**** BEGIN LOGGING AT Sun Apr 14 02:59:58 2013 Apr 14 09:00:40 Any reason for why OE is sticking to Gtk3 version 3.4.4 ? Apr 14 09:06:33 jkroon_: it's wairing for at-spi2-atk to be merged Apr 14 09:06:44 and that is waiting for some site entries added Apr 14 09:07:09 JaMa, ok, glad to hear its in the pipeline Apr 14 09:07:43 http://patches.openembedded.org/patch/43077/ Apr 14 09:53:20 xserver-nodm-init fires up the pointer calibration program on boot. I have a custom x session that starts up gtk3-demo, without paying any respect to that the pointercalibration is running aswell Apr 14 10:00:21 I saw the same thing when building with packagegroup-core-x11-sato. It would be nice if the x session would wait for the pointer calibration program to finish, before starting Apr 14 10:01:03 But maybe I'm missing some good reasons for why it done as it is today ? Apr 14 10:05:40 provide pointercal-xinput bbappend for you MACHINE Apr 14 10:06:03 -> save calibration data from running device and add it to BSP layer Apr 14 10:30:37 yeah.. not quite what I want though, best for me would be if the calibration could run undisturbed, if no calib data existed yet Apr 14 11:44:57 I could use some up2date docs for writing bitbake recipes... Apr 14 11:45:38 jkroon_: I'd recommend looking at the yocto project documentation, specifically the reference and development manuals Apr 14 11:45:54 bluelightning, thanks, will do Apr 14 11:46:00 if you're using master, the "in progress" versions are the most applicable Apr 14 11:54:54 Hmm. I was doing "DESTDIR=${D} oe_runmake install" in the do_install() function. Changing that to just "DESTDIR=${D} make install" improves things. DESTDIR didnt get exported to my Makefile otherwise. But I guess I'm doing something else wrong.. Apr 14 11:57:06 EXTRA_OEMAKE variable Apr 14 11:57:13 but thats nothing new Apr 14 11:57:44 than you do no need to overwrite do_install Apr 14 11:57:51 right Apr 14 12:17:54 I get "Variable FILES is set as not being package specific" warnings, even though I use FILES_${PN} in my bitbake recipe Apr 14 12:18:38 I've tried "-c cleansstate" to get rid of any old files Apr 14 12:19:23 jkroon_: can I see your recipe? Apr 14 12:19:44 bluelightning, absolutley, let me upload it somewhere... Apr 14 12:20:53 bluelightning, http://pastebin.com/bubfSbTd Apr 14 12:22:19 if that's really all that's in the file I can't see how that error could come up Apr 14 12:22:31 it's definitely warning you about that recipe specifically? Apr 14 12:22:40 and you don't have the same named recipe elsewhere? Apr 14 12:24:33 I have : test.bb test-release.bb test-session.bb. It's test.bb that is complaining. Maybe I need to suffix my files with version numbers ? Apr 14 12:24:50 all three are different packages Apr 14 12:25:17 Or do I have stale files in the build system that I need to cleanup first ? Apr 14 12:25:41 jkroon_: no, as long as you don't use _ in the name except to separate the name and the version, those names should be fine Apr 14 12:26:50 ok Apr 14 12:27:15 "cleaning out" files isn't going to help, no Apr 14 12:38:20 hmm.... I was actually missing a FILES_${PN} in test-session.bb... Apr 14 12:38:56 jkroon_: normally you shouldn't need to set FILES at all Apr 14 12:40:55 jkroon_: if it were installing to ${bindir} rather than /usr/local/bin/ then you wouldn't Apr 14 12:49:50 bluelightning, aha Apr 14 12:50:04 let me try changing to ${bindir} then Apr 14 12:55:12 a side note, I get a test-dbg and test-dev rpm package, but