**** BEGIN LOGGING AT Fri Oct 07 02:59:58 2016 Oct 07 06:53:19 join #poky Oct 07 06:56:37 Hi, is this a good place to ask newbie yocto 1.8 questions? I'm trying to get this layer added: https://github.com/meta-qt5/meta-qt5 Oct 07 06:57:42 in the build dir, I give command: bitbake -c devshell qt5 Oct 07 06:58:47 and I am getting error: ERROR: ParseError at /......./ventana-yocto/sources/meta-qt5-master/classes/qmake5_base.bbclass:50: Could not inherit file classes/remove-libtool.bbclass Oct 07 07:00:00 Now I am going to "start from scratch" getting the latest yocto for gateworks Ventana, seeing that it builds, getting the qt layer again etc. Oct 07 07:00:50 but this is kinda groping in the dark, so if someone can educate me a bit about how to go about doing it, that would be helpful. Or if someone has a better idea on how to get Qt5 compiled, that'd be fine too. Oct 07 07:02:03 Background: I have the Gatworks Ventana board flashed and running and X11 working (just by getting their yocto, configuring and building it, and flashing), and now my next need is to get Qt5 on it. Oct 07 07:03:27 is there a known problem that doing a "PACKAGECONFIG +=" from a bbappend actually overwrites the PACKAGECONFIG and sets it only to the values I try to add through my bbappend? Oct 07 07:03:47 while if I use PACKAGECONFIG_append actually does the append as expected Oct 07 07:10:19 it looks like "+=" , "=+" , ".=" , "=." are broken when used with PACKAGECONFIG Oct 07 07:16:31 gartin: usually some recipes set PACKAGECONFIG ??= Oct 07 07:16:58 so if you just want to add to the defaults then use append/prepend Oct 07 07:17:12 PACKAGECONFIG_append = " foo" Oct 07 07:17:25 yeah, it works with _append Oct 07 07:17:37 but I would have expected it to work with += also Oct 07 07:17:52 there is difference between += Oct 07 07:17:58 here it will set it as well Oct 07 07:18:14 and ??= is only a fallback when nothing sets it Oct 07 07:18:19 the only difference I was aware was that _append is like .= Oct 07 07:18:46 the evaluation order is different Oct 07 07:21:43 so basically when there is a weak assignment, += will override the weak assignment whereas _append will not? Oct 07 08:07:43 good morning guys. I have created my own layer and placed there an image which requires another one. After building the image I don't see symlinks to the latest built image in the images directory. Could someone help me with this issue? Oct 07 08:39:58 Hi, i am building image meta-intel (intel-core-32) and bitbake throws an error: linux-yocto-4.4.14... do_kernel_sanity_check: does not match of kernel being build (4.4.13) please update PV variable ... Oct 07 08:40:46 where is PV variable ? Oct 07 08:43:22 Has anyone ever tried to use versioned dependencies on poky-1.8? It's described extensively in the RDEPENDS section of the Poky manual, but I cannot seem to get it to do anything. The format I followed is equal to that in the docs: RDEPENDS_${PN} = "foo (>= 1.2)" Oct 07 08:45:39 Erratum: 1.8 should be 1.6 Oct 07 08:48:41 Even relying on a bogus version for which no recipes exist does not result in unsatisfied dependencies. Oct 07 09:00:42 Rootert: *runtime*, they'll have an effect at image generation not build time Oct 07 09:03:16 rburton: Thanks, that's true. Even though the manual does not state so, I deduced from the gnutls.inc that it should be the same with the DEPENDS variable. Is that correct? Oct 07 09:04:29 gnutls.inc example can be found here: http://git.yoctoproject.org/cgit/cgit.cgi/poky/tree/meta/recipes-support/gnutls/gnutls.inc?h=daisy#n4 Oct 07 