On Tue, 2021-02-09 at 16:37 +0100, Tomasz Dziendzielski wrote: > > Thanks for the answer. > > > I've been giving this some thought and I really don't like having > > to > > remember to manually add special cases like this, it sounds/looks > > like > > something we should really fix in a better way. > > The other way that comes to my mind would be to loop over all > dependencies and check if they inherit image.bbclass which would > increase build time. Another problem is that we don't have direct > access to other recipes' metadata, so we would need to prepare such > mechanism.
That isn't something we want to support. > > Perhaps we should just skip license files if they don't exist? We > > could > > perhaps show a warning if the match isn't for a recipe called *- > > image? > > There might be recipes inheriting image.bbclass that don't have > "image" in its name (which is also my case). I can prepare the change > to just print a warning instead of error if the recipe can't find a > license manifest for any recipe, not only the ones called *-image. > Unfortunately that way we could not catch missing manifests with > recipes that should provide them. Would such change be accepted? I think the warn/error issue isn't the real solution, "good" builds shouldn't show warnings and this case would. I guess the other option is to enable populate_lic for image recipes rather than deleting the task, then there would be some kind of license information available. Not sure whether that makes sense or not (and whether it should be the image recipe license or the sum of the licences making up that image is a secondary difficult question)... Cheers, Richard
-=-=-=-=-=-=-=-=-=-=-=- Links: You receive all messages sent to this group. View/Reply Online (#147866): https://lists.openembedded.org/g/openembedded-core/message/147866 Mute This Topic: https://lists.openembedded.org/mt/80377785/21656 Group Owner: openembedded-core+ow...@lists.openembedded.org Unsubscribe: https://lists.openembedded.org/g/openembedded-core/unsub [arch...@mail-archive.com] -=-=-=-=-=-=-=-=-=-=-=-