plain test package Apr 14 12:55:21 i have a lot to learn about EO :-( Apr 14 12:55:25 OE even Apr 14 13:21:34 jkroon you are on a good way Apr 14 14:02:44 i dont get it. does bitbake scan the do_install() before installation, and since I dont mention ${bindir} in there, it complais about FILES not being set package specific Apr 14 14:07:11 im using RPMs, if that should matter Apr 14 14:25:03 jkroon_: nothing like that, no Apr 14 14:31:24 jkroon_: I'd suggest you use bitbake -e | less and then search for FILES to see where your errant value is being set Apr 14 14:34:29 bluelightning, ok, I can find where it is being set.. Apr 14 14:37:47 It under a "# computed:" section Apr 14 14:47:56 jkroon why do you still bother with do_install? Apr 14 14:49:04 can I just remove it ? will OE automatically pass DESTDIR to "make install" ? Apr 14 14:49:21 use EXTRA_OEMAKE variable Apr 14 14:49:32 you can git grep for it in other recipes Apr 14 14:50:09 yeah.. right now it just adds to the complexity for me, i dont know what it does Apr 14 14:50:26 ???????? Apr 14 14:50:28 interestingly, renaming the recipe from test.bb to testt.bb seems to fix things Apr 14 14:50:41 :-) Apr 14 14:50:48 passing extra arguments like DISTMOO=foolo ot oemake call Apr 14 14:50:56 which than is passed to make Apr 14 14:51:12 so using own do_install extends complexity Apr 14 14:51:20 ok Apr 14 14:51:47 do I need my own do_compile() ? I have no idea what the default functions look like Apr 14 14:56:41 ok.. finallly! I was using a distro, also called "test", test.conf Apr 14 14:56:53 so test.conf and test.bb, not ok ? Apr 14 14:57:12 better use it with version Apr 14 14:57:28 test_git.bb or test_01.bb Apr 14 15:04:10 jkroon_: right, now I realise what is happening Apr 14 15:04:31 jkroon_: since DISTRO="test", "test" is being treated as an override Apr 14 15:05:02 woglinde: hey Apr 14 15:05:04 so FILES_${PN} becomes FILES_test and since test is an override, it just becomes FILES Apr 14 15:05:04 gm all Apr 14 15:05:12 hi khem Apr 14 15:05:26 we ought to have a warning when PN is in overrides Apr 14 15:05:34 woglinde: I am hitting https://github.com/woglinde/meta-java/issues/24 Apr 14 15:05:36 again with master Apr 14 15:05:50 I have no idea what could be wrong Apr 14 15:05:58 it works most of the time Apr 14 15:06:11 but once in a while on clean build it does break like that sitll Apr 14 15:06:22 khem I build with master today Apr 14 15:06:24 this time no -ccleansstate is helping :( Apr 14 15:06:25 and had no problems Apr 14 15:06:40 woglinde: whats your parallel build value Apr 14 15:07:19 j2 Apr 14 15:07:27 bbthreads too Apr 14 15:07:53 here its 20 Apr 14 15:07:57 uh hm Apr 14 15:08:02 bluelightning: http://pastebin.com/tBJ9fg2N Apr 14 15:08:03 bluelightning, should I file an improvement-bug for that or something ? Apr 14 15:08:17 jkroon_: I'm about to file it Apr 14 15:08:27 bluelightning, cool, thanks Apr 14 15:09:16 khem can you check if the classes are 0 bytesß Apr 14 15:09:41 woglinde: I also see http://pastebin.com/NRf8jPGh Apr 14 15:10:07 woglinde: hmmm so checcking in sysroot would be enough ? Apr 14 15:11:03 khem ah yes multilib stuff Apr 14 15:11:18 or multi machine build Apr 14 15:11:47 woglinde: ./classpath-native/0.99-r6.0/classpath-0.99/.pc/sun-security-getproperty.patch/sun/security/action/GetPropertyAction.java Apr 14 15:11:51 is 0 Apr 14 15:12:10 which is a patch Apr 14 15:12:11 thats a patch Apr 14 15:12:37 classes are intressting Apr 14 15:12:43 ./classpath-initial/0.93-r6.1/classpath-0.93/lib/resources Apr 14 15:12:43 ./classpath-initial/0.93-r6.1/classpath-0.93/lib/compile-classes Apr 14 15:13:52 well this one is not multilib and single machine Apr 14 15:14:15 I said multimachine Apr 14 15:14:28 when building for qemuarm and than beagle Apr 14 15:14:44 jo jkridner Apr 14 15:14:51 woglinde: single machine as I said Apr 14 15:15:02 hm Apr 14 15:15:03 strange Apr 14 15:15:30 ah okay Apr 14 15:15:43 maybee for intial I should disable the tools Apr 14 15:15:45 let me see Apr 14 15:15:48 yes Apr 14 15:15:56 unless they are needed Apr 14 15:18:36 next phase is to merge native packages with normal for openjdk Apr 14 15:18:57 I will make two tickets Apr 14 15:19:57 ok Apr 14 15:21:13 hm but with the other problem Apr 14 15:21:20 I have no machine to test with -j20 Apr 14 15:21:52 its some concurrency I guess Apr 14 15:22:03 usually I cleaned gnumail-native and it woudl work Apr 14 15:22:06 this time no luck Apr 14 15:22:09 its consistent Apr 14 15:24:16 khem maybbe its gnumail target Apr 14 15:24:23 wild guess Apr 14 15:25:08 ah Apr 14 15:25:20 * khem shoots in dark Apr 14 15:25:55 or at which recipe it occurs? Apr 14 15:27:06 woglinde: log4j1.2-native Apr 14 15:27:34 woglinde: I cleaned gnumail-native gnujaf-native Apr 14 15:27:37 no avail Apr 14 15:27:48 hm classpath-initial has --disable-tools Apr 14 15:28:32 khem can you do -c build gnumail-native Apr 14 15:28:43 and than check if there are classes with 0 bytes Apr 14 15:28:54 ok Apr 14 15:28:54 | 11. ERROR in src/main/java/org/apache/log4j/net/SMTPAppender.java (at line 33) Apr 14 15:28:54 | import javax.mail.Authenticator; Apr 14 15:28:54 | ^^^^^^^^^^ Apr 14 15:29:00 this is the first error I get Apr 14 15:29:06 so its clearly something in gnumail Apr 14 15:29:17 but who can inject it no idea Apr 14 15:30:20 hm I should update jamvm too Apr 14 15:30:30 $ find -type f -size 0 Apr 14 15:30:30 ./1.1.2-r1/mail-1.1.2/.pc/datadir_java.patch/.timestamp Apr 14 15:30:32 it has now java8 support and some internal changes Apr 14 15:30:42 khem hm strange Apr 14 15:31:03 maybee the jar does not get exchange Apr 14 15:31:15 so you always stick with the faulty Apr 14 15:31:33 you mean in the sysroot ? Apr 14 15:31:58 yes Apr 14 15:33:21 it does see http://pastebin.com/zegpwSrk Apr 14 15:33:40 the datestamp on gnumail jars is from a minute ago Apr 14 15:34:19 and nothing has changed in my machine Apr 14 15:34:30 usual updates for oe-core and other layers Apr 14 15:34:37 but no changes in meta-java Apr 14 15:34:52 hm hm Apr 14 15:35:12 khem are you build on ka6sox cluster? Apr 14 15:35:26 its work Apr 14 15:35:31 machine Apr 14 15:35:37 okay Apr 14 15:35:42 hi bildman Apr 14 15:35:44 which I (ab)use Apr 14 15:35:49 ups blindman Apr 14 15:36:17 khem can you md5sum the jars in sysroot Apr 14 15:36:29 and in the noew build dir? Apr 14 15:36:35 actual build dir Apr 14 15:37:56 woglinde: http://pastebin.com/RgnfVNEs Apr 14 15:39:14 woglinde: md5 is identical in sysroot and builddir Apr 14 15:40:16 and still the error Apr 14 15:40:35 khem can you add -j1 to gnumail build? Apr 14 15:41:10 moin Apr 14 15:43:17 something is messed up in systemd as well for me with recent commits Apr 14 15:43:25 its doesnt find ttyS0 Apr 14 15:43:47 woglinde: PARALLEL_MAKE = "" did not help either :( Apr 14 15:44:25 hm sorry Apr 14 15:44:32 than I do not know at the moment Apr 14 15:44:41 and its hard to debug without access Apr 14 15:46:28 woglinde: who builds activation.jar Apr 14 15:49:44 woglinde: | DEBUG: Executing shell function do_compile Apr 14 15:49:44 | incorrect classpath: -d Apr 14 15:49:55 classpath may be is wrong somehow ? Apr 14 15:50:16 activation should be gnumail too Apr 14 15:50:45 hm okay I see now classpath-initial has no configure option to disable tools Apr 14 15:50:50 yes, so I copies the jars from a working build Apr 14 15:50:53 and it still fails Apr 14 15:51:07 woglinde: manually del them I guess ? Apr 14 15:52:27 hm yes I have the funtion in Apr 14 15:52:33 but why does it not work Apr 14 15:52:48 rm ${D}${bindir}/${i} Apr 14 15:54:49 args Apr 14 15:56:52 bluelightning can I now remove all my PR's in meta-java or will that break upgrade paths? Apr 14 15:59:39 you need to keep them Apr 14 15:59:49 remove only when PV is incremented Apr 14 16:02:14 interesting, meta-handheld has: conf/machine/include/zaurus.inc:MACHINE_ESSENTIAL_EXTRA_RDEPENDS_append = " kernel-base kernel-image" Apr 14 16:02:38 but packagegroup.*boot is not rebuild when new kernel version is built Apr 14 16:03:06 causing do_rootfs to fail Apr 14 16:03:07 | * satisfy_dependencies_for: Cannot satisfy the following dependencies for packagegroup-core-boot: Apr 14 16:03:10 | * kernel-3.4.0 * kernel-image-3.4.0 * Apr 14 16:05:56 thanks jama Apr 14 16:07:46 woglinde: hmmm strange indeed, so I copied the gnumail jars from a working build and it worked Apr 14 16:08:25 JaMa: building now for spitz Apr 14 16:09:04 pulled right now, collie still builds fine l-y 3.8 Apr 14 16:09:24 woglinde: you said new jamvm ? Apr 14 16:09:35 may be it solves relocatibility issues ? Apr 14 16:09:38 hi khem woglinde JaMa btw ;) Apr 14 16:09:46 ant_home: great, thanks Apr 14 16:09:48 woglinde: see http://old.nabble.com/Re:-Problems-with-JamVM-(I-am-a-beginner)please-help-me-out!-p17404599.html Apr 14 16:09:51 ant_home: hi Apr 14 16:10:00 JaMa: systemd is broke on me Apr 14 16:10:09 I have to figure out why :( Apr 14 16:10:13 ant_home: I was still building my old linux.inc based recipe Apr 14 16:10:16 khem: how? Apr 14 16:10:52 khem: btw have you seen my systemd "fix" for external toolchains? what do you think about HOST_PREFIX/TARGET_PREFIX inconsistency? Apr 14 16:10:58 khem this have nothing to do with this problem I think Apr 14 16:10:59 it does end up with Apr 14 16:11:00 [ TIME ] Timed out waiting for device dev-ttyS0.device. Apr 14 16:11:00 [DEPEND] Dependency failed for Serial Getty on ttyS0. Apr 14 16:11:27 hm so stupid "for i in " bug fixed Apr 14 16:11:36 need to copy and to push it Apr 14 16:11:41 woglinde: that problem in that link is a sstate prolblem for meta-java Apr 14 16:11:55 woglinde: push Apr 14 16:12:02 woglinde: if you want me to take a look Apr 14 16:12:06 I can I have time now Apr 14 16:12:48 khem: I have that too, but that wasn't fatal for me, fatal was hanging run-postinsts.service Apr 14 16:16:27 When using a git repo as source for a package, can I specify "use latest revision" instead a hardcoded sha1 commit ? Apr 14 16:16:55 khem hm but you made patches for it? Apr 14 16:19:16 jkroon_: AUTOREV is what you're looking for Apr 14 16:19:53 jkroon_: but don't write it directly in recipe, use revision known to work in recipe and set AUTOREV from some DISTRO .inc file outside Apr 14 16:20:24 JaMa, ok Apr 14 16:20:36 JaMa: my issue is console does not start Apr 14 16:20:49 JaMa: and the device does not have network so ssh is not an option Apr 14 16:21:38 JaMa: I have a hunch its something botched up in oe-cor Apr 14 16:21:40 e Apr 14 16:21:46 or may be my