**** BEGIN LOGGING AT Thu Nov 22 02:59:58 2012 Nov 22 08:36:45 good morning Nov 22 09:09:55 morning Nov 22 09:10:16 mornig all, hi afournier Nov 22 09:15:14 hi silvio Nov 22 09:49:13 morning all Nov 22 10:04:43 morning bluelightning Nov 22 10:11:25 someone seeing php-native and evolution-data-server failing since yesterday? Nov 22 10:20:30 JaMa: any suggestions on how to handle the cron job in your suggested solution? Nov 22 10:20:37 (for ntp) Nov 22 10:21:44 bluelightning: maybe ntpdate shell wrapper sourcing /etc/default/ntp and showing that echo when servers not defined? Nov 22 10:22:04 bluelightning: same wrapper could be used from .service/cron as well as manually from cmdline Nov 22 10:22:49 yeah that would be the best way, but where to put it? Nov 22 10:23:37  Nov 22 10:23:42 is there shared package by both ntpd as well as ntpdate? for /etc/default/ntp ? Nov 22 10:23:54 wrapper could be in ntpdate package I think Nov 22 10:24:23 JaMa: there isn't unfortunately, but this is only needed for ntpdate - ntpd has its own config file which lists the servers Nov 22 10:24:41 by "where to put it" I mean where on the filesystem... Nov 22 10:24:54 ah maybe new ntpdate-config? Nov 22 10:25:07 ah /etc/default? it's used in some scripts already Nov 22 10:25:21 sudo wireshark Nov 22 10:25:26 oops Nov 22 10:26:24 JaMa: /etc/default/ntpdate for the script that sets the variable, but I don't think that's the right place for a script that performs any action Nov 22 10:26:43 ah ok, now I understand you :) Nov 22 10:27:04 just ${bindir}? Nov 22 10:27:43 seems like there ought to be somewhere else, but I can't think of any more suitable location... Nov 22 10:28:21 maybe libexec or something like that? Nov 22 10:28:39 but if it's supposed to be used manually too then bindir seems better Nov 22 10:32:09 JaMa: I sent you an e-mail to hear your opinion. If you agree I'd send an RFC Nov 22 10:34:59 ah I thought it was sent off-list by accident.. Nov 22 10:35:39 I didn't want to delay them :) Nov 22 10:35:58 it's generic issue Nov 22 10:36:05 did you check if some benchmark depends on some lib provided only by meta-oe? if yes then it won't help much to separate benchmark layer Nov 22 10:36:06 JaMa: how about ${bindir}/ntpdate-sync? Nov 22 10:36:14 bluelightning: looks good Nov 22 10:36:19 ok, thanks Nov 22 10:36:29 bluelightning: I've asked JaMa about that meta-oe split out Nov 22 10:36:47 ant_work: sorry, which one? Nov 22 10:37:04 oh benchmark split out Nov 22 10:37:07 was: meta-openembedded/meta-oe is confusing Nov 22 10:37:21 oh right Nov 22 10:37:22 yes Nov 22 10:37:24 I don't think that splitting every few recipes does not work well unless they are really well separated/standalone Nov 22 10:38:19 many devs percieve meta-oe as a jump in the dark Nov 22 10:38:44 some just include a couple of layers of meta-openembedded (me) Nov 22 10:38:48 then they should report their problems on ML Nov 22 10:39:05 at the very least, meta-oe needs to have no bbappends and no overlayed recipes Nov 22 10:39:33 i.e. no change in behaviour when the user adds it to bblayers.conf Nov 22 10:40:04 bluelightning: what about stuff like qt4 bbappends? Nov 22 10:40:49 JaMa: I'm not convinced we should have those frankly - why can't those be in distro layers? it's not guaranteed that when I add meta-oe I want qt4 to be rebuilt just to get postgres/mysql support Nov 22 10:41:06 levonmaa: pong Nov 22 10:41:19 having those there is effectively setting policy, that's not what meta-oe should be doing Nov 22 10:41:47 or at least it shouldn't be trying to do that *and* add extra recipes that people want independently of any policy Nov 22 10:41:54 bluelightning: distro still can change QT_CONFIG_FLAGS, so policy is still in distro Nov 22 10:42:11 it just updates default value because it has postgres/mysql