**** BEGIN LOGGING AT Sat Sep 10 02:59:58 2011 Sep 10 05:33:49 hi wonder anyone know any tutorial on creating a binary file for a target board using openembedded? Sep 10 06:27:09 anyone know where is cross toolchain is under in the tmp directory? Sep 10 08:28:19 i have a patch fail for 11239 Sep 10 08:28:30 anyone that help? Sep 10 08:44:36 hi GNUtoo|laptop Sep 10 08:45:00 hi Sep 10 08:46:31 I've fixed the meta-toolchain-qte problem Sep 10 08:46:39 I'm testing and patches are coming Sep 10 08:50:59 ok Sep 10 08:54:07 hi regarding the patch for 2.20.1 how do i go about doing it? Sep 10 09:26:38 hi regarding the patch for 2.20.1 how do i go about doing it? Sep 10 09:27:20 While compiling native=perl in bitbake Open Embedded i am getting undefined reference to `floor' Sep 10 10:03:35 morning Sep 10 10:05:05 hi otavio Sep 10 10:06:09 what is the thing around qt.conf ? Sep 10 10:09:51 ericben: ah; just replied to the ml Sep 10 10:10:00 ericben: take a look on the reply first hhehe Sep 10 10:12:04 otavio: done Sep 10 10:12:21 but qt.conf is not in the sdk so I don't understand everything ;-) Sep 10 10:12:33 ericben: I've been doing a qt sdk at O.S. Systems and for it to work I have a qt.conf with ... Sep 10 10:12:50 do you add qt.conf manuall y? Sep 10 10:13:19 find . -iname qt.conf in /usr/local/angstrom-eglibc-x86_64-armv5te doesn't return anything here Sep 10 10:13:28 ericben: http://paste.debian.net/129180/ Sep 10 10:13:46 ericben: ahh so you fixed the issue :-) Sep 10 10:14:32 otavio: strange when you say sdk do you mean the toolchain inside oe or a meta-toolchain-qte ? Sep 10 10:14:41 meta-toolchain Sep 10 10:14:56 ok so I should have qt.conf as I didn't change anything concerning it ! Sep 10 10:15:29 ericben: but I have this in my meta-toolchain-ossystems.bb Sep 10 10:15:36 ericben: not on oe-core one Sep 10 10:15:40 otavio: ah OK I understand ;-D Sep 10 10:15:51 ericben: can you use: qmake -query Sep 10 10:15:57 ericben: and paste the output? Sep 10 10:16:57 what I can say is that with the patch I sent I mange to compile qt app in shell and in qtcreator, and you're right, qtmake query returns the right paths http://paste.debian.net/129182/ Sep 10 10:17:02 ericben: I have the impression if you build two sdks for two different targets your qmake will report wrong paths Sep 10 10:17:17 otavio: true Sep 10 10:18:04 otavio: in fact no as in angstrom the sdk path has the target in the name to that would make 2 different sdk Sep 10 10:18:18 it is in /usr/local/angstrom-eglibc-x86_64-armv5te Sep 10 10:18:23 ericben: you might try to make qmake to use TARGET_SYS instead the expanded value Sep 10 10:18:55 ericben: ohh indeed Sep 10 10:19:28 otavio: I tried using TARGET_SYS but from memeory that fails because when in nativesdk we dont have references to armv5te in my case Sep 10 10:20:03 ericben: it won't work completely Sep 10 10:20:21 ericben: http://paste.debian.net/129183/ Sep 10 10:20:28 ericben: this is my generated qt.conf Sep 10 10:20:33 oh maybe I don't talk about the same thing, I spent several hours on this an my brain is quite fuzzy Sep 10 10:20:53 ericben: with this I've been compile many complex applications Sep 10 10:21:33 otavio: what won't work ? Sep 10 10:22:03 ericben: you don't include the files from the targetsys neither the libraries Sep 10 10:22:19 ericben: the path has sysroot for native Sep 10 10:22:32 ericben: when you really want to use and link with target Sep 10 10:22:37 otavio: in fact qmake is using OE_QMAKE in the makefiles Sep 10 10:23:03 I mean OE_QMAKE_* variables in the makefiles Sep 10 10:23:17 ericben: ?? Sep 10 10:24:38 otavio: and in the end in the Makefile there are references to /usr/local/angstrom-eglibc-x86_64-armv5te/sysroots/armv5te-angstrom-linux-gnueabi/ Sep 10 10:24:58 hi ericben Sep 10 10:25:02 hi woglinde Sep 10 10:25:10 ericben: yes but this doesn't work