end Apr 14 16:21:51 but thats less likely Apr 14 16:22:23 JaMa: I havent looked at oe mls for past 2 days Apr 14 16:22:30 ah right, I'm just lucky to have more ways to access Apr 14 16:22:45 khem: then you will be happy to see patches there :) Apr 14 16:22:57 JaMa: did you fix it ? Apr 14 16:23:02 if yes point me to patch Apr 14 16:23:03 not this one Apr 14 16:23:07 ah Apr 14 16:23:24 I wonder what broke it Apr 14 16:23:30 prolly those post installs Apr 14 16:23:49 I need to cut a release branch based on dylan Apr 14 16:23:53 and its not looking good Apr 14 16:23:57 fwiw I was able to reproduce this failing ttyS0 also with oe-core only and some core-image Apr 14 16:24:06 hmmm Apr 14 16:24:17 systemd-image on angstrom works all fine Apr 14 16:24:22 so probably not issue on your end Apr 14 16:24:28 and my failure is based on poky Apr 14 16:24:46 hmm ok Apr 14 16:24:53 well I'm not sure if I haven't "fixed" it in my case by changing qemu params for serial in the end Apr 14 16:25:05 so if I add systemd to DISTRO_FEATURES and do bitbake core-image-minimal will I see it ? Apr 14 16:25:32 JaMa: mine is not qemu its a real box :) Apr 14 16:25:56 I guessed that :) Apr 14 16:26:03 and ross probably came and went Apr 14 16:26:06 :) Apr 14 16:26:19 so you need a bit of solder too :) Apr 14 16:26:27 and folks have scripts running Apr 14 16:26:32 which groks console Apr 14 16:26:36 and automake stuff Apr 14 16:26:46 so having no console is dead meat for me Apr 14 16:27:44 jama point me to your external tc patch please Apr 14 16:27:58 if you have a minute, please reply on my whitespace thread http://lists.linuxtogo.org/pipermail/openembedded-devel/2013-April/044989.html Apr 14 16:28:32 khem: external tc + systemd http://lists.linuxtogo.org/pipermail/openembedded-core/2013-April/037985.html Apr 14 16:32:29 khem: which llvm is "better"? the one in meta-extra seems pretty old and unmaintained and llvm2.[89] also provide native Apr 14 16:33:37 JaMa: STRINGS and GPERF should be added to bitbake.conf Apr 14 16:33:42 not to systemd recipe Apr 14 16:33:52 it would fix many other recipes Apr 14 16:34:03 but exporting them is not enough Apr 14 16:34:10 that's the problem Apr 14 16:35:10 and I don't know where is boundary between AC_PATH_TOOL and AC_CHECK_TOOL because some TC can provide e.g. GPERF and some no Apr 14 16:35:38 that sed -i is needed yes Apr 14 16:35:51 in systemd Apr 14 16:36:10 well it should fail if they dont Apr 14 16:36:22 or disable it completely in systemd Apr 14 16:36:42 external tc class can stub it Apr 14 16:36:44 if needed Apr 14 16:36:48 many ways Apr 14 16:37:11 JaMa: the one you send for llvm is better Apr 14 16:37:15 but it does not fail in configure.. it finds tool on host and fail much later in do_compile Apr 14 16:37:19 but I wanted to bring to your notie Apr 14 16:37:32 fair enough Apr 14 16:39:03 JaMa: and use AC_PATH_TOOL Apr 14 16:39:05 bluelightning ping Apr 14 16:39:28 JaMa: and if not found bail out Apr 14 16:39:43 I think even systemd upstream will accept such a patch Apr 14 16:39:46 but how many other packages are using AC_CHECK_TOOL .. Apr 14 16:40:04 JaMa: we cant end the world hunger :) so one at a time Apr 14 16:40:08 khem: fixing HOST_PREFIX to match TARGET_PREFIX should fix all Apr 14 16:40:22 well not the world hunger :) Apr 14 16:40:45 JaMa: for crosstoolchains like gcc HOST_PREFIX != TARGET_PREFIX Apr 14 16:41:29 may be BUILD_PREFIX could be = HOST_PREFIX in some cases when you run the compiler on same host you built it