On Sun, Mar 1, 2015 at 8:35 PM, Saleem Abdulrasool <[email protected]> wrote: > I think you missed one of the core points of all this work. This is > intended to remove multi build for the C class. The use of multi build will > persist, but for python, ruby, (perl?) ABIs, not the C class.
Just imagine I wrote multibuild_c there. > Does something like `cave resolve -m cross -E i686 wine' not work for that? > Yes, we would need to enhance the current label handling to understand > cross. However, given that the previous should allow you to build a > cross-compiled wine (as an example), I think that we can live with this for > the short term. But all the build+run dependencies need to be cross compiled too and whenever I did that, it installed wine to ::installed-i686 and everything else to ::installed. I may be grossly misunderstanding/misusing -E, but there doesn't seem to be _any_ documentation for it afaict. (well not actually wine, but it's the same for everything that has dependencies) > I think that we can have a slightly lower bar for the merge. My intention > here is two fold. One, it reduces the barrier to entry, allowing more > people to contribute towards it. cross has languished a bit as of late, and > I would like to see that change. The second objective is to reduce the > overhead of maintaining cross. There were issues that were introduced into > cross due to merging. Having to maintain paludis and arbor is a fair amount > of work. There is no reason to not also branch all the other repositories, > which only increases the amount of effort required to keep everything in > sync. Agreed on all points. If the above actually works I'm fine with merging. I just don't want to break Exherbo for anyone who uses multibuild_c: 32 to get skype, wine or something else to work, as I imagine that that would upset some ;-) Cheers Benedikt _______________________________________________ Exherbo-dev mailing list [email protected] http://lists.exherbo.org/mailman/listinfo/exherbo-dev
