but let me build and see Mar 19 20:47:42 likewise can you add a DEPENDS = "libcap2" to the recipe Mar 19 20:47:52 then test ? Mar 19 20:48:56 effem: no I want the minimal way. I looked at buildroot and they do what I thought was best. Mar 19 20:49:26 likewise: if you have anohter/better solution, fine with me Mar 19 20:49:30 the problem seems that tmp/deploy/glibc/ipk is empty after the second run... Mar 19 20:49:40 but at least i understand why it does not work Mar 19 20:49:44 eFfeM: I would like to propose this: Mar 19 20:49:48 do_compile () { Mar 19 20:49:50 oe_runmake CC="${BUILD_CC}" CFLAGS="" LDFLAGS="" -C lib/libcap _makenames Mar 19 20:49:51 oe_runmake all Mar 19 20:50:08 mpoullet1 surprises me Mar 19 20:50:27 because we only use the _makenames tools (I guess) Mar 19 20:50:41 mpoullet1: oh I had that problem once... let me remember Mar 19 20:50:58 mpoullet1: Are you using TMPDIR in your top level make or something similar? Mar 19 20:51:14 likewise: is it BUILD_CC or HOST_CC; also I am not sure if there is not a macro for host CFLAGS and LDFLAGS Mar 19 20:51:27 likewise: TMPDIR is /home/mpoullet/oe/tmp Mar 19 20:51:38 mpoullet1: where do you set TMPDIR?? Mar 19 20:51:59 TMPDIR is in local.conf Mar 19 20:52:12 likewise: in my local.conf Mar 19 20:52:46 likewise: BUILD_CC is ok Mar 19 20:52:55 but there is also BUILD_CFLAGS Mar 19 20:52:55 mpoullet1: do you have a variable called "TOPDIR" somewhere? Mar 19 20:52:56 * mpoullet1 local.conf here: http://pastebin.com/d7bc1a787 Mar 19 20:53:15 and BUILD_LDFLAGS Mar 19 20:53:27 likewise: guess you need to use those Mar 19 20:53:37 eFfeM: thanks, yes! Mar 19 20:53:50 likewise: I don't think so... Mar 19 20:55:56 mpoullet1: then I don't know what happens... sorry Mar 19 20:56:16 likewise: that's fine, thanks for having a look Mar 19 20:56:59 * mpoullet1 is trying a new second step packaged-staging right now, let's see Mar 19 20:57:02 03Frans Meulenbroeks  07org.openembedded.dev * r35cd9a8643 10openembedded.git/packages/perl/libtermreadkey-perl-native_2.30.bb: libtermreadkey-perl-native: created native too Mar 19 20:57:05 03Frans Meulenbroeks  07org.openembedded.dev * r683b85ef29 10openembedded.git/conf/checksums.ini: checksums: corrected twig checksum (copied one digit too little) Mar 19 20:57:06 03Frans Meulenbroeks  07org.openembedded.dev * r0d52a4446f 10openembedded.git/packages/perl/libfile-slurp-perl-native_9999.13.bb: libfile-slurp-perl-native: added native Mar 19 20:57:07 03Frans Meulenbroeks  07org.openembedded.dev * r27313cf6a9 10openembedded.git/packages/perl/libxml-twig-perl-native_3.33.bb: libxml-twig-perl-native: created native version Mar 19 20:57:09 03Frans Meulenbroeks  07org.openembedded.dev * r729a49bf3d 10openembedded.git/packages/perl/libxml-writer-perl-native_0.606.bb: libxml-writer-perl-native: created native version Mar 19 20:57:12 03Frans Meulenbroeks  07org.openembedded.dev * rdcfe7349b3 10openembedded.git/: Merge branch 'org.openembedded.dev' of git@git.openembedded.net:openembedded into org.openembedded.dev Mar 19 20:57:14 hey, cia woke up Mar 19 20:57:15 03Frans Meulenbroeks  07org.openembedded.dev * rfa1a9e9e96 10openembedded.git/ (conf/checksums.ini packages/xmltv/xmltv_0.5.54.bb): Mar 19 20:57:18 xmltv: created: Functionality related to the XMLTV file format for TV listings Mar 19 20:57:20 note: this is not tested yet; will do so asap, but wanted to have it archived Mar 19 20:57:24 03Frans Meulenbroeks  07org.openembedded.dev * r299069ed99 10openembedded.git/: Merge branch 'org.openembedded.dev' of