09:09:02 depends doesn't support versions, no Oct 07 09:09:08 (patches welcome!) Oct 07 09:11:08 Ok, thanks for the clarification Oct 07 09:30:48 What's the cleanest way to do install a directory and all its subdirectories? Oct 07 09:32:20 /* at the top level is not enough Oct 07 09:36:32 rburton? Oct 07 09:44:15 could somebody help me with a symlinks? Oct 07 12:11:27 ok, I have opened bitbake -c devshell Oct 07 12:11:31 I have simple test.cpp, which has empty main and single include: #include Oct 07 12:11:56 I try to compile it with command: arm-poky-linux-gnueabi-g++ -c test.cpp Oct 07 12:12:20 I get error: test.cpp:1:19: fatal error: stdio.h: No such file or directory Oct 07 12:12:46 what could possibly be the reason? why the default include paths don't include stdio.h? Oct 07 12:16:07 ah, sysroot is not defined, I guess that is the reason Oct 07 12:17:28 ok, so this works: $CXX -c test.cpp Oct 07 12:17:42 next step, how do I get Qt configure to use that Oct 07 12:26:58 hyde there should be qt recipes existing Oct 07 12:28:17 yeah, I tried it, but I got an error with it Oct 07 12:28:35 so rather than trying to patch someone elses system into working, I thought I'd do it myself Oct 07 12:29:27 copypaste from earlier today in this channel: Oct 07 12:29:39 Hi, is this a good place to ask newbie yocto 1.8 questions? I'm trying to get this layer added: https://github.com/meta-qt5/meta-qt5 Oct 07 12:29:47 in the build dir, I give command: bitbake -c devshell qt5 Oct 07 12:29:54 and I am getting error: ERROR: ParseError at /......./ventana-yocto/sources/meta-qt5-master/classes/qmake5_base.bbclass:50: Could not inherit file classes/remove-libtool.bbclass Oct 07 12:30:12 ... Oct 07 12:30:49 rob_w: ^ Oct 07 12:31:01 well i am not using devshell alot .. maybe it misses something .. basicly a bitbake qt5 should set your chain up for all following recipes you create Oct 07 12:31:19 what would it take to make image-buildinfo detect that recipe changes were finally committed, and that it matters to rebuild the image for /etc/build to get refreshed ? Oct 07 12:32:42 hyde .. maybe go with the example recipes inside this meta-qt5 Oct 07 12:33:34 rob_w: I think I'll give it a shot later, but now I'll try to get my own Qt configured anyway, since I'm quite far along with that Oct 07 12:34:42 row_w: actully, not possible. I get that ParseError above when ever I do anything with bitbake, if I have that Qt layer there. Oct 07 12:37:08 I get that if I just have htis added to conf/bblayers.conf: ${BSPDIR}/sources/meta-qt5-master \ Oct 07 12:40:49 hyde: does that mean that you're trying to use the master branch of meta-qt5 together with yocto 1.8? Oct 07 12:44:05 basically yes Oct 07 12:44:59 ernstp: if you have info (or even better, link to relevant README which I failed to find) on what I *should* use, I'd be delighted to hear it :) Oct 07 12:46:12 hyde: well I'm just thinking that you should use the 1.8 branch of meta-qt5 Oct 07 12:46:28 what is that called... fido? Oct 07 12:46:44 master would be for yocto 2.2 Oct 07 12:47:31 ernstp: not sure. poky? too many names ;) Oct 07 12:47:39 anyway, thanks for the idea, I'll try to check Oct 07 12:47:49 (what is "poky" exactly?) Oct 07 12:47:52 1.8 is old... Oct 07 12:47:58 2.1 is the current Oct 07 12:48:14 Fido Oct 07 12:49:28 1.8 is what the board vendor has available Oct 07 12:52:27 a newbie question: what exactly is the preferred method of adding this kind of layer? Oct 07 12:52:50 now I downloaded it as zip and unzipped under sources/ Oct 07 12:53:33 is it ok to just git clone under