**** BEGIN LOGGING AT Thu Jun 13 02:59:59 2013 Jun 13 05:46:51 hi i was getting an error quilt-native_0.60.bb, do_configure) failed with exit code '1' Jun 13 05:47:18 i tried to look for it using google and found https://bugzilla.yoctoproject.org/show_bug.cgi?id=969 Jun 13 05:47:19 Bug 969: normal, Medium, 1.1, qing.he, RESOLVED FIXED, [quilt] do_configure fail when building quilt on opensuse 11.3 Jun 13 05:47:29 i was on ubuntu Jun 13 09:27:00 Hi my bitbake has trouble it sais unable to bundle what can the causes be ? Jun 13 09:42:00 hi what could cause "Failure expanding variable create_bundle_files: ShellSyntaxError: expecting here-document name, got '<<'"? Jun 13 10:02:07 condition: $ grep create_bundle_files * -R Jun 13 10:02:07 meta/recipes-core/images/build-appliance-image_8.0.bb:create_bundle_files () { Jun 13 10:02:17 condition: did you change something in that file? Jun 13 10:06:06 no ... it was a merging conflict with a git repo Jun 13 10:06:20 but thx anyway Jun 13 10:09:54 RP: for getting qmake running on target, I had to source /usr/share/qt4/environment-setup Jun 13 10:10:50 Net147: bluelightning would probably know more about that than I do... Jun 13 10:11:05 bluelightning: ^ Jun 13 10:11:10 bluelightning: it's normal right? Jun 13 10:11:20 Net147: yes Jun 13 10:11:47 bluelightning: it's just that I never had to do that for normal native qt development Jun 13 10:14:49 bluelightning: it's not mentioned in the Yocto documentation either for running qmake on target device, only for cross compiling Jun 13 10:15:18 Net147: right, it's not a use case we have covered particularly well; typically we expect people to cross-build Jun 13 10:15:44 the environment setup has been that way in OE for a long time, I'm not sure of the specifics Jun 13 10:16:55 bluelightning: well it's easy enough to workaround. I just create symbolic link to it from /etc/profile.d/qt4-dev.sh Jun 13 10:53:02 Anyone know anything about systemd in poky? More specifically, why is networking.service masked out (in meta/recipes-core/systemd/systemd-compat-units.bb)? What is supposed to setup the network instead? Jun 13 10:54:02 Saur: I am using systemd, but I use connman for networking Jun 13 10:57:32 the networking.service of systemd is not working ... so far i know ... you need to use a seperat network deamon ... look on the archlinux page and scroll down a bit to see there fix Jun 13 10:58:29 Saur: https://wiki.archlinux.org/index.php/Network_Configuration#Manual_connection_at_boot_using_systemd Jun 13 11:02:16 Net147, condition: Thanks. Jun 13 11:03:26 condition: what systemd version was networking.service last tested with? systemd was updated recently to 204 Jun 13 11:06:45 i think it was somthing like 195ish Jun 13 11:07:19 Saur: i think it was somthing like 195ish Jun 13 11:11:26 Saur: well last i know some time ago i had the same problem because i had to sync a start prosess of some programs that need the network for clustering ... i made a service that used the network init script to start the network expicitly befor the programs ... but this is just a work around that will backfire later on ... Jun 13 11:12:05 Hmm, how hard is it to get e.g. connman to work? Maybe that's an easier solution... Jun 13 11:38:42 Saur: are you using static or DHCP? Jun 13 11:40:05 Net147: DHCP currently, but in the end it should be configurable Jun 13 11:42:12 Saur: I use connman because I need to handle cable being disconnected/connected all the time, APIPA, DHCP and static. connman defaults to DHCP but I can configure it to static using connmanctl. Jun 13 11:44:23 Saur: I guess NetworkManager can do something similar as well but I didn't like the NetworkManager configuration interface at the time Jun 13 11:45:58 