git@git.openembedded.net:openembedded into org.openembedded.dev Mar 19 20:57:27 03Frans Meulenbroeks  07org.openembedded.dev * r6a21d65d28 10openembedded.git/: Merge branch 'org.openembedded.dev' of git@git.openembedded.net:openembedded into org.openembedded.dev Mar 19 20:58:34 03Frans Meulenbroeks  07org.openembedded.dev * r7ad21b9e93 10openembedded.git/packages/ (9 files in 2 dirs): Mar 19 20:58:34 moved some files from packages to recipes Mar 19 20:58:34 these files were pushed after the packages to recipes move Mar 19 21:00:24 mpoullet1: i do not have anything after your ti tools comment, so no INHERIT += "packaged-staging" Mar 19 21:00:44 but i do have a BB_FILES Mar 19 21:01:02 actually I do have a thing after the ti comments: DL_DIR Mar 19 21:01:38 eFfeM: are you sure because I see it in pastebin: # packaged-staging INHERIT += "packaged-staging" DEPLOY_DIR_PSTAGE = "${OE_HOME}/archives" DL_DIR = "${OE_HOME}/sources" Mar 19 21:01:44 mpoullet1: do you have anything at all in that ipk dir? Mar 19 21:02:07 i have in ipk: all armv7a beagleboard i686 Packages Packages.filelist Packages.gz Packages.stamps Mar 19 21:02:10 after the first long build I've a lot of ipks in the archives dir yes Mar 19 21:02:17 first 4 are dirs Mar 19 21:03:06 mpoullet1: my local.conf: http://filebin.ca/sgszjz/local.conf Mar 19 21:03:24 eFfeM: no I don't have this structure... Mar 19 21:04:06 in my DEPLOY_DIR_PSTAGE I've only 3 dirs: Mar 19 21:04:20 ./angstromglibc-home-mpoullet-oe-tmp-home-mpoullet-oe-tmp ./angstromglibc ./pstaging_lists Mar 19 21:04:32 hmm odd, i'm tempted to suggest to remove all in tmp and do a rebuild, but it takes a very long time Mar 19 21:04:42 and a lot of ipks there Mar 19 21:04:53 all beginning with staging-***.ipk Mar 19 21:05:16 where does DEPLOY_DIR_PSTAGE point to? Mar 19 21:06:06 DEPLOY_DIR_PSTAGE points outside of the tmp, to ~/oe/archives Mar 19 21:06:08 ah, found it: archives Mar 19 21:06:24 the difference is that I do not have that var Mar 19 21:06:42 not sure what these two lines do: Mar 19 21:06:42 DEPLOY_DIR_PSTAGE Mar 19 21:06:45 oops Mar 19 21:06:54 these two Mar 19 21:07:03 INHERIT += "packaged-staging" Mar 19 21:07:03 DEPLOY_DIR_PSTAGE = "${OE_HOME}/archives" Mar 19 21:07:23 well I'm trying to do a packaged-staging build Mar 19 21:07:29 I don't have these, and i have no ipk's who'se names start with staging Mar 19 21:07:36 why do you want that? Mar 19 21:07:47 that means I do a first run Mar 19 21:07:58 ??? Mar 19 21:08:09 then got all the binaries host+target in DEPLOY_DIR_PSTAGE Mar 19 21:08:22 so that I can then remove the tmp dir Mar 19 21:08:26 i just created a local.conf like the one I gave above then did bitbake nano or bitbake console-image Mar 19 21:08:36 and still rebuild the staging area in a few minutes Mar 19 21:08:57 reducing the build time for the next build Mar 19 21:08:58 that is uncharted territory for me, guess you need an oe wiz Mar 19 21:09:16 something like 2 hours the first time Mar 19 21:09:23 yeah, i understand what you are aiming at (and if you know how it works, please tell me, I'd like to do that as well) Mar 19 21:09:28 and only 15 minutes the second time Mar 19 21:09:51 * eFfeM now every once in a while removes tmp just before going to sleep then does a rebuild over night Mar 19 21:10:10 * mpoullet1 too and it sucks Mar 19 21:10:27 i don't do it that often ... Mar 19 21:10:31 so I'd like to try the packaged-staging alternative Mar 19 21:11:48 i fully understand Mar 19 21:11:56 * eFfeM would want that too :-) Mar 19 21:12:21 well a lot of people on this