> Note that these are all *deliberate design choices* in OpenBSD and its ports > tree, > not a limitation of the tool.
It follows the 'eat our own dogfood' principle. We only have so many machines and developers around to eat our own dogfood, so we don't do cross compilations. That would require more machines, or more people watching more machines, or looked at from the other side, it would mean less watching of the specific cases that matter the most (ie. native). > Those all come from lack of manpower with respect to expected quality of the > results. Right. We run on many architectures, because it helps improve the quality. Running via cross compilers? That's does not improve the quality of the resulting native output in any way. t might improves the quality of the cross compilation environment, or the compiler itself, but that is not where our core responsibilities lie. And anyways, it is rather apparent that those who have that as a core responsibility also have far fewer cross-targets in mind than might be useful (ie. walk off their map, and you'll step in mud).
