**** BEGIN LOGGING AT Wed Dec 16 02:59:57 2009 Dec 16 03:35:40 Hello, Can anyone tell me what the .x at the end of the name of all the ipk packages mean? For instance, libsoup-2.4-1_2.27.91-r1.4_armv7a.ipk: What does the .4 mean? I remember that Koen told me that it's related to glibc and on Angstrom, it's now .5. On my distro, it's still .4 though I have upgraded to openembedded very recently Dec 16 03:36:36 it's revision r1.4. Dec 16 03:36:52 aka PR variable in the recipe Dec 16 03:37:46 but r1 is the package revision which appears as PR = "r1" in the recipe file Dec 16 03:37:58 yes, but then, what's the .4? Where does it come from? Dec 16 03:39:14 run bitbake -e | grep PR= Dec 16 03:40:29 Got it. Thanks! :-) Dec 16 03:40:48 np Dec 16 05:04:08 * kergoth plays with firmware Linux Dec 16 08:12:57 03Martin Jansa  07org.openembedded.dev * r5452b155e1 10openembedded.git/recipes/mesa/ (mesa-dri_7.6.bb mesa-dri_git.bb mesa-xlib_7.6.bb mesa_7.6.bb): mesa: add 7.6 from xorg-7.5 branch, update git version with different SRC_URI for om-gta02 (glamo patches) Dec 16 08:24:10 03Martin Jansa  07org.openembedded.dev * r3e97c5a64c 10openembedded.git/conf/machine/ (om-gta01.conf om-gta02.conf): om-gta0(1|2): set virtual/libgl provider without shr override Dec 16 08:26:55 03Martin Jansa  07org.openembedded.dev * rfdb5cf6eed 10openembedded.git/recipes/xorg-xserver/xorg-xserver-common.inc: xorg-xserver-common.inc: depend on virtual/libgl instead of mesa Dec 16 09:27:47 morning Dec 16 09:28:09 morning XorA Dec 16 09:28:35 thanks for the reminder, will write a TSC announcement Dec 16 09:31:05 yo XorA Dec 16 09:31:36 hi XorA Dec 16 09:33:41 mickey|office: awesome Dec 16 09:41:07 good morning all Dec 16 09:54:13 morning mckoan Dec 16 10:17:51 mickey|office: good morning Dec 16 10:24:43 morning pb! Dec 16 10:25:38 good morning Dec 16 10:29:29 * pb___ working at home today, car still broken Dec 16 10:30:22 bitbake new-car Dec 16 10:31:32 yeah I would like to push recipe bmw_5.3.5.bb Dec 16 10:32:02 heh. sadly I think my problem is with the hardware, not the software. Dec 16 10:32:40 actually, that's not quite true, the original hardware defect is probably fairly minor: it is the software that makes it much more annoying Dec 16 10:41:51 pb___: could be worse, could be a microsft car :-) Dec 16 10:42:20 heh, indeed Dec 16 10:42:44 pb___: or a gnu car, then it would need some free fuel which is gonna be hard to get hold of Dec 16 10:43:19 I guess that would be biodiesel :-) Dec 16 10:45:59 * XorA imagines a BSD car would be the hotrod of cars Dec 16 10:48:03 * pb___ fears the gentoo car Dec 16 10:48:55 would go real fast, but would break randomly due so some buggy over optimised building :-) Dec 16 10:49:28 USE="windows doors racingtires -trunk -radio -sparetire" emerge -av bmw-5.3.5 Dec 16 10:49:28 and half the uses would have -engine in their USEFLAGS because thats the heaviest bit of a car Dec 16 10:50:10 * XorA survived the wonders of Glasgow roads in the car yesterday Dec 16 11:00:36 is it possible to retrieve the current PR variable of a dependency in a BitBake recipe? Dec 16 11:01:16 so if I in recipe A depends on recipe B, and want's to use the PR value of recipe B Dec 16 11:01:23 in recipe A Dec 16 11:05:52 not easily Dec 16 11:06:14 I think you could do it in theory by grubbing around in tmp/pkgdata, but there's no API for it. Dec 16 11:06:28 hmm... Dec 16 11:06:57 BitBake must have this information, as it knows if it needs to rebuild recipe B when it has a PR bump Dec 16 11:07:15 in cases this has happened