channel say it works... Mar 19 21:12:31 anyone here that uses ppc603e (like lite5200, efika, n1200, storcenter) or any of the other 'powerpc' architectures? It looks like for those arch's angstrom distro uses gcc4.2.4 and I'm curious if anyone has used 4.3.x with success Mar 19 21:12:35 but until now not for me... Mar 19 21:19:30 eFfeM: see e.g. http://pokylinux.org/blog/index.php/2008/07/packaged-staging-and-what-it-gives/ Mar 19 21:20:30 the first line (starting with "Many people" applies to me :p Mar 19 21:21:22 eFfeM: well I've seen this yesterday;-)! Mar 19 21:21:48 so you did your first build and now you removed tmp and want to do a fresh build Mar 19 21:22:03 eFfeM: exactly Mar 19 21:22:19 and it seems to work at 90% Mar 19 21:22:30 it does build stagemanager-native Mar 19 21:22:32 ? Mar 19 21:22:34 only the last steps completely fail Mar 19 21:22:53 what pacakge did you build the first time ? Mar 19 21:23:20 I suspect an issue with opkg native not finding the ipks... Mar 19 21:23:38 I'm doing only an experiment with console-image for BB Mar 19 21:23:52 that seems a good start! Mar 19 21:24:26 during the second build, it does about 2000 tasks in ~10mn Mar 19 21:24:38 none fails ? Mar 19 21:25:24 no, I suppose it simply that the ipk in the staging area is up to date and do mostly nothing, as explained in the blog article Mar 19 21:25:57 but at the end it needs to run opkg native to pick up the ipks and create the rootfs Mar 19 21:26:00 i think you need someone with more knowledge on the build process Mar 19 21:26:04 and that fails... Mar 19 21:27:20 well maybe someone will show up who knows Mar 19 21:27:25 * mpoullet1 new build still fails Mar 19 21:27:31 and here I am Mar 19 21:27:39 anyway, I'll give it a try, but not now, maybe tomorrow night Mar 19 21:27:52 2245 of 2249 tasks ok and bang ! Mar 19 21:28:00 duh Mar 19 21:28:03 heh, hey Russ Mar 19 21:28:11 hey kergoth Mar 19 21:28:23 mpoullet1: i'm doing lots of test builds with pstage, but my repo is too modified to be very comparable to your setup Mar 19 21:28:28 i'll see if i can do a more pristine build Mar 19 21:29:37 kergoth: well I'm using the dev branch and build console-image for the BB, with no modifications at all Mar 19 21:29:57 k, i'll give that a shot Mar 19 21:29:59 need to sign off, now, back tomorrow, gl Mar 19 21:30:15 eFfeM: gb! Mar 19 21:53:09 * Jay7 -> sleep Mar 19 21:59:42 for those who are following my struggles with the packaged-staging method, I've post a new thread on the ml with all the needed infos, bye and thanks to you all! Mar 19 22:19:32 mpoullet: FWIW rebuild from pstage was fully functional months ago...then broke (at least for me). Lately I retried moving pstage out of tmp but as you see there are some issues... Mar 19 22:20:04 I did not insist 'cause I have other issues happening ... Mar 19 22:23:01 kergoth: tha'ts an input for you: building images for Zaurus, I stumble in build errors (packages not found). I thought it could be originated by parallel_make and bb_number_threads but...strangely both failing packages are pulled in as EXTRA_IMAGEDEPENDS += Mar 19 22:23:12 bell ringing? Mar 19 22:23:27 i haven't touched a zaurus in like 5 years Mar 19 22:23:29 don't ask me :) Mar 19 22:23:49 well...not the answer I expected from you Mar 19 22:23:57 it's NOT Zaurus-specific :-) Mar 19 22:24:02 it's OE Mar 19 22:24:38 I solve building the two packages *before* the images Mar 19 22:24:49 but stil... Mar 19 22:28:19 "stumble in build errors" isn't exactly enough to judge from Mar 19 22:28:28 and i haven't