**** BEGIN LOGGING AT Fri Jul 14 03:00:04 2017 Jul 14 03:22:37 Can anyone help me with adding ISC DHCP server to yocto image? I'm setting IMAGE_INSTALL_append = " dhcp" in my image class. It fails to build with error: " * opkg_prepare_url_for_install: Couldn't find anything to satisfy 'dhcp'. Jul 14 03:22:38 " Jul 14 07:05:28 rburton_: technically not oe specific, but the fail case behavior that it fixes was specific to mingw Jul 14 07:06:09 sorry /oe spec/mingw spec/ Jul 14 07:11:03 paintenzero: http://cgit.openembedded.org/openembedded-core/tree/meta/recipes-connectivity/dhcp/dhcp.inc suggests the server is in dhcp-server Jul 14 07:15:40 hello guys.. Has anyone a suggestion on how to mount also the boot partition on the target after booting ? Jul 14 07:19:30 fberg: besides adding it to the fstab you mean, or what? Jul 14 07:31:03 hello Leto Jul 14 07:31:16 neverpanic: thanks! That worked! Jul 14 07:31:31 running df -h i can't see the boot partition Jul 14 07:32:15 but running fdisk -l I can see the two memory blocks related to the two partitions: boot and rootfs Jul 14 07:33:29 running cat /etc/fstab there is a commented line related to the boot memory block Jul 14 07:34:19 Sounds like you want to uncomment this line. Jul 14 08:13:42 rburton_: have you encountered "msgfmt: cannot locate ITS rules" in combination with meson? Jul 14 08:14:07 There's https://github.com/mesonbuild/meson/issues/1565 about it, but I have no idea how that applies to building with meson.bbclass. Jul 14 08:28:00 rburton_: I think I understand - we need /usr/share/gettext/its/polkit.its in recipe-sysroot-native's /usr/share. It's part of polkit. But depending on polkit-native is overkill (and not possible at the moment). Jul 14 08:28:22 Perhaps OE's gettext-native should provide it? Jul 14 08:46:36 RP: I noticed that you ended up as assignee for the sstate fetcher issue (https://bugzilla.yoctoproject.org/show_bug.cgi?id=11782). I missed the part of bug triage where that was discussed. Do you actually plan to work on that or just figure out what to do? With some guidance on how this should be solved I can try to work on it. Jul 14 08:46:36 Bug 11782: normal, Medium+, 2.4 M3, richard.purdie, NEW , sstate.bbclass: handle networking errors better Jul 14 08:48:26 If my first impression is right, the current approach is to create one TCP connection for each HEAD request, instead of using HTTP keep-alive. Is there a technical reason for that? It seems inefficient to me. Jul 14 08:57:23 There are many articles about how to add options in .config files in linux and busybox using append classes and .cfg files. But how do I unset options? Jul 14 10:17:43 pohly: i was under the impression that it used keep-alive and connection recycling Jul 14 11:08:03 rburton: then I need to check a bit more. Jul 14 11:10:09 rburton: about the meson/gettext/polkit.its issue: should gettext perhaps be invoked so that it searches for .its/.loc files both in the recipe-sysroot-native and recipe-sysroot? fwupd correctly DEPENDS on polkit, so the files should be installed in recipe-sysroot. How is that handled for non-meson-based recipes? Jul 14 11:10:41 not aware of this issue, what's up? Jul 14 11:11:11 rburton: see backlog ^^^ "msgfmt: cannot locate ITS rules" Jul 14 11:12:50 yeah it should search both i guess Jul 14 11:12:58 no idea how/if that works with autotools Jul 14 12:44:15 how are the patchwork commit checks implemented? I wonder if it would be difficult to make git push hooks out of the most common mistakes Jul 14 12:57:28 jku: yeah not a bad idea actually. repo is patchtest on git.yocto. Jul 14 12:57:51 or if not commit hooks, create-pull-request can invoke the tests Jul 14 13:06:57 yeah it does seem