405 2010-10-07 08:57:07 +0000 Add a mechanism to do pre-build sanity based on specific configurations 2026-07-02 15:27:32 +0000 1 1 7 Build System, Metadata & Runtime OE-Core core 0.9 x86 Multiple ACCEPTED Medium enhancement 6.99 2407 1 mark.hatle unassigned gzatko jessica.zhang kevin.tian leonardo.sandoval.gonzalez liezhi.yang mariano.lopez meta.mr.watcher meta.watcher randy.macleod richard.purdie sgw song.liu tim.orling --- --- Yes (doc changes required) oldest_to_newest 10196 0 mark.hatle 2010-10-07 08:57:07 +0000 It would be nice if the recipes involved in a build had a way to specific additional system sanity checks. This was a component using a unique host-system resource could add a sanity check that would only fail if that specific component were to be built. 16929 1 jessica.zhang 2011-10-19 18:41:26 +0000 *** Bug 1616 has been marked as a duplicate of this bug. *** 19291 2 scott.a.garman 2012-03-21 17:20:53 +0000 Fixed in poky master and OE-core: http://git.yoctoproject.org/cgit/cgit.cgi/poky/commit/?id=96dde9513c08430e12f4f1f6e6e54ca2a244bfcd 19294 3 scott.a.garman 2012-03-21 17:32:49 +0000 Sorry, that last comment was intended for another bug. 19296 4 scott.a.garman 2012-03-21 17:34:54 +0000 Mark, due to the age of this bug I've lost track of what you had in mind when this feature was first being discussed. Can you provide more details and some specific examples of these types of checks? 21194 5 mark.hatle 2012-05-08 15:46:43 +0000 What I would like is for specific recipes to implement sanity checks specific for those recipes. The sanity checks would be run very early by the normal system sanity checking class, and not deferred until normal task/recipe execution. (Note, I'm really not sure if this is possible due to the system not necessarily knowing what tasks will be executed!) If this is not possible, the next best alternative would be provide a standard sanity checking task, and a standard template to enable it, if provided by the recipe. An example of the type of check requested would be QEMU checking for the NVIDIA libGL drivers. If the user is not going to build or use QEMU, then there should be no reason to check for the libGL components and/or create a failure condition. Proposed change: * Add a do_sanity or similar task to recipes with unique host requirements * Add a mechanism for the recipe to trigger the sanity check As architecture work remains to determine the implementation, an estimate is unclear to me. I will set it to a fairly large number due to this uncertainty. 75629 6 mark.hatle 2017-08-03 19:51:02 +0000 Current idea would be a standard method that would do a simply sanity check during the anonymous python processing time. If it failed, would return an error which in turn would blacklist the package with a specific error message indicating why it won't work. 87368 7 gzatko 2020-05-31 10:57:54 +0000 @Richard, does this one make sense to you? Please add your thoughts about implementation if yes. If I understand the proposal we want to add a new standard task that would define some requirements on host machine necessary to build the recipe's packages. But it's not quite clear to me who would define a dependency graph for each of them. 106080 8 tim.orling 2026-07-02 15:27:32 +0000 Moving to 6.99.