run into anything similar Mar 19 22:29:04 sorry..there are several bugs about 'zaurus-updater not found' Mar 19 22:29:17 ah..and days ago this one sneaked in my logs: Mar 19 22:29:25 sh: rm: command not found Mar 19 22:29:45 was reported by others too iirc Mar 19 22:31:47 could it be EXTRA_IMAGEDEPENDS is processed too late? Mar 19 22:32:49 it seems just in image.bbclass Mar 19 22:35:04 kergoth: at least for 'rm' I kept some minimal logs http://rafb.net/p/bSR0D469.html Mar 19 22:35:28 during staging Mar 19 23:04:12 kergoth: if you're still there: http://bugs.openembedded.net/show_bug.cgi?id=4234 Mar 19 23:44:40 my build is getting stuck on freetype do_compile with the error: Mar 19 23:44:43 tmp/work/armv5te-angstrom-linux-gnueabi/freetype-2.3.5-r2/freetype-2.3.5/builds/unix/arm-angstrom-linux-gnueabi-libtool: No such file or directory Mar 19 23:45:16 there is libtool in that directory but not arm-angstrom-linux-gnueabi-libtool Mar 19 23:45:42 is this just a naming problem? how can I get around it? Mar 19 23:52:31 03Robert Schuster  07org.openembedded.dev * ra5efb484c5 10openembedded.git/recipes/classpath/classpath_0.98.bb: classpath 0.98: New recipe. Mar 19 23:52:35 03Robert Schuster  07org.openembedded.dev * r748445f26f 10openembedded.git/recipes/cacao/ (cacao-native_0.99.3.bb cacao-native_0.99.4.bb): cacao-native: Updated to 0.99.4. Mar 19 23:52:36 03Robert Schuster  07org.openembedded.dev * rd7807ce287 10openembedded.git/conf/checksums.ini: checksums.ini: Added classpath 0.98 entry. Mar 19 23:52:45 03Robert Schuster  07org.openembedded.dev * rbb5fa5808d 10openembedded.git/conf/checksums.ini: checksums.ini: Added cacao 0.99.4 entry. Mar 19 23:52:45 03Robert Schuster  07org.openembedded.dev * rcf5fb3e95f 10openembedded.git/recipes/cacao/ (cacao_0.99.3.bb files/cacao-codegen-arm3.patch): cacao 0.99.3: Added patch for PR 99. Mar 19 23:52:47 03Robert Schuster  07org.openembedded.dev * r665ae4e542 10openembedded.git/recipes/classpath/ (5 files in 2 dirs): Mar 19 23:52:47 classpath-native: Clean up Mar 19 23:52:49 - removed 0.96.1, 0.97.2: Unable to guarantee that these work with current VMs Mar 19 23:52:51 - added 0.98 Mar 19 23:52:53 - removed superfluous patches Mar 19 23:52:55 03Robert Schuster  07org.openembedded.dev * ra5bb2d013b 10openembedded.git/conf/distro/chinook-compat.conf: chinook-compat: Updated versions of packages that where removed from OE. Mar 19 23:53:02 03Robert Schuster  07org.openembedded.dev * r6771de5ce1 10openembedded.git/recipes/maemo4/libhildonmime_1.10.0.bb: libhildon-mime: Replaced dependency with gnome-vfs. Mar 19 23:53:05 03Robert Schuster  07org.openembedded.dev * r74740f99be 10openembedded.git/recipes/gnome/libart-lgpl.inc: Mar 19 23:53:08 libart-lgpl.inc: Build fix. Mar 19 23:53:10 - enforce 'mkdir' executable location Mar 19 23:53:14 - add install-native to depends Mar 19 23:53:16 - enforce 'install' executable location Mar 19 23:53:18 03Robert Schuster  07org.openembedded.dev * r52efc66214 10openembedded.git/conf/distro/chinook-compat.conf: chinook-compat.conf: Fix automake-native preference. Mar 19 23:53:21 03Robert Schuster  07org.openembedded.dev * r37a7bc04d0 10openembedded.git/recipes/maemo4/hildon-thumbnail_0.14.bb: hildon-thumbnail 0.14: Fixed dependencies. Mar 19 23:53:26 03Robert Schuster  07org.openembedded.dev * r924b543d96 10openembedded.git/recipes/maemo4/ (libosso/libosso-dont-use-inline.patch libosso_2.13.bb): Mar 19 23:53:29 libosso 2.13: Fixed build Mar 19 23:53:31 - updated libosso-dont-use-inline.patch to really make some Mar 19 23:53:33 