on Apr 14 16:42:34 but adding GPERF and STRINGS to bb.conf patch should be nice if you can send it to core ml Apr 14 16:43:06 woglinde: openjdk-7-jre takes like 30 mins and its single threaded build :( Apr 14 16:43:38 my whole other stuff is done in 40 mins and its waiting for openjdk Apr 14 16:51:03 woglinde: can I use jamvm to provide JRE Apr 14 16:51:12 woglinde: instead of openjdk ? Apr 14 16:51:13 I wonder Apr 14 16:51:30 my target is ppc and I am only usin zerovm Apr 14 16:51:34 which sucks anyway Apr 14 17:01:55 ant_home: there are some wrong dependencies between modules Apr 14 17:03:50 khem theoretical you can Apr 14 17:04:09 ant_home: they depend only on kernel-3.8.4-yocto-standard, not on required modules like they did with linux.inc based kernels (and maybe other l-y versions too) Apr 14 17:04:30 with classpath Apr 14 17:15:15 woglinde: EXTRA_OECONF_append = " --enable-zero=no" Apr 14 17:15:15 WITH_ADDITIONAL_VMS_powerpc = "jamvm" Apr 14 17:15:15 WITH_ADDITIONAL_VMS = "" Apr 14 17:15:23 that should do it Apr 14 17:16:09 woglinde: the graph at the end here http://jamvm.sourceforge.net make me think that jamvm could be more optimizing Apr 14 17:16:15 than zerovm Apr 14 17:16:21 pwd Apr 14 17:16:22 clear Apr 14 17:17:57 hm yes ppc could run without zero Apr 14 17:18:38 khem but the graph is old Apr 14 17:18:51 sourcecode is on berlios Apr 14 17:19:09 woglinde: I have to try Apr 14 17:19:20 currently operations take like 20 to 30 mins Apr 14 17:19:34 may be jamvm can reduce it to say 15mns Apr 14 17:19:39 would be a win Apr 14 17:19:56 woglinde: where is jamvm hosted Apr 14 17:20:22 as I said sourcecode is on berlios Apr 14 17:23:03 woglinde: yeah I see its git3.berlios.de Apr 14 17:23:06 thx Apr 14 17:25:45 JaMa, yes, that spitz_pm patch is upstream. Is cruft of 3.6 Apr 14 17:27:29 * JaMa now trying linux-yocto-tiny-kexecboot Apr 14 17:27:50 * JaMa haven't upgraded spitz for ages Apr 14 17:28:16 the kernel on kexecboot.org has a more recent, dev version of kexecboot Apr 14 17:28:29 ant_home: is there easy way to make kexecboot verbose again? to see why it does not boot? Apr 14 17:29:12 just coinfigure it with --enable-debug Apr 14 17:29:23 and kernel? Apr 14 17:29:27 I was using CMDLINE_DEBUG = "debug" Apr 14 17:29:36 but that probably doesn't work with l-y Apr 14 17:29:44 ah, you mean kernel... re-enable printk and bug at least Apr 14 17:30:08 yes that what linux.inc did for me :) Apr 14 17:30:26 on my 3200 all is ok Apr 14 17:30:54 which kernel doesn't boot? Apr 14 17:30:59 l-y-3.8 Apr 14 17:32:07 hm, apart the huge size it did boot on all my devices Apr 14 17:33:23 I did last round some weeks ago, though Apr 14 17:38:55 oh disabling sysvinit in oe-core is a pain you need two knobs Apr 14 17:39:14 ant_home: the same with kexecboot-kernel from kexecboot.org Apr 14 17:39:38 you mean it cannot launch a 3.8 kernel? Apr 14 17:39:39 ant_home: only difference is that with newer kexecboot screen wents blank before it hangs Apr 14 17:44:06 and self built linux-yocto-tiny-kexecboot does not start kexecboot maybe because of systemd Apr 14 17:53:56 hmm, isnt Adwait gtk3 theme available in any of the layers ? Apr 14 18:05:18 ant_home: any idea why /lib/modules are in initramfs image? http://pastebin.com/n9eES17A Apr 14 18:05:26 and var/lib/opkg too Apr 14 19:25:36 JaMa: I'm back now Apr 14 19:25:42 pls see http://paste.debian.net/249555/ Apr 14 19:27:37 I think I'll have to check the dummy regulators...dromede said are needed now by ads7846 Apr 14 