**** BEGIN LOGGING AT Tue Mar 20 03:00:07 2018 Mar 20 09:23:31 which task is responsible for building dependencies found in DEPENDS ? Mar 20 09:24:47 wakko: no task at all, given the nomenclature that a task is one of the things in listtasks for the recipe. Mar 20 09:24:56 the dependencies and task ordering is done by bitnake. Mar 20 09:25:15 bitbake, even. Mar 20 09:26:29 because i wrote an independent image class added a task and placed noexec on everyother tasks, and it looks like DEPENDS is ignored Mar 20 09:27:22 hmhm Mar 20 09:27:31 maybe i should do something like do_mytask[depends] = "${DEPENDS}" ? Mar 20 09:27:36 i would expect it to be honoured, yes. Mar 20 09:27:44 with :do_build on every thing Mar 20 09:28:19 probably if there is nothing to build or compile, it is assumed that there as no dependencies needed. but thats pure guesswork. Mar 20 09:36:25 apparently it's honoured, but not with my overrides, stuffin DEPENDS in machine conf and it's done :) Mar 20 11:34:59 hi. anyone got info on how to set up an oe tree for generating gtk-doc for packages? Mar 20 12:50:43 davidsf: if you're using a modern release then add api-documentation to DISTRO_FEATURES and ensure the recipe that is using gtk-doc, inherits gtk-doc. Mar 20 12:53:32 rburton: ok, thanks. I will try that Mar 20 15:31:27 rburton: on my system, gtk-doc works for glib-2.0 but not for gstreamer1.0 Mar 20 15:31:38 for gstreamer, pkg-config-native is not used in configure when checking for gtk-doc Mar 20 15:31:50 (it is used for glib, I see there is a patch for it in gtk-doc recipes) Mar 20 15:31:57 is this a known problem for gstreamer? Mar 20 15:32:11 master? rocko? old release? Mar 20 15:32:34 (works in master for me) Mar 20 15:32:38 eg /usr/share/gtk-doc/html/gstreamer-1.0/GstControlSource.html Mar 20 15:36:41 rburton: old release. pyro I think Mar 20 15:39:06 yeah probably wasn't expected to work there. gst ships a gtk-doc.m4 we remove now, you'll need to backport that Mar 20 15:39:56 ah, ok. thanks Mar 20 16:39:20 khem: it looks like github.com/riscv/meta-riscv is more up-to-date than github.com/kraj/meta-riscv? Mar 20 17:12:22 tlwoerner: yes thats the new home, its official place for all riscv Mar 20 17:12:54 khem: okay, thanks some people on #riscv saw your talk at ELC and were/are asking questions Mar 20 17:13:14 I will turn kraj repo to a fork as soon as the freedom-u500 patches are reviewed and proposed to official layer Mar 20 19:13:27 hmm, anyone know why -cross tend to have PN .= -${TARGET_ARCH} while cross-canadian have -${TRANSLATED_TARGET_ARCH}, when the only difference is - vs -? Mar 20 19:13:38 guessing they should probably all use the latter to avoid issues with overrides Mar 20 19:16:50 kergoth: I've wondered that too Mar 20 19:19:55 i'm guessing it's just a legacy remnant we should fix up Mar 20 19:21:25 this is.. odd. i'm getting qemu illegal instructions with gobject-introspection.. but the machine is qemux86. if anything can run qemu, it should be that! Mar 20 19:21:26 hah Mar 20 19:37:34 kergoth: qemu trying to run eg ssseeeee3 instructions is is emulating a pentium1? Mar 20 19:37:56 that's the usual problem, the bsp can control what qemu options are passed Mar 20 19:44:47 would hope qemux86 woudn't be guilty of setting the wrong qemu options, but i suppose it's possible Mar 20 20:09:44 kergoth: if you can replicate then the resulting core dump should tell you what instruction was failing Mar 20 20:11:36 kergoth: QB_CPU is the variable fwiw, QB_CPU_x86 = -cpu pentium2 Mar 20 20:11:55 i *think* there was a bug where that and the compile flags didn't match Mar 20 20:12:20 kergoth: do you have oe-core 20b3574749420a1fef2cb2e0579584453dd4c5c5? Mar 20 20:13:40 yep. weird. disabling gobject-instrospction-data and will revisit later. **** ENDING LOGGING AT Wed Mar 21 03:00:02 2018