formerly hidden functions visible Mar 19 23:53:35 03Robert Schuster  07org.openembedded.dev * rfa0e1045e2 10openembedded.git/recipes/glib-2.0/glib.inc: glib-2.0.inc: Provide a change in packaging for chinook-compat specifically. Mar 19 23:53:42 03Robert Schuster  07org.openembedded.dev * r595952e9bb 10openembedded.git/recipes/cacao/ (cacao_0.99.4.bb cacaoh-native_0.99.4.bb): Mar 19 23:53:45 cacao 0.99.4: New recipe. Mar 19 23:53:47 cacaoh-native 0.99.4: New recipe. Mar 19 23:53:49 03Robert Schuster  07org.openembedded.dev * r35b91db2b1 10openembedded.git/recipes/swt/ (swt-gtk.inc swt3.4-gtk-hildon_3.4.bb): Mar 19 23:53:52 swt3.4-gtk.inc: Depends on classpath explicitly. Mar 19 23:53:56 swt3.4-gtk-hildon 3.4: Increase PR, since it is affected by glib-2.0 change. Mar 19 23:53:58 03Robert Schuster  07org.openembedded.dev * rff058c0300 10openembedded.git/recipes/cacao/cacao-native_0.99.4.bb: cacao-native 0.99.4: Added note about unusability on amd64 hosts. Mar 19 23:54:01 03Robert Schuster  07org.openembedded.dev * r7f17340d80 10openembedded.git/conf/distro/chinook-compat.conf: chinook-compat.conf: Declare preference on automake-native 1.9.6. Mar 19 23:54:08 03Robert Schuster  07org.openembedded.dev * rcb6d67df8f 10openembedded.git/conf/distro/chinook-compat.conf: chinook-compat.conf: Prefer fakeroot-native 1.9.6. Mar 19 23:54:11 03Robert Schuster  07org.openembedded.dev * r3eaf623918 10openembedded.git/recipes/automake/ (automake-native_1.9.6.bb automake_1.9.6.bb): Mar 19 23:54:14 automake 1.9.6: Retrieve recipe from history. Mar 19 23:54:16 automake-native 1.9.6: Retrieve recipe from history. Mar 19 23:54:20 03Robert Schuster  07org.openembedded.dev * r3b2c009653 10openembedded.git/recipes/cacao/cacao-native_0.99.3.bb: cacao-native 0.99.3: Retrieve recipe back. Mar 20 00:04:26 03Robert Schuster  07org.openembedded.dev * r1731b184f1 10openembedded.git/recipes/tasks/task-maemo.bb: task-maemo.bb: Changed osso-gnome-vfs2 to gnome-vfs. Mar 20 00:04:49 03Robert Schuster  07org.openembedded.dev * r0ee4b03ea5 10openembedded.git/recipes/maemo4/ (2 files in 2 dirs): Mar 20 00:04:49 osso-gnome-vfs2: Removed. Mar 20 00:04:49 - its basically the same as gnome-vfs and Mar 20 00:04:49 conflicts with it in staging Mar 20 00:08:40 all the metadata in OE licensed under MIT, correct? Mar 20 00:09:38 denix0: it is. however consider patches for applications/libraries and the programs license Mar 20 00:09:54 s/and/under/ Mar 20 00:11:29 that's clear. I'm more concerned about packages, which don't pull any sources Mar 20 00:12:02 e.g. angstrom-version or initscripts or base-files Mar 20 00:12:34 are those packages under MIT as well? Mar 20 00:14:34 denix0: AFAIK yes, but if in doubt. ask koen. he's the angstrom maintainer Mar 20 00:16:44 fine, I'll ask him. but the other two are not angstrom specific, right? Mar 20 00:18:16 good question. It is possible that they have been taken from another project (however then there should be some kind of credit) Mar 20 00:19:23 base-files.bb says GPL in the LICENSE field... Mar 20 00:19:45 angstrom-version.bb says MIT... Mar 20 00:20:21 I guess I'll use that info for now and ask around Mar 20 00:24:42 03Robert Schuster  07org.openembedded.dev * r1be4426d9d 10openembedded.git/conf/distro/include/preferred-om-2008-versions.inc: preferred-om-2008-versions.inc: Update jamvm preference to 1.5.2. Mar 20 00:34:06 what's PD license? Mar 20 00:34:32 ah, public domain... Mar 20 00:35:01 which is non existent in some areas, like Europe... **** ENDING LOGGING AT Fri Mar 20 03:00:02 2009