HPA> POSIX.