doable... Need to clone two repos and install multiple dependencies so not the simplest thing ever Jul 14 13:48:33 rburton: do you know how GFileAttribute determines the file attributes? It does not work as expected for a .xml.gz file, returning standard::content-type: text/plain instead of application/gzip as expected by appstream-glib. Jul 14 13:48:57 The result is the same as in https://github.com/hughsie/fwupd/issues/38 -> fwupd fails to parse the update xml file. Jul 14 13:51:08 strace to the rescue.... gvfs-info looks for /usr/share/mime which doesn't exist. Jul 14 13:51:31 Any idea anyone which recipe in OE provides this? Jul 14 13:52:27 It's not libmagic. "file" works alright. Jul 14 13:57:35 shared-mime-info looks promising... Jul 14 14:06:51 Yep, that's it. Jul 14 14:09:51 yeah, shared mime info, glad you got there :) Jul 14 14:10:05 there should be a recommends on vfs for that Jul 14 14:10:19 gvfs, even. if there isn't please submit a patch. Jul 14 14:11:10 I had installed gvfs and glib, of which neither pulled in shared-mime-info. Jul 14 14:11:56 Should the recommends be for glib (because g_file_info_get_attribute_string needs it) or gvfs? IMHO it should be glib. Jul 14 14:13:05 glib is a bit keen. gvfs definitely wants it Jul 14 14:13:19 Someone must have noticed this before - RDEPENDS_glib-2.0-ptest contains shared-mime-info. Jul 14 14:13:19 maybe a recommends on glib will be okay, its not a huuuge package Jul 14 14:14:04 Let's make it a recommends of glib-2.0. I'd prefer to have it working by default. Jul 14 14:16:01 yeah fine by me Jul 14 14:18:15 https://bugzilla.yoctoproject.org/show_bug.cgi?id=11792 Jul 14 14:18:16 Bug 11792: normal, Undecided, ---, ross.burton, NEW , glib: RRECOMMENDS shared-mime-info Jul 14 14:19:02 can you send a patch too? :) Jul 14 15:16:48 RP: i'm assuming sstate cleans out the inputdirs before running the real task? Jul 14 16:08:49 I'm trying to do a clean install using rpms on the target, but there seem to be some circular dependencies. For example busybox postinst uses update-alternatives which is a script which calls sed which is provided by busybox. Jul 14 16:13:03 there will always be circular deps in the base system like that Jul 14 16:13:15 if you're basically bootstrapping then you need to be careful Jul 14 16:32:12 Just wondering what others do in the situation where a 'bootstrap' install is required. Repairing a damaged install in the field, for example. Jul 14 16:33:37 you'd have to use force arguments to ignore the dependency in one direction before you satisfy it with the other Jul 14 16:34:00 in this case it's tougher, of course, since busybox is a requirement for basically everything Jul 14 16:44:50 do_rootfs seems to get around this by providing a path to the sysroot in its wrapper script Jul 14 16:57:51 RP: I wish error.yp.org was able to display such details as https://karkhaz.github.io/tuscan/ Jul 14 16:58:13 e.g. https://karkhaz.github.io/tuscan/vanilla/bonnie++.html Jul 14 16:58:44 clickable errors, surfing is so easy and time saving Jul 14 17:30:52 AB maintenance is complete. Jul 14 18:49:25 rburton: ;-) Jul 14 19:01:09 thanks halstead Jul 14 19:09:16 Thank you for the maintenance window rburton. :) Jul 14 20:18:21 hi, i'm trying to build two kernel modules which depend on a third. is there some provision for this? i don't see anything that pops out from reading module.bbclass Jul 14 20:18:36 should i just add a do_compile prepend step and copy the module.symvers to the right place? Jul 14 20:18:47 and then set up the DEPENDS so that the common module gets built fast? Jul 14 20:18:48 first? Jul 14 20:19:01 couldn't find a readily available example in the repos **** ENDING LOGGING AT Sat Jul 15 03:00:04 2017