Nice, connman started and I got an IP address :) Jun 13 11:46:29 Too bad it pulled in python... Jun 13 11:48:02 Saur: python is only used for the connman-tests package I thought Jun 13 11:49:25 Saur: did it actually install Python on the target? Jun 13 11:50:17 Net147: Hmm, doesn't look like it. Just assumed it did as it built it, and the image size increased with ~1 MB... Jun 13 11:53:59 Saur: Python is larger than 1MB Jun 13 11:54:33 Net147: Well, good think it wasn't installed then :) Jun 13 12:22:03 hi everybody Jun 13 12:22:36 hi whats up ? Jun 13 12:29:25 I have a question about source files (SRC_URI) Jun 13 12:29:55 it seems that when I add a shared library in my sources files the library is automatically added to my sysroot. And when I add an another time the same library, the do_fetch part of my recipe doesn't load an other time the library in my new build directory. Is it the real mechanism of the fetch part of my recipes? Jun 13 14:47:04 meh, that python auto dependency stuff proved to have some annoying corner cases, e.g. imports of this-or-that via a try/except block Jun 13 14:49:31 kergoth: I figured it might be a bit painful Jun 13 14:49:39 these sort of things usually are :) Jun 13 14:50:15 kergoth: presumably even when all that is ironed out it will still suffer from the issue of having RDEPENDS on items which haven't been built? (since presumably this will need to run at packaging time) Jun 13 14:50:58 yeah, either the deps will have to live in DEPENDS or RDEPENDS, one of the two, just as with shlibs. not as useful as it might be, but still useful for all hte deps on the individual packages from the python build itself Jun 13 14:51:02 * kergoth shrugs Jun 13 14:51:10 at worst it was an interesting exercise :) Jun 13 14:55:50 * kergoth gets food Jun 13 15:00:22 kergoth: right, it should definitely be helpful Jun 13 15:00:57 as a generic improvement for this kind of thing it would be nice if the system could report where added dependencies aren't going to be satisfied earlier than do_rootfs Jun 13 15:05:16 https://wiki.yoctoproject.org/wiki/Yocto_Project_v1.5_Status#Milestone_1 Jun 13 15:08:30 https://wiki.yoctoproject.org/wiki/Fullpass_Test_Report_for_Yocto_1.5_M1_2013-05-28-2_Build Jun 13 15:16:13 https://wiki.yoctoproject.org/wiki/Performance_Test Jun 13 15:24:06 RP, bluelightning: I'm thinking it would be useful to support foo[bar] syntax in vardeps flags. thoughts? Jun 13 15:24:36 kergoth: that sounds a bit scary to me... Jun 13 15:25:32 kergoth: can you expand on that a bit? Jun 13 15:25:53 we have multiple cases where flags are overloaded as dictionaries, either we should support that case, in which case we need the ability to use vardeps so the checksums know when functions need to be rerun based on the flags changing, or we need to stop overloading them entirely across the board Jun 13 15:25:58 afaict anyway Jun 13 15:26:31 * kergoth ponders Jun 13 15:32:05 kergoth: I know I kind of hit this when I was thinking about how we could possibly handle an extension to PACKAGECONFIG to support a choice for a config option as opposed to just on/off Jun 13 15:32:10 kergoth: certainly something to think about but it needs some careful thought Jun 13 15:32:42 kergoth: this is the DISTRO_FEATURES type use case? Jun 13 18:30:28 RP: https://gist.github.com/kergoth/5776139 is an example of why I think we need the aforementioned ability to add var flags to vardeps Jun 13 18:31:39 RP: not a priority, but it does seem like a gap that should be filled Jun 13 18:43:09 Hi can someone help me get ff7 cross-compiler and native-compiler for my Yocto environment? Jun 13 18:49:46 Hi does anyone know why there isn't a cross-compiler with Yocto? (arm-poky-linux-gnueabi-f77) Jun 13 18:54:52 see LANGUAGES ?