On 16 Jun 2014, at 14:10, Alexander Pyhalov <a...@rsu.ru> wrote: > On 06/16/2014 16:40, Andy Stormont wrote: >> I have a couple of nits to pick: > > > Yes, these are two 'rude hacks'. > .cleaned rule is necessary to force re-applying patch on update. So that we > can see when patch is not applied after git pull. This also should help when > patch file is changed. Without this target it would create garbage or failed > to apply. It's because we don't do "real clean" (i.e. rm -fr $(SOURCE_DIR)) > on clean target. >
I would suggest collapsing the .%ed, .cleaned and .patched rules into one then. > And we use ".downloaded" file to find out if we pulled something new. But > ".cleaned" target will destroy it, so we recreate it to avoid illumos-gate > rebuild on each "gmake publish”. Never-mind my initial objections - I misread the diff. Looks good. Andy. > > >> - Having a .cleaned stamp seems like a bad idea. If it were really clean >> there shouldn’t be anything like that around at all. >> - The .cleaned rule depends on .downloaded but it also creates .downloaded. >> Does this still work with a clean build? >> > > > -- > Best regards, > Alexander Pyhalov, > system administrator of Computer Center of Southern Federal University _______________________________________________ oi-dev mailing list oi-dev@openindiana.org http://openindiana.org/mailman/listinfo/oi-dev