Wes Felter (wmf.delete@this.felter.org) on August 5, 2014 8:19 pm wrote:
> Patrick Chase (patrickjchase.delete@this.gmail.com) on August 4, 2014 7:51 pm wrote:
>
> > Speaking of urban legends, I've always liked the one about Apple's "super-secret" project
> > to create an x86 port of OSX (hint: It was on x86 long before they bought it).
>
> I also assumed that Apple had maintained the x86 version of NeXTSTEP/OS X all along, but then I read
> this: http://www.quora.com/Apple-company/How-does-Apple-keep-secrets-so-well/answers/1280472?srid=i1
> which makes it sound like the x86 code went unmaintained for a while until one guy secretly revived it.

The kernel and BSD userland were maintained on x86 for quite some time after the release of PowerPC Mac OS X. This was public knowledge, thanks to the Darwin open source project. Hardware support was limited, they clearly weren't aiming to do much beyond keeping it running on a few reference machines.

The layers on top of BSD are what I believe fell into a very low maintenance state. It wasn't ever 100% dead though, I remember reading something (or maybe it was part of the announcement keynote?) in which a senior exec said that Apple's internal applications groups were required to maintain bi-endian discipline all along. This put them in a good position to port the full stack, even if it hadn't been completely maintained. No doubt this is how that guy was able to revive it by himself.

However, there's a huge difference between initial bringup and the final shipping product. I'd hazard a guess that they had to put in a lot of work on two areas in particular: the Carbon APIs (descended from the Classic MacOS Toolbox, not nearly as endian neutral by design as Cocoa) and performance tuning (e.g. porting all the AltiVec code to SSE).