**** BEGIN LOGGING AT Sun Mar 09 02:59:58 2014
Mar 09 19:23:17 <Jin^eLD> hi
Mar 09 19:23:49 <Jin^eLD> how does yocto decide what postinst scripts should be run on target during first boot and which ones are run during image generation?
Mar 09 19:24:58 <Jin^eLD> I have a couple of recipes where I would require that the postinst hooks are run during first boot
Mar 09 19:25:25 <nerdboy> your own recipes, or existing ones?
Mar 09 19:25:31 <Jin^eLD> my own
Mar 09 19:26:44 <Jin^eLD> I'm migrating from 1.5 year old angstrom to latest yocto dora and I noticed that it now tries to do as much postinsts as possible during image generation as opposed to how it was, where it ran them all on the target
Mar 09 19:26:50 <nerdboy> the your package_postinst should use $D
Mar 09 19:27:18 <Jin^eLD> as in if [ "x$D" != "x" ]; then exit 0; fi ?
Mar 09 19:27:56 <Jin^eLD> or how exactly?
Mar 09 19:28:08 <nerdboy> without the "if" should force it to run in the image rootfs
Mar 09 19:28:27 <Jin^eLD> hm? I think I do not fully understand you..
Mar 09 19:28:37 <Jin^eLD> you mean the presence of "$D" as such will already tell it to run it from the rootfs?
Mar 09 19:29:12 <nerdboy> one sec
Mar 09 19:33:03 <nerdboy> the postinst/postrm functions become a shell script on the target
Mar 09 19:33:25 <Jin^eLD> yes, I know that
Mar 09 19:34:18 <nerdboy> so "if test "x$D" != "x"; then exit 1" makes the "else" run on target
Mar 09 19:34:40 <Jin^eLD> ah, cool, thanks, so I should just exit 1 on the host
Mar 09 19:36:13 <nerdboy> without any else and $D set it should only run during the build since it gets marked as "installed"
Mar 09 19:36:29 <Jin^eLD> ok I get it, the exit 1 was the key
Mar 09 19:53:22 <bluelightning> Jin^eLD: it tries to run all, the ones that fail during do_rootfs on the host are deferred to first boot
Mar 09 19:54:22 <Jin^eLD> thx
Mar 09 21:11:24 <Jin^eLD> bluelightning: but now when I added the exit 1 thing it fails to build the image somehow
Mar 09 21:12:00 <Jin^eLD> https://pastebin.mozilla.org/4531692
Mar 09 21:12:14 <Jin^eLD> and the only thing I see are opkg errors caused but scripts that did not run
Mar 09 21:12:28 <Jin^eLD> or am I not allowed to fail in preinst?
Mar 09 21:13:13 <Jin^eLD> well makes sense I guess
Mar 09 21:14:15 <bluelightning> yeah preinst probably can't fail
Mar 09 23:20:55 <nerdboy> as long as it works correctly with both $D set and unset, then it should be safe without the if test, since it should always work in do_rootfs and never need to run on first boot
Mar 09 23:21:09 <nerdboy> or am i missing something?
**** ENDING LOGGING AT Mon Mar 10 02:59:58 2014