before recipe B has dumped a file with the version into staging Dec 16 11:14:39 EsbenH: when parsing recipe B, Bitbake does not hold the data of recipe A. Dec 16 11:14:58 EsbenH: it has also not selected what A is going to be Dec 16 11:16:03 but when parsing recipe A, does it know the PV and PR of recipe B? Dec 16 11:19:27 no Dec 16 11:19:33 it is parsing one recipes at a time Dec 16 11:33:26 hello all. I have a zaurus SL-5500 and an SL-5600 i got them to play around with oe and now i need money. any idea how much they are worth? Dec 16 11:33:51 would be willing to make a deal with a dev(s) that could use them for testing and whatnot Dec 16 11:37:43 i'm afraid these models are so old that you won't get anything except shipping costs Dec 16 11:42:19 last I bought a 5500 was about 4 years ago and it was 20 GBP Dec 16 11:42:51 the 3200 I sold 2 years ago reached 80 GBP Dec 16 11:46:06 03Martin Jansa  07org.openembedded.dev * rb0c1ffe3b6 10openembedded.git/conf/checksums.ini: checksums.conf: add checksums for packages added to xorg-7.5 branch Dec 16 11:46:27 bummer. found a few 5500's on amazon going for 150 to 200 Dec 16 11:46:30 no 5600s Dec 16 11:46:47 iirc i paid about 350 for both of them Dec 16 11:46:54 they are cool little machines Dec 16 11:48:30 last time i looked into zaurus' there was a company selling brand new ones for use in a warehouse stock application (using rfids iirc) they were trying to get like 500 or 600 USD i think Dec 16 11:57:15 Hmm, sounds like it might be the wrong direction I am looking in... Dec 16 12:11:13 I made some changes in the tzdata recipe. Instead of copying the default timezone to /etc/localtime I created a link pointing at one. When I build the recipe I get: *** Error: CONTROL/conffiles mentions conffile /etc/localtime which does not exist. I never seen this kind of message before. Dec 16 12:17:41 jovox_: look in tzdata.inc Dec 16 12:17:54 jovox_: I dont think its valid to list a link as a conffile Dec 16 12:18:47 tzdata.inc? I don't have one. At least not in the tzdata dir. Dec 16 12:20:23 nope, no tzdata.inc in the recipes dir Dec 16 12:20:29 tzdata.inc is located in your openembedded dir /recipes/tzdata/ Dec 16 12:20:57 tsjsieb, not in my branch Dec 16 12:21:20 I'm in stable/2009 Dec 16 12:23:16 All I can think of is the line CONFFILES_${PN} = "${sysconfdir}/timezone ${sysconfdir}/localtime" in the tzdata_2007k.bb Dec 16 12:24:23 I ah, removed the ${sysconfdir}/localtime and it works Dec 16 12:25:42 Is possible to convert a debian packet directly to an ipk package? I founded the "usbmount" app, and this app only exist for debian. And I don't found the source code to build a recipe Dec 16 12:29:05 roliveira_: opkg is able to handle debian packages. Dec 16 12:29:43 roliveira_: you might have to force he dependencies if package names or version numbering do not match. Dec 16 12:35:02 florian: Do you know if exist an HAL event Handler to OpenEmbedded? I tried "ivman", but i found some text explain that the current version of ivman doesn't work with the versions of hal Dec 16 12:51:22 roliveira_: There is hal in OE but I do have no idea about ivman. Dec 16 13:03:46 florian: Ivman is application to mount and unmount all device, and if device is unmounted, the ivman tries to mount them. I wrote some udev rules, but it didn't work. I have an big trouble in my machine. I don't know why, but sometimes the thumb drive is unmount suddenly, like an power saving, but i haven't acpi enable Dec 16 14:06:44 Hi everybody Dec 16 14:06:49 Hi Dec 16 14:06:57 how are you ? Dec 16 14:07:06 Fine thx, and you ?? Dec 16 14:07:15 Fine ;) Dec 16 14:13:56 Well, we found a problem with mono 2.4.2.3 ... we compiled with success, we installed with opkg and we get this : "opkg: Cannot create symlink from ../gac/Mono.CompilerServices.SymbolWriter/1.0.5000.0__0738eb9f132ed756/Mono.CompilerServices.SymbolWriter.dll to '../gac/Mono.CompilerServices.SymbolWriter/1.0.5000.0__0738eb9f132ed756/Mono.CompilerServices.SymbolW': No such file or directory " We suppose it's a problem with the path nam Dec 16 14:15:09 it seems that your path are truncated to 100 characters Dec 16 14:15:42 there may be a miss configuration of your opkg ? Dec 16 14:18:04 what kind of configuration file ??? Dec 16 14:22:19 somebody alive ? Dec 16 14:22:25 nope Dec 16 14:22:29 ^^ Dec 16 14:22:36 * XorA has been dead many years Dec 16 14:23:08 c'est dommage Dec 16 14:23:17 c'est fromage! Dec 16 14:23:33 français ?? Dec 16 14:23:40 oui Dec 16 14:23:51 un peu Dec 16 14:24:05 =) Dec 16 14:27:57 Please, someone know this problem with opkg and limitation with the path lenght ? Dec 16 14:33:31 morning Dec 16 14:35:44 morning Dec 16 14:50:39 morning Dec 16 14:51:58 gm Dec 16 14:52:28 hi Marcin Dec 16 14:52:43 firmware Linux is cute. Dec 16 14:55:24 hi, any idea why if I put dbus-x11 on the RDEPENDS of a task for an image, I still don't get dbus-launch on the final image? Dec 16 14:57:01 ddompe: add to IMAGE_INSTALL Dec 16 15:02:01 hrw: thanks for the help Dec 16 15:02:18 is a bit confusing because in some cases you can use RDEPENDS Dec 16 15:02:26 (and many recipes do it that way Dec 16 15:02:52 what are the semantic differences that make RDEPENDS vrs IMAGE_INSTALL works? Dec 16 15:04:13 ddompe: did you rebuild the task? Dec 16 15:04:57 not sure (I did many things), let me retry that Dec 16 15:07:02 ddompe: without the rebuild the task.ipk file wont have the new Depends: Dec 16 15:07:41 XorA: I'm a bit new still, my procedure for rebuild so far is "bitbake mytask -c clean" Dec 16 15:07:44 and rebake... Dec 16 15:07:46 ddompe: how is it confusing? Dec 16 15:08:02 there is some better way? Dec 16 15:08:16 IMAGE_INSTALL is a list of packages that the image uses to populate the filesystem. The RDEPENDS of the image recipe also includes the items in IMAGE_INSTALL, to ensure bitbake gets the build order right. Dec 16 15:08:21 ddompe: thats the way Dec 16 15:08:26 kergoth: well, it is confusing because sometimes doesn't work (but it maybe that I'm failing to rebuild the task) Dec 16 15:09:03 that would do it, yes. RDEPENDS flow into the binary packages, and the latter won't update per changes to the former without bumping PR or rebuilding the task Dec 16 15:09:06 s/task/recipe/ Dec 16 15:09:06 kergoth: cool, my IMAGE_INSTALL only have one task, maybe I'm just failing to build the task Dec 16 15:09:12 * kergoth nods Dec 16 15:14:25 *ddompe is rebuiling his recipes Dec 16 15:27:14 bon allez fini la branlette Dec 16 15:28:53 çà tue ce chan Dec 16 15:29:11 coin coin coin Dec 16 15:29:33 craquage Dec 16 15:29:57 mono coin.exe not found Dec 16 15:36:04 bien Dec 16 15:36:19 de toutes façon, vu qu'il n'y a personne je donne la solution en français Dec 16 15:36:30 pour faire sur winform sur une board en mono et tout Dec 16 15:39:54 il faut se cross compiler à la mano libgdiplus et tout, on l'envois sur la board, et ça ne marche tj pas Dec 16 15:39:54 même avec le dllmap Dec 16 15:48:11 03Marcin Juszkiewicz  07org.openembedded.dev * rbedc7cb2e8 10openembedded.git/ (classes/java.bbclass conf/bitbake.conf): (log message