= "c,c++${FORTRAN}${JAVA}" in gcc-configure-common.inc. Jun 13 18:55:01 gcc-4.7 and gcc-4.8 .inc set FORTRAN = "" for some reason Jun 13 18:55:14 the default value of FORTRAN in gcc-configure-common is: FORTRAN ?= ",f77" Jun 13 18:59:49 thanks Jun 13 18:59:55 i'll give that a try Jun 13 19:00:57 not sure what you intend to try, but as i just said, gcc-4.7 and gcc-4.8 set FORTRAN="". you'd havec to modify them, or set FORTRAN with an override, otherwise the recipe will blow away any value of FORTRAN you set in a .conf, or you'd have to try a different, likely unsupported, gcc version Jun 13 19:01:22 presumably there's a reason it was explicitly disabled for those compilers, but I don't know what that reason would be Jun 13 19:07:26 Thanks I'll try to force it to ",fortran" Jun 13 19:16:32 cool, good luck Jun 13 19:32:07 @kergoth, thanks for your help. I have arm-poky-linux-gnueabi-gfortran binary now Jun 13 19:32:17 I need to do the same for the native compiler Jun 13 19:32:47 cool Jun 13 19:32:49 np Jun 13 19:35:59 darn it doesn't work for native gcc Jun 13 19:36:05 i get this bitbake error: Jun 13 19:36:06 ERROR: This autoconf log indicates errors, it looked at host include and/or library paths while determining system capabilities. Jun 13 19:36:15 Rerun configure task after fixing this. The path was '/home/appusr/poky/build/tmp/work/armv7a-vfp-neon-poky-linux-gnueabi/libtool-cross-2.4.2-r5.1/libtool-2.4.2' Jun 13 23:49:21 goddamnit bitbake. now it's saying 'd' isn't defined in a ${@} block. what? Jun 13 23:52:17 I'm guessing in sstate / siggen / vardep evaluation context not all the context setup is complete for the evaluation or something Jun 13 23:52:21 grr Jun 13 23:54:22 I've got my system up and running, and I specified rpm packages, but I don't have the rpm program itself. Should this have been compiled somewhere that I'm just not seeing? Jun 13 23:55:36 jayemar: yes, if you specified PACKAGE_CLASSES in your local.conf as "package_rpm", then bitbake will build the rpm binaries Jun 13 23:56:49 @sgw I have the binaries, but not the rpm 'program' itself Jun 13 23:57:42 I see this (rpm-5.4.9-0.20120508.src.rpm) in my downloads folder which makes me think I could compile it from source, but it seems weird that that wouldn't have been done as part of the build process Jun 14 00:00:11 jayemar: right, it would have downloaded and built it for your host /native machine and for the target if you do package management on the target. Jun 14 00:06:33 sgw_: hmm. I've got 'PACKAGE_CLASSES ?= "package_rpm"' in my local.conf file, but I can't find the rpm binary in the build products. Jun 14 00:07:12 take a look in your tmp/sysroots/x86_64-linux/usr/bin dir Jun 14 00:15:49 there is an 'rpm' file there, but it references rpm.real which is an x86 binary; I cross-compiled for ARM, so I'm looking for an ARM rpm binary Jun 14 00:16:25 jayemar: Ah, sorry mis-understood, thought you where talking about the host side. Jun 14 00:16:51 jayemar: what was your target image that you created? Jun 14 00:18:27 sgw_: no problem, I appreciate the help. I created an ARMv7 image for the BeagleBone Jun 14 00:18:29 jayemar: it's packaged up and installed in the target image if you have package-managment IMAGE_FEATURE Jun 14 00:20:40 the target rpm will be built in the tmp/work/armv7.../rpm/5.4.9/ Jun 14 00:20:43 is IMAGE_FEATURE a parameter in the local.conf file? Because I didn't include that there Jun 14 00:21:04 jayemar: what image are you build for the beaglebone? Jun 14 00:22:31 sgw_: I'm not sure I understand what you mean by what image...? Jun 14 00:22:52 what's your final target? Jun 14 00:24:57 the BeagleBone, an ARM am3359 Cortex A8 board **** ENDING LOGGING AT Fri Jun 14 02:59:58 2013