19:30:50 JaMa: S98run-postinsts[1119]: /etc/rcS.d/S98run-postinsts: line 2: /etc/default/rcS: No such file or Apr 14 19:31:45 JaMa: btw. tty issue is fixed by DISTRO_FEATURES_BACKFILL_CONSIDERED_append = " sysvinit" Apr 14 19:31:49 in my case Apr 14 19:32:35 hm... BUG: mapping for 0x00000000 at 0xff000000 out of vmalloc space Apr 14 19:32:53 kind of bad news Apr 14 19:34:00 khem: interesting I have DISTRO_FEATURES_BACKFILL_CONSIDERED too and /etc/default/rcS was there iirc, because running /etc/rcS.d/S98run-postinsts manually worked fine (after removing .service file to boot it) Apr 14 19:35:13 in my case /etc/default/rcS is not there Apr 14 19:35:19 if I touch it then it runs fine Apr 14 19:35:26 so somethin to fix Apr 14 19:37:07 just checked buildhistory/images/qemux86_64/eglibc/core-image-minimal/files-in-image.txt and it was there when I was testing it Apr 14 19:37:23 khem: I hope that wasn't my fault :( Apr 14 19:38:10 but in images/qemuarm/eglibc/core-image-base/files-in-image.txt it isn't there Apr 14 19:38:19 so yes, there is bug in some combinations Apr 14 19:39:05 JaMa: apart spitz specific issues, yes, there are oddities with modules when switching between linux and linux-yocto provider. Apr 14 19:39:18 and again, yes, the initramfs tends to be inflated Apr 14 19:39:44 in my case, distroless, it is still small enough Apr 14 19:40:00 khem: yes, it was one of my commits that added that; I don't think it's even necessary Apr 14 19:40:21 I'll prepare a patch now Apr 14 19:40:46 bluelightning: thx Apr 14 19:41:45 bluelightning: if you cook quickly, I have setup to test it out Apr 14 19:42:09 bluelightning: btw. I sent you a lin about some warnings I am seeing Apr 14 19:42:18 what is all that about ? Apr 14 19:43:10 khem: sent Apr 14 19:43:38 khem: er when did you send that? Apr 14 19:43:50 in the morning prolly few hrs ago Apr 14 19:44:42 its a patebin.com link addressed to you Apr 14 19:44:49 so how do I fix those Apr 14 19:45:30 http://pastebin.com/tBJ9fg2N Apr 14 19:48:40 bluelightning: testing your patch now Apr 14 19:48:53 will send Tested-by once it passes my tests Apr 14 19:55:56 khem: to fix those, don't use VARIABLENAME_recipename, use VARIABLENAME_${PN} Apr 14 19:55:59 khem: most of them are fixed by patches on ML Apr 14 19:56:20 only the first from meta-java looks new to me Apr 14 19:58:44 JaMa: cool ok Apr 14 19:59:11 hm? Apr 14 19:59:43 too much churn for systemd towards the tail end of release ... not good Apr 14 20:00:38 JaMa: I will take in and test those too Apr 14 20:00:45 but dont wait for me to commit them Apr 14 20:00:55 warnings or systemd changes? Apr 14 20:01:00 yes... we could have done with more testing earlier Apr 14 20:01:01 warnings Apr 14 20:01:02 I'm waiting for jenkins builds :/ Apr 14 20:01:09 OK Apr 14 20:01:19 building 40 hours already :/ Apr 14 20:01:32 but we will make it. Apr 14 20:02:02 bluelightning: we could have done testing earlier if discussions about systemd weren't stalled on oe-core ML :/ Apr 14 20:02:52 * JaMa read "[OE-core] RFE: make the init manager an image feature (again)" thread again today and there are still valid unanswered questions Apr 14 20:03:21 JaMa: yes, however, I think we have made progress Apr 14 20:03:26 which is what matters Apr 14 20:03:57 well it depends if it's improvement from 1.3 or not Apr 14 20:04:33 sofar I see only more issues with systemd then before Apr 14 20:05:07 JaMa: hopefully dust will settle with all work you and all of us are putting in Apr 14 20:05:13