available Nov 22 10:42:29 JaMa: sure you can change it, after you realise meta-oe has altered it Nov 22 10:43:33 I'm going to push harder on this now especially as there is now a danny branch Nov 22 10:45:39 I pose my question again: what is the point having e.g. meta-openembedded/meta-multimedia and meta-openembedded/meta-oe/recipes-multimedia ? Nov 22 10:45:48 same with recipes-graphics etc. etc. Nov 22 10:46:05 i often get unrecognized options: --disable-silent-rules, --disable-dependency-tracking, --with-libtool-sysroot Nov 22 10:46:07 is it normal ? Nov 22 10:46:50 ant_work: that's just a case of nobody getting around to moving those things Nov 22 10:47:13 ant_work: they should be moved Nov 22 10:47:17 bluelightning: worth an RFC Nov 22 10:47:45 I wouldn't think it needs to be an RFC, the layers got created, they're in the same repo... there shouldn't be a problem Nov 22 10:48:24 we just need someone to send patches :) Nov 22 10:49:04 afournier: are you overriding do_configure? Nov 22 10:49:19 yes Nov 22 10:49:43 afournier: why you do that? Nov 22 10:50:03 to fix the Makefile most of the time Nov 22 10:50:08 afournier: ok, I think it's likely that the configure script distributed with the source doesn't support the options that we expect and that's why you get that message Nov 22 10:50:12 afournier: do_configure_append Nov 22 10:50:30 afournier: by using own do_configure you make life harder for others Nov 22 10:50:48 hrw: sometimes i also fix the configure Nov 22 10:50:58 afournier: fix configure.{in,ac} instead? Nov 22 10:51:33 afournier: I am working on armv8 stuff nowadays. Each time I see do_configure() in autotools based recipe I kill a kitty. Nov 22 10:51:45 hrw: trying Nov 22 10:51:52 afournier: my neighbours are affraid life of their pets... Nov 22 10:52:22 bluelightning: btw, first commit in meta-oe says: "move layer into meta-oe in preparation for future splits Nov 22 10:52:22 . As per TSC decision" Nov 22 10:52:32 bluelightning: did you tried php 5.4.4 again maybe? Nov 22 10:52:50 hrw: I haven't had time to look at it yet I'm afraid Nov 22 10:53:47 ok Nov 22 10:54:01 I am fighting update to 5.4.8 Nov 22 10:54:21 cause 5.4.4 ftfbs in crazy way for aarch64 Nov 22 10:54:30 hrw: do you see some build issues probably related to libxml changes? Nov 22 10:54:38 hrw: fails for me since yesterday Nov 22 10:54:56 JaMa: love you! tell me more Nov 22 10:54:57 | /OE/jansa-test/shr-core/tmp-eglibc/work/x86_64-linux/php-native/5.3.6-r2.0/php-5.3.6/ext/dom/node.c: In function 'dom_canonicalization': Nov 22 10:55:00 | /OE/jansa-test/shr-core/tmp-eglibc/work/x86_64-linux/php-native/5.3.6-r2.0/php-5.3.6/ext/dom/node.c:1903:21: error: dereferencing pointer to incomplete type Nov 22 10:55:01 yes! Nov 22 10:55:03 | /OE/jansa-test/shr-core/tmp-eglibc/work/x86_64-linux/php-native/5.3.6-r2.0/php-5.3.6/ext/dom/node.c:1905:5: error: dereferencing pointer to incomplete type Nov 22 10:55:06 | make: *** [ext/dom/node.lo] Error 1 Nov 22 10:55:27 then not aarch64 fault :) Nov 22 10:55:35 the same in evolution-data-server Nov 22 10:55:43 next month I will freeze all Linaro OE builds to one set of revs and then update monthly Nov 22 10:56:11 * JaMa will stick to danny with official SHR builds Nov 22 10:56:28 JaMa: we have many things already merged into master Nov 22 10:56:56 and php/mysqli extension segfaults for me so want to check newer php first Nov 22 11:13:21 morning Nov 22 11:13:30 hi ericben Nov 22 11:13:33 hi woglinde Nov 22 11:15:20 concerning PACKAGECONFIG for example in php we have PACKAGECONFIG[mysql] PACKAGECONFIG[pgsql] : how is done the choice ? is it using a DISTRO_FEATURE or by setting PACKAGECONFIG in local conf files ? Nov 22 11:23:49 bluelightning: hi, for modphp: what is the irght way to set PACKAGECONFIG ? Nov 22 11:25:17 ericben: you can either set