On May 19, 2016, at 5:50 AM, René J.V. Bertin wrote: > One could, the real question is, should we? I think a main MacPorts guideline > has always been to keep dependencies as simple as possible by not splitting > up ports even if they can. Qt has always allowed (and advised) building each > component separately, but it's only since recently that port:qt5 has actually > implemented that. I don't think that's a big advantage except for users who > install only a very limited set of Qt5-dependent ports, and IIRC someone > (Ryan?) lodged a formal-looking request not to do this on trac.
I don't think that was me. I have no problem with qt5 being split up to better match upstream's distribution model. Our guidelines used to be to include in the port by default everything that another port might expect to exist. Then the subport feature got implemented, and I don't think anyone really wrote any revised guidelines after that, so I don't know what our guidelines would be at this time. If it makes sense to split something into a subport, do it. For example, I split libnetpbm out from the netpbm port, because libnetpbm has a more liberal license than the rest of netpbm, and other ports that need to depend on libnetpbm have no need for the other parts of netpbm, so splitting libnetpbm out into its own port with its own license allowed other ports to become distributable. But I haven't understood why splitting things the way Debian/Ubuntu do with their -dev packages would be helpful to us in any way. _______________________________________________ macports-dev mailing list macports-dev@lists.macosforge.org https://lists.macosforge.org/mailman/listinfo/macports-dev