sources/ ? Oct 07 13:07:14 I use repo to clone it under sources like boundary suggests: https://github.com/boundarydevices/boundary-bsp-platform Oct 07 13:31:03 ok, I got it to start building something related to Qt :) Oct 07 13:31:40 ernstp: thanks for pointing out how the branches refer to yocto versions by name, and needing the right version's branch! Oct 07 13:32:02 hyde: np :-) Oct 07 16:07:54 RP: have you noticed that no tasks are showing progress bars ever since the elapsed time was added? Oct 07 16:07:55 or anyone else? Oct 07 16:12:02 kergoth: I'd swear I'd seen progress bars Oct 07 16:12:55 i see overall build progress bars, but haven't seen one task one since my last update Oct 07 16:12:57 * kergoth shrugs Oct 07 16:13:22 kergoth: I'm sure I saw fetch ones the other day... Oct 07 16:13:23 specifically linux-yocto:do_fetch used to show one, but no longer does Oct 07 16:13:46 kergoth: cmake's do_compile has a progress bar Oct 07 16:13:47 kergoth: it can depend on which fetch module its using, I did notice that Oct 07 16:14:06 kergoth: btw, have you any idea why this would happen: http://autobuilder.yocto.io:8010/builders/nightly-arm/builds/115/steps/BuildImages/logs/stdio Oct 07 16:14:08 also, ui event printing is screwy since the recent changes Oct 07 16:14:22 i ^c ^c and see NOTE: messages from a *task* Oct 07 16:14:36 NOTE: i686-pc-linux-gnu-objdump -p /scratch/dogwood/minnowmax-psplash-test/build/tmp/work/corei7-64-mel-linux/pulseaudio/9.0-r0/packages-split/pulseaudio-module-always-sink/usr/lib64/pulse-9.0/modules/module-always-sink.so Oct 07 16:14:36 NOTE: Checking Package: glibc-binary-localedata-hi-in Oct 07 16:14:36 NOTE: Sending SIGTERM to remaining 4 tasks Oct 07 16:14:38 kergoth: note the log corruption. The log/cooker/ file goes dead after corruption starts Oct 07 16:15:14 kergoth: I've not noticed it in general but see the above bug Oct 07 16:15:39 * kergoth tests cmake:do_compile Oct 07 16:19:46 rburton: ah, that one has a progress bar, thanks. maybe i just haven't done much fetching recently. though i do know linux-yocto isn't showing one, even though it's git.. Oct 07 16:20:56 RP: Huh, don't know offhand, haven't seen that before. I can't think of how it'd lose track of a task process like that.. that's similar to what happened when i intentionally raised an exception from TaskStarted without the added exception handling, but I can't see how that would be related, unless an event handler is freaking out Oct 07 16:20:57 hmm Oct 07 16:21:08 and in that case there'd be a traceback, so i'm fresh out of ideas Oct 07 16:38:17 kergoth: rburton and I are struggling too. We keep seeing this Oct 07 16:39:24 I'd add debug output in knotty to keep track of its process accounting. try to figure out if an event is going MIA? Oct 07 16:39:25 * kergoth shrugs Oct 07 16:44:49 "keep" is generous, "annoyingly rarely but enough to cause problems" is more accurate Oct 07 17:02:50 halstead: when you're done with autobuilder work, can you fire ross/mut on the new cluster? Oct 07 17:03:19 rburton, Yes. It will take longer than usual because I'm working on FTP. Oct 07 17:03:29 halstead: no problem at all Oct 07 19:26:50 Is there anything unique necessary to make a Yocto kernel boot on an Intel Mac? It's an 64-bit EFI-capable kernel, but there is no console output whatsoever. Oct 07 19:27:12 Bootloader (ipxe) loads vmlinuz and initramfs, and the computer halts at that point. Oct 07 19:27:38 I do have CONFIG_EARLY_PRINTK_EFI=y **** BEGIN LOGGING AT Fri Oct 07 20:22:30 2016 **** ENDING LOGGING AT Sat Oct 08 02:59:58 2016