trimmed) Dec 16 15:48:11 java.bbclass: stage all JAR files outside of ARCH related directories Dec 16 15:48:11 When Java things are build for more then one architecture strange things Dec 16 15:48:11 happen: Dec 16 15:48:11 - first architecture builds fine Dec 16 15:48:15 - next architectures fails to build Dec 16 15:48:17 Thats due to lack of JAR files as they were built as PACKAGE_ARCH="all" Dec 16 16:06:22 hi, just to tell(no time to fix it) that there are staging issues for nfs scripts: Dec 16 16:06:43 http://pastebin.com/m6baef257 Dec 16 17:10:11 hello Dec 16 17:10:18 anybody there Dec 16 17:10:20 ? Dec 16 17:10:29 I have a quick question Dec 16 17:11:46 yes Dec 16 17:12:12 I'm building Angstrom with openembedded for a Beagleboard Dec 16 17:12:26 good choice Dec 16 17:12:48 everything works fine but I would like to customize files like /etc/hosts, is it possible with bitbake or do I have to do it after the image is created ? Dec 16 17:13:16 marc252: you can alter them in the metadata, recipes such as basefiles and netbase Dec 16 17:13:46 marc252: basically make a custom branch and modify to hearts delite Dec 16 17:13:49 delight Dec 16 17:14:34 * XorA obviously listening to too much early 90s techno Dec 16 17:14:35 ok XorA, let me take a quick look because I think I allready tried that Dec 16 17:15:09 marc252: youll ever thave to bump PR in recipe or -c clean it to make it work Dec 16 17:15:38 XorA I'm looking at openembedded/recipes/meta but I can't find any base or simmilar Dec 16 17:16:37 recipes/base-files and recipes/netbade Dec 16 17:16:50 ok, I'm kind of dumb :-) Dec 16 17:16:55 opkg search /etc/hosts on the target to see what it belongs so Dec 16 17:17:15 great!! I didn't realize that!! Dec 16 17:17:58 bye Dec 16 17:20:04 Thanks XorA, that works beautifully!! Dec 16 17:21:28 now, how can I do the same thing to create a bunch of different images without having to make the same thing over and over. For example imagine I want to install 30 different beagleboards and each one with a different /etc/hostname, can I pass a cmd line argument to bitbake to change that file? Dec 16 17:23:17 marc252: erk Dec 16 17:23:38 sorry? Dec 16 17:24:20 marc252: hmmm, I think you would need a ROOTFS_POSTPROCESS with reference to an environment variable Dec 16 17:25:12 ROOTFS_POSTPROCESS_COMMAND += "echo ${OE_HOSTNAME} >${IMAGE_ROOTFS}/etc/hostname;" Dec 16 17:25:26 BB_EXTRA_WHITE += "OE_HOSTNAME" Dec 16 17:25:29 for local.conf Dec 16 17:25:44 then OE_HOSTNAME="blah" bitbake console-image Dec 16 17:25:51 this is untested Dec 16 17:26:15 ok, I'll give it a try Dec 16 17:26:51 BB_ENV_EXTRAWHITE is actually the variable name Dec 16 17:27:09 should be in the manual somewhere Dec 16 17:27:58 hello! Dec 16 17:28:20 anyway its time for the to exchange some alcohol tokens Dec 16 17:29:13 ? Dec 16 17:31:19 beer Dec 16 17:31:40 * XorA|gone doesnt drink much beer Dec 16 17:31:55 or other substitute of your choice Dec 16 17:31:59 but I have many alcohol tokens to exchange Dec 16 17:34:01 ok, I'll stick to coffee meanwhile :-) Dec 16 17:52:29 bye Dec 16 18:10:30 yet another strange problem. This is from my recipe: ln -s /data/dc/work ${D}/qmdc/dc/work and ln -s /data/dc/work ${D}${sysconfdir}/network.config. When building the recipe I get the message "the following files were installed but not shipped in any package: work". The first link (/etc/network.config) is being installed but not the first. I don't see any diffrence between them? Dec 16 18:14:02 Ah! I needed to add it to FILES Dec 16 18:33:57 Anyone may help