Let me put it this way. If I do a pkgsrc bulk build, what do I build it on?
Previously there was the XServe platform that I could have in a rack along my other infrastructure. They ditched that. The old MacPro's you could install in a rack, though they were massively overpriced for what it was a purely CPU-bound job. Now the new MacPro's are not rackable any more.
Mac Mini's were never great, but they were cheap enough you could buy a few and small enough that you could put somewhere, although it sucked.
Again, you're missing the point. Mac's are built for workstation style development. The distributed, shared infrastructure that's en vogue today isn't a factor if you're building iOS & OSX apps. And, if you dedicate the time and infrastructure, you can build those up to stronger processes with Xcode and it's distcc.
And where does the software that powers the workstations come from? Out of thin air? Where do you compile it? I am not talking about iOS and OS X bundles, I am talking about fundamental Unix system software (which pkgsrc provides). And what if you do continuous integration? Where do you run that?
Lately I've been playing with a setup that builds a cross-clang on FreeBSD to cross-compile OS X software using the libraries inside the XCode sdks. The setup itself is working reasonably well, but most autoconf scripts don't work well with cross-compilers, so it's not workable for pkgsrc right now (though it is workable for my own C software). Unfortunately, this would not help with integration testing...
Previously there was the XServe platform that I could have in a rack along my other infrastructure. They ditched that. The old MacPro's you could install in a rack, though they were massively overpriced for what it was a purely CPU-bound job. Now the new MacPro's are not rackable any more.
Mac Mini's were never great, but they were cheap enough you could buy a few and small enough that you could put somewhere, although it sucked.