**** BEGIN LOGGING AT Thu Mar 22 03:00:03 2018 **** BEGIN LOGGING AT Thu Mar 22 03:33:06 2018 Mar 22 10:47:04 Is there something in yocto to help me create systemd mount units? Mar 22 10:47:29 Today they are generated from my /etc/fstab, but I need the have something I can add-wants on Mar 22 13:44:26 there's the "%" character for defining that a bbappend is applied to any bb recipe version. but is there something for making sure the bbappend file is applied *up to* a certain version? Mar 22 14:03:42 dv_: % can be used like 2.% Mar 22 14:04:27 but this would work only for 2.x , right? not 1.x ? Mar 22 14:10:14 yes Mar 22 14:10:31 no, you can'd do foo_<2.%.bbappend Mar 22 14:52:52 rburton: but perhaps I can do it in the recipe? can I do for example, to add a packageconfig[foo] = "--enable-foo,--disable-foo,foo" line only if the version is 5 or newer Mar 22 15:37:42 dv_: you *can*, yes Mar 22 15:37:47 should you? don't know. Mar 22 15:38:17 get PV, compare, d.addVarFlag() if required Mar 22 16:24:58 writing another recipe is simpler Mar 22 16:26:31 rburton: I am seeing a wierd behaviour with qemu imagetest where its trying to check if system is booted. It is booted fine but somehow the mechanism to grok this info is not working and it things system is not booted and bails out. This is using qemu from git/master Mar 22 16:28:42 RP and I talked about it and he mentioned it could be the serial consoles we open to poke at system Mar 22 16:29:02 I wonder if we are carrying local patches for qemu to support this Mar 22 16:29:14 don't think we are, no Mar 22 16:29:22 if you're using systemd not that master is a bit broken Mar 22 17:12:15 its sysvinit Mar 22 18:09:41 rburton: i'm getting a "xorg-abi-" insane warning and not sure how to remedy it Mar 22 18:10:02 Package xf86-video-armsoc-meson contains Xorg driver (armsoc_drv.so) but no xorg-abi- dependencies [xorg-driver-abi] Mar 22 18:12:01 oh... rdepends Mar 22 18:14:37 hmm... still not sure what to put in there, there are no xorg-abi-* recipes --> http://layers.openembedded.org/layerindex/branch/master/recipes/?q=xorg-abi Mar 22 18:24:03 rburton, distcc recipes, are you seeing failures in master ? Mar 22 18:24:20 * armpit branch name changes 24 days ago Mar 22 18:25:00 * armpit wonders if there is a patch yet Mar 22 18:26:24 * armpit nope rocko only, master moving to 3.3 Mar 22 18:41:16 tlwoerner: xorg-abi- are provided by other recipes, not recipes themselves, iirc, though you'd think the layer index would also search provides.. Mar 22 18:47:52 kergoth: but xorg-abi's don't appear as packages, i can only find them in pkgdata/runtime files (i've never been here before) Mar 22 20:12:00 armpit: master isn't taking the upgrade so i just picked your fix Mar 22 20:12:58 tlwoerner: xorg-abi things are all rdepends/provides, cant remmeber how it works now as it was years ago but basically its to stop you running a driver against a xserver with incompatible ABIs Mar 22 20:27:01 rburton: the "magic" was to 'require recipes-graphics/xorg-driver/xorg-driver-video.inc', what happens after that is a mystery, but the check goes away Mar 22 20:38:44 tlwoerner: _add_xorg_abi_depends runs and does Magic Mar 22 20:38:58 tlwoerner: (adds rdepends based on the abi version of the xserver that it linked to) Mar 22 20:42:35 just wish i could remember why this was a thing Mar 22 20:43:07 well i remember a binary driver we used to explicitly mark with the right abi version so we didn't build images with incompatible xserver/driver pairs Mar 22 20:53:58 rburton, distcc will need a v2 Mar 22 21:40:13 are there any wic kickstarter files for flip-flop rootfs Mar 22 22:03:51 khem: https://github.com/updatehub/meta-updatehub-raspberrypi/blob/master/wic/updatehub.rpi.wks Mar 22 22:04:27 i've seen it done for use with swupdate too, not sure if it's in a public layer offhand though Mar 22 22:06:23 is it possible to specify a PROVIDES conditionally on a PACKAGECONFIG? Mar 22 22:06:54 of course, if you write the inline or anonymous python to do so Mar 22 22:06:55 DEPENDS = "${@bb.utils.contains{'PACKAGECONFIG', 'abc', 'virtual/abc', '', d)} Mar 22 22:08:06 kergoth: i'm just concerned about the timing of when the full PACKAGECONFIG will be known (i.e. a user can edit a PACKAGECONFIG from conf/local.conf, will that occur too late?) Mar 22 22:09:12 .conf files are parsed long before any recipes, by definition Mar 22 22:09:49 ok, even all the += and _appends and _remove's ? Mar 22 22:09:54 most modifications would use _append or overrides, both of which are processed at variable expansion time Mar 22 22:10:07 ok, thanks Mar 22 22:10:14 i.e. when it's used, which is after parsing is complete in the case of DEPENDS/PROVIDES Mar 22 22:13:16 oops, in my example i wrote 'DEPENDS = ...' that should have been 'PROVIDES = ...' :-) Mar 22 23:01:00 you could us packageconfig syntax to specify depends and rdepends Mar 22 23:33:47 khem: haha, i meant PROVIDES not R/DEPENDS :-) Mar 23 00:58:50 khem: the question with shallow-since is whether we make the main clone in downloads/git2/ shallow, or put it elsewhere. if the latter, adds complexity, if the former, need to remember to fetch —unshallow when a rev is MIA. leaning toward the former Mar 23 00:59:38 hmm Mar 23 01:24:06 kergoth: I think we need to prepare main clone all the time or atleast have a way to do so Mar 23 01:24:48 i expect we'd create it as shallow if and only if the full clone doesn't exist, and then still create a shallow mirror tarball for mirror population from there Mar 23 02:21:54 Hmmm does _remove honor regexps Mar 23 02:22:14 kergoth: yeah that seems right **** ENDING LOGGING AT Fri Mar 23 03:00:03 2018