> I completely understand why this is not happening. But a lesser goal would be > in the category “I'm sorry, Dave, but I can't do that.” IOTW, when the > dependencies of packages are in conflict, it is better to get an error > message than an apparently successful build that doesn't work. This does > happen with conflicting direct dependencies, does it not? If package A needs > gtk2 +quartz and package B needs gtk2 +x11, you will not be able to install > both A and B at the same time? (At least not without encountering numerous > warnings and using -f to get past them.) But when these dependencies are > indirect, life gets complicated, and the conflict hard or impossible to > detect. (Though I am still not sure I quite understand why.) > > – Harald
I agree that a warning/error when an install would conflict would be nice. But I am confused about the underlying problem. 1. The X11 and Quartz versions of the "either/or" installs either place or use files at the same locations, and you cannot have two different files at the same place (well, you *CAN*, in some unix flavors; this is typically found in multiple machine OS's that have a hidden directory that looks like a file. Mac OS does not support this.) 2. But there are some ports that work with both flavors, and adapt -- so they have a way to use/provide both kinds of service (gtk I think was one of them) -- so some libraries can provide both services ... but one file? Am I misunderstanding something here? --- Beyond that, it sounds like the x11 vs quartz system would need to check all installed thingies that use one or the other, and give a warning if there are any of the wrong flavor installed, and either a "here's how to switch them all to this flavor" or "some cannot be switched to this flavor". === My big question: Why are two different API's -- the quartz and the x11 libraries -- being installed in the same location?!? --- Entertaining minecraft videos http://YouTube.com/keybounce