it in a bbappend, or you can use PACKAGECONFIG_pn-modphp in local/distro config Nov 22 11:25:32 bbappend probably makes it more obvious though Nov 22 11:26:02 bluelightning: OK thanks. Would that make sense to generete it on the fly using DISTRO_FEATURES ? Nov 22 11:26:28 only if there is a system-wide DISTRO_FEATURES item that would make sense to set it from... Nov 22 11:26:38 I'm not sure there is in this case Nov 22 11:27:01 ok, in some packages that's the way it's done. Thanks Nov 22 11:27:08 indeed, we do set from DISTRO_FEATURES for other cases where there is Nov 22 11:34:31 hmm, there is no end to problems with this recipe :( Nov 22 11:35:17 we have ${PN}-bin and ${PN}-utils in ntp - only one seems to actually have files in it... Nov 22 11:35:41 ok, 5.4.8 failed same way Nov 22 11:35:51 | /home/hrw/HDD/devel/canonical/aarch64/build/arm-gcc-build/tmp-eglibc/sysroots/x86_64-linux/usr/libexec/aarch64-oe-linux/gcc/aarch64-oe-linux/4.7.3/ld: /home/hrw/HDD/devel/canonical/aarch64/build/arm-gcc-build/tmp-eglibc/sysroots/genericarmv8/usr/lib/../lib/libc-client.a(osdep.o)(.text+0x8ef4): unresolvable R_AARCH64_ADR_PREL_PG_HI21 relocation against symbol `daylight@@GLIBC_2.16' Nov 22 12:56:55 symbol daylight lol Nov 22 14:55:33 Don't spose someone could help me with trying to rebuild a kernel. I can't get the uImage places into the deploy directory to be updated, it seems to be coming from a cache somewhere. I do bitbake -c clean virtual/kernel; bitbake -c menuconfig virtual/kernel; (make changes); bitbake -c compile virtual/kernel; bitbake -c deploy virtual/kernel. But the created uImage does not have my menuconfig changes Nov 22 14:57:28 furthermore, I've tried updating the defconfig for the kernel in use (which I've confirmed the changes are there), and no matter what 'cleaning'/rebuilding I do, I can't get the generated uImage to be updated (I checked /proc/config.gz). This is all on a beaglebone Nov 22 14:59:46 i've created own recipes which use svn with user/pass in the URL - this info is stored in IPK in controls in control.tar.gz - is there a way to drop credentials in this file? Nov 22 15:09:20 yay, I get to revert ross' glib commit :) Nov 22 15:10:05 not entirely true. just bits. Nov 22 15:15:24 xanium4333: which version are you using? Nov 22 15:30:15 did anyone include eglib-locale in an rpm build lately? I get "Unable to resolve package eglibc-locale". it works with ipk. Nov 22 15:30:59 a build from master, that is Nov 22 15:35:35 bluelightning: this is actually an angstrom 2012.05 setup, ie all the OE stuff checked out by their setup-scripts Nov 22 15:36:17 I'm now trying koen's instructions again (https://groups.google.com/forum/#!msg/beagleboard/UlL4e0_O_UE/vAmLNmkXc1sJ) just to make sure... Nov 22 15:37:05 the config file places in the deploy DIR does not have my changes within it, and I'm *sure* my changes are in the defconfig (and I've checked the work/kernel/git/.config file as well Nov 22 15:41:44 silly question but i'll ask anyway. how to install meta-toolchain-openmoko? i already successfully installed meta-toolchain. do i need core-image-minimal? Nov 22 15:52:37 nope, still didn't work. The config file in the deploy directory does not have my changes (which show up in bitbake virtual/kernel -c menuconfig) Nov 22 16:02:17 are you sure you replaced the right defconfig ? Nov 22 16:05:41 morning Nov 22 16:07:29 xanium4333: did you try -c cleansstate ? Nov 22 16:08:53 your problem looks like this: https://bugzilla.yoctoproject.org/show_bug.cgi?id=2256 Nov 22 16:11:29 morning although here it is evening :) Nov 22 16:11:50 aloniv: only 11:11 here :) Nov 22 16:13:45 chouimat: i'm having trouble with openembedded. can you help? Nov 22 16:26:21 aloniv: please ask your question Nov 22 16:27:23 re Nov 22 16:29:32 bluelightning - trying to get