me, please? I have a trouble with udev, I installed udev in the distro, but when i restart the machine the udev doesn't start automatically, because doesn't have any script in the /etc/rc.d folders. To solve that i have to run the command update.rc. But I have to do that automatically when the distro was builded. How can I do that? Dec 16 18:38:31 03Chris Larson  07org.openembedded.dev * r05019cad32 10openembedded.git/recipes/coreutils/ (10 files): (log message trimmed) Dec 16 18:38:31 coreutils: don't patch the autotools files in the native recipes Dec 16 18:38:31 If we patch autotools files in the native recipes and build them, the make Dec 16 18:38:31 results in the make rules trying to rerun autoconf, automake, aclocal, etc, Dec 16 18:38:33 and those don't yet exist when coreutils-native is built. So we split out the Dec 16 18:38:35 common target (not native) specific bits so that native recipes aren't Dec 16 18:38:37 including the target recipes, so we can avoid applying those patches for Dec 16 20:31:10 roliveira: you have to enable package management and then it should run post install hooks Dec 16 20:33:06 roliveira: which distro are you using Dec 16 20:33:13 khem: Thanks for the answer, I've already enabled opkg packet manager. Dec 16 20:33:47 khem: The distro is OE, I made my distro.conf like the documentation Dec 16 20:35:37 khem: And I put the flag "udev" in the IMAGE_INSTALL Dec 16 20:35:46 roliveira: I meant what is DISTO set yo Dec 16 20:35:47 to Dec 16 20:37:06 khem: I set the minimal distro and the machine is i486 Dec 16 20:38:36 roliveira: ok that should already have ONLINE_PACKAGE_MANAGEMENT set to full Dec 16 20:40:41 khem: there is this line there IPKG_VARIANT = "opkg-nogpg" Dec 16 20:42:04 roliveira: you can try to change to opkg Dec 16 20:42:13 and see if that helps Dec 16 20:42:39 khem: I will try, thanks Dec 16 21:46:54 03Yuri Bushmelev  07org.openembedded.dev * r4e9de808a7 10openembedded.git/recipes/qemu/ (2 files in 2 dirs): Patch added to fix QEMU 0.10.3 compilation on distros with FORTIFY_SOURCE enabled. Dec 16 21:47:08 03Yuri Bushmelev  07org.openembedded.dev * r04fc3f1a63 10openembedded.git/: Merge branch 'jay7-akita' into org.openembedded.dev Dec 16 21:47:54 I hope I'll not break anything :) Dec 16 21:48:11 my test builds was fine at least Dec 16 21:57:34 hello Dec 16 21:58:54 im new to the whole bitbake/OE system but I thought i read somewhere that it was based off of Portage Dec 16 21:59:10 to the build files support anything like "eclasses" in gentoo/portage? Dec 16 21:59:14 originally it was, i doubt there's anything left of those remnants Dec 16 21:59:17 yes, bbclasses. Dec 16 21:59:20 read the OE manual Dec 16 21:59:22 awesome Dec 16 22:00:04 unlike portage, it doesn't really have USE yet (though its on the todo), handles crosscompilation, doesn't care what your build machine's distro is, and uses a directly parsed file format (which supports python tasks and inline python snippets) Dec 16 22:00:30 portage handles crosscompilation Dec 16 22:00:31 but yeah Dec 16 22:00:43 not well :) Dec 16 22:00:44 i need to setup an OE environment for Palm Pre development Dec 16 22:01:02 when we started, it didn't support it at all, of course. lots of build tools have come quite a long way since then Dec 16 22:01:12 yeah im sure Dec 16 22:01:31 but anyway, im glad is has bbclasses Dec 16 22:01:36 s/is/it/ Dec 16 22:01:50 kergoth: hey FWIW I have generated a toolchain for amd64/x86_64 successfully Dec 16 22:02:00 ah, nice Dec 16 22:02:04 kergoth: but its not multilib yet