for cmake integration Sep 10 10:25:19 otavio: maybe Sep 10 10:25:24 ericben: cmake uses qmake -query to discovery the right paths Sep 10 10:25:39 otavio: but actually without this patch a simple qt helloworld doesn't cmpile Sep 10 10:25:48 ericben: the FindQt4.cmake module uses it Sep 10 10:25:50 ericben: yes Sep 10 10:25:59 ericben: because you fixed most paths Sep 10 10:26:04 ericben: but not all Sep 10 10:26:14 so maybe we need this patch + soem more hacks to get the right qmake -query Sep 10 10:26:20 or add qt.conf Sep 10 10:26:36 ericben: but in fact I think it does require a qt.conf Sep 10 10:26:55 ericben: I will prepare a patch for it Sep 10 10:27:07 ok, the best thing would be that you send a patch adding qt.conf on top of my actual patch which fix mkspec & plain qt apps compilation Sep 10 10:27:23 so that we will have qt app + qtcreator integraton + cmake working Sep 10 10:28:19 btw I'm testing a patch which add diginotar blacklist, and an other one which updates qt to 4.7.4 Sep 10 10:28:30 but compile time is quite long to test all that ;-) Sep 10 10:28:44 ericben: I think qtcreator will work even better with the qt.conf Sep 10 10:28:56 ericben: since it will have the full paths available Sep 10 10:29:15 ericben: NICE :-D Sep 10 10:30:32 otavio: waht is strange is that in the present case qtcreator find all the tree under sysroots/armv5te-angstrom-linux-gnueabi but it can't get the path from qmake -query so I think it is using export QMAKESPEC= or export OE_QMAKE_LIBDIR_QT or export OE_QMAKE_INCDIR_QT to get the right paths Sep 10 10:30:59 ericben: yes Sep 10 10:31:22 ericben: I spent a long time to understand why cmake used to fail ... this all is quite confusing Sep 10 10:31:25 ericben: hehe Sep 10 10:31:54 otavio: http://paste.debian.net/129185/ here is the env file which adds the missing path (but doesn't fix the cmake problem) Sep 10 10:32:04 ericben what was the workaround for udev while kernel.org is still dead? Sep 10 10:32:10 otavio: yes I remember we already discussed that concerning .dev Sep 10 10:32:42 git cherry-pick 9075f7559b212cb89db69d8854b48124751e6f80 Sep 10 10:32:44 args Sep 10 10:33:03 woglinde: in meta-oe/recipes-core/udev/udev_173.bb replace URI with SRC_URI = "git://mirrors.arsc.edu/kernel/scm/linux/hotplug/udev.git;protocol=http \ Sep 10 10:33:28 okay Sep 10 10:33:29 thanks Sep 10 10:34:16 that's incredible how kenrel.org's absence is breaking things ! Sep 10 10:34:32 indeed Sep 10 10:35:30 and android development is difficult without git.android.kernel.org Sep 10 10:36:19 and and our git recipe fetching from kernel.org are dead. A git mirror functionnality in git fetcher may be a good thing for the guture Sep 10 10:36:23 future Sep 10 10:37:25 ok see you later, will go to eat and take the sun while there is some (and while qt compiles;-) bye Sep 10 10:37:25 yes Sep 10 10:37:37 ericben: thanks :-) Sep 10 10:37:42 ericben: enjoy the sun Sep 10 18:26:18 the checksum of pax-utils from meta-oe seems to be wrong, i downloaded it twice Sep 10 18:38:36 Can someone try to reproduce a built failure in e2fsprogs? Sep 10 18:56:52 otavio: I can try Sep 10 18:57:12 ericben: take a look on ml Sep 10 18:57:21 ericben: i sent more details Sep 10 18:57:32 ericben: xserver-xorg also fails in the same case Sep 10 18:59:12 otavio: I just saw the message Sep 10 18:59:19 ericben: supposely it ought to work Sep 10 18:59:37 otavio: what fail ? build or packaging ? Sep 10 18:59:44 ericben: linking Sep 10 18:59:50 ericben: in both cases, linking Sep 10 19:00:02 ericben: end of compile task Sep 10 19:03:48 otavio: this seems to be a OE failure configure --prefix=/opt/foo; make works in e2fstools Sep 10 19:03:52 out of OE Sep 10 19:04:43 otavio: I think you should provide the failure log to the ML so that oe-core/yocto experts can have a look **** ENDING LOGGING AT Sun Sep 11 02:59:56 2011