Portability of C Programs and the Unix System (1978)
bell-labs.com
bell-labs.com
Sometime between 1978 and about 1997, people figured out how to tie C code strictly to a single architecture. About '97, I sat next to some people doing a big C and I think X11 system on HP-UX. By 2000, they were trying to port it to something else, Sun Solaris, probably. They couldn't do it. They never got this system to run on something other than HP-UX.
The arc of portability is long, but bends toward "not" I think. It's basically unheard of for people to write about the portability of anything these days, C, C++, whatever. There's Windows, and there's Linux, both 64-bit, little-endian, with a stack that grows down from high addresses, and a heap that grows up. Both have the same sort of ideas about system calls, as near as I can tell. Portability is dead.