Op 19 apr. 2013, om 14:59 heeft Richard Purdie <[email protected]> het volgende geschreven:
> On Mon, 2013-04-15 at 08:59 -0400, Colin Walters wrote: >> On Mon, 2013-04-15 at 13:22 +0100, Richard Purdie wrote: >>> On Mon, 2013-04-15 at 08:12 -0400, Colin Walters wrote: >>>> On Mon, 2013-04-15 at 12:31 +0100, Richard Purdie wrote: >>>> >>>>> It will make our builds work again for now until the next time someone >>>>> upgrades libpng and and then it will potentially silently start using an >>>>> old version in some builds :(. >>>> >>>> So something like the attached on top of the previous patch? >>>> >>>> If this looks good I'll push both to master. >>> >>> Looks good to me, thanks! >> >> Both pushed to git, thanks. > > Much appreciated! > >> <nag> >> If OE was building from git as it should, your original patch could have >> been a ready-to-upstream "git format-patch" style instead of the default >> "Upstream-Status: inappropriate" non-git patches =/ >> </nag> > > Noted, although I doubt we'll be using git SRC_URIs for everything any > time soon for a variety of reasons. I would love to see the option of > building most things from bleeding source but that would require a lot > of work which we don't have the bandwidth for. And would introduce some dependency hell (gtk-doc-native needed for every little gnome thing *and* udev) but more importantly it will expose us to that stupid, stupid, stupid practice of using git submodules to drag in autoconf stuff. Gstreamer is the worst offender, but there are a lot of others. I try to have recipe build from a git tag instead of a tarball whenever possible, but it hurts a lot if it's one of that gtk-doc users, you can't disable that completely. regards, Koen _______________________________________________ Openembedded-core mailing list [email protected] http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-core