openmoko environment. Nov 22 16:29:50 oh right I see your question above Nov 22 16:29:50 on wiki they say to bitbake meta-toolchain-openmoko Nov 22 16:30:02 i bitbaked meta-toolchain Nov 22 16:30:09 bitbaked :) Nov 22 16:30:40 what now? Nov 22 16:31:13 I would suggest following their instructions (i.e. bitbake meta-toolchain-openmoko) and if you have problems you probably need to ask in #openmoko-cdevel Nov 22 16:31:13 on this page Nov 22 16:31:21 http://www.openembedded.org/wiki/SDK/Toolchain/Application_Developer_Toolkit Nov 22 16:31:31 they claim i get some tarball which i can't find Nov 22 16:31:44 meta-toolchain-openmoko doesn't exist Nov 22 16:32:32 I guess those are outdated instructions... Nov 22 16:33:35 aloniv: http://build.shr-project.org/shr-core/sdk/ Nov 22 16:33:47 thanks but i want older openmoko... Nov 22 16:33:56 om2009 Nov 22 16:34:24 aloniv: http://downloads.openmoko.org/developer/toolchains/ Nov 22 16:34:27 sadly never had much luck with shr Nov 22 16:35:16 these are compiled i think Nov 22 16:36:06 if you want to compile them, then you have to checkout very old version of oe-classic Nov 22 16:36:25 but a lot of URLs etc won't be accessible now I guess Nov 22 16:39:45 unfortunately the compiled version only works on distros based on ubuntu 8.04 or 8.10 Nov 22 16:40:07 well what do you expect from dead old toolchain :) Nov 22 16:40:11 and i don;t really want to install e.g. gnewsense for this Nov 22 16:40:17 *don't Nov 22 16:54:21 bluelightning: thanks a lot for ntp changes Nov 22 16:54:40 JaMa: you're welcome... took quite a lot more time than I had planned :/ Nov 22 16:54:51 bluelightning: there is small typo in commit message "/etc/default/ntpdate.default" should be without .default suffix but in code it's right Nov 22 16:55:12 argh... oops Nov 22 16:55:51 and patch for meta-systemd still in queue? Nov 22 16:56:04 I just sent it Nov 22 16:56:31 including otavio's patches as somehow they did not all hit patchwork Nov 22 16:57:07 and is it ntp-date or ntpdate package now? Nov 22 16:57:25 ntpdate Nov 22 16:57:48 the previous patch that changed it was reverted; I was going to revert the name anyway though Nov 22 16:58:03 good Nov 22 16:58:12 I also prefer ntpdate Nov 22 16:58:39 as do I... meta-opie even broke on that change Nov 22 16:59:09 worth noting though that ntpdate itself may go away in future according to upstream Nov 22 16:59:16 we'll deal with that when the time comes though Nov 22 17:45:57 is there a way to add default user to a rootfs? Nov 22 17:46:42 chouimat: useradd.bbclass should help you Nov 22 17:46:53 JaMa: thanks Nov 22 17:50:00 JaMa: the problem is I'm stuck with classic for this project Nov 22 17:57:01 chouimat: in that case, it's probably a matter of creating a shell function with calls to the useradd command and adding a call to the function to ROOTFS_POSTPROCESS_COMMAND Nov 22 18:03:27 thanks Nov 22 19:13:07 tsjsieb, I think you're not here, but here's a 'thanks' for the hint about sstate, just trying it now, and I think it'll solve the problem Nov 22 19:27:59 is there any reason I can't update bitbake, if I'm using angstrom's 'standard' environment made by ./oebb.sh? Nov 23 00:36:20 in Paul E's Fosdem talk, he lists: Web-based layer index (searchable) <-- has the search interface been created? Nov 23 00:38:59 this one: https://archive.fosdem.org/2012/schedule/event/758/143_fosdem_2012_yocto_oe.pdf Nov 23 01:20:19 vmeson, he has not got to that yet :) Nov 23 01:22:56 Crofton|work: thx. I also looking for guidelines wrt layer priority - are there reserved priorities? maxim = 10? I guess I should play with some extra yocto layers . Nov 23 01:23:19 and bitbake-layers to see what is being overlayed... Nov 23 01:23:23 yeah Nov 23 01:25:30 ah, that's not so bad after all. **** ENDING LOGGING AT Fri Nov 23 02:59:58 2012