Dec 16 22:02:57 multilib should be well thought out Dec 16 22:03:10 mips ppc and x86 Dec 16 22:03:12 we left it mostly stock, using /lib & /lib64 rather than /lib32 & /lib. had a few things to fix in some recipe packaging to account for it, force the opkg path, etc Dec 16 22:03:28 mips though.. haven't dared :) Dec 16 22:03:31 mips has 3 multilibs so Romans approach is limited Dec 16 22:03:42 yeah, need a generic solution Dec 16 22:04:02 and ideally we'd be able to do something like bbclassextend to do builds of variants of the recipes for each abi, eventually Dec 16 22:04:06 I think multilib needs to be a machine feature Dec 16 22:04:22 true I was thinking about that today Dec 16 22:04:46 unfortunately you'd need one class per abi, really, as things stand today, can't generate multiple variants from one class yet Dec 16 22:17:44 03Klaus Kurzmann  07org.openembedded.dev * r877f8c1953 10openembedded.git/recipes/tasks/task-shr-feed.bb: Dec 16 22:17:44 task-shr-feed: add obex-data-server Dec 16 22:17:44 Signed-off-by: Klaus Kurzmann Dec 16 23:19:25 damn, that initial parse time is crazy :| Dec 16 23:38:16 kergoth: question on overrides Dec 16 23:38:40 kergoth: I have EXTRA_OECONF_append_x86_64 = .... Dec 16 23:38:55 now will be treat x86_64 as one work for arch Dec 16 23:39:20 yes, it doesn't split them by _ and do it that way, it looks for the overrides at the end with a _ before explicitly Dec 16 23:39:49 ok EXTRA_OECONF_append_x86_64 would only do it for x86_64 then ? Dec 16 23:39:54 yep Dec 16 23:40:24 well, if "64" was in overrides, then itd do it there too, but i don't think "64" ever shows up as an override :) Dec 16 23:40:36 ok Dec 17 01:41:38 03Khem Raj  07org.openembedded.dev * rda29e9aaa0 10openembedded.git/ (5 files in 3 dirs): Dec 17 01:41:38 iptables_1.4.6.bb: Recipe for version 1.4.6. Dec 17 01:41:38 * Switch to INC_PR. Dec 17 01:41:38 * inherit autotools Dec 17 01:41:38 Signed-off-by: Khem Raj Dec 17 01:41:53 03Khem Raj  07org.openembedded.dev * ra68775b072 10openembedded.git/recipes/linux-libc-headers/ (2 files in 2 dirs): Dec 17 01:41:56 linux-libc-headers_2.6.32.bb: New recipe. Dec 17 01:42:00 * Patch to export sa_family_t to userspace. Dec 17 01:42:02 Signed-off-by: Khem Raj Dec 17 02:25:30 kergoth: I am getting unset TARGET_ARCH when I do bitbake -e Dec 17 02:25:41 do you know where this happens ? Dec 17 02:25:43 khem: it's likely flagged as 'unexport' Dec 17 02:25:50 hmm Dec 17 02:25:50 see base.bbclass / bitbake.conf, one of those Dec 17 02:26:04 iirc its done for recipes that flake out when certain vars are set Dec 17 02:26:13 but the thing is, setting it globally is fucking stupid, it should be in the recipes that need them.. Dec 17 02:26:18 heh Dec 17 02:26:19 oh right Dec 17 02:26:39 also, its annoying that it affects bitbake -e at all, that should be fixed.. Dec 17 02:26:51 bitbake -e is for the user, not builds :) Dec 17 02:27:07 I want to use it in gcc recipes to find out what the target is in a python func Dec 17 02:27:22 yeah I was using it for checking Dec 17 02:27:27 you could grep out the unexpanded version, its unlikely there are var refs Dec 17 02:28:05 it dumps them as commented.. I've done that to extract MACHINE before Dec 17 02:29:00 http://khem.pastey.net/130319 Dec 17 02:29:04 this is my function Dec 17 02:29:20 * kergoth finds bitbake -e really useful for shell scripts.. you can eval "$(bitbake -e | grep -Ev '(^MACHINE|DISTRO|WHATEVER)=')", then use the vars in the shell script.. $MACHINE, whatever :) **** ENDING LOGGING AT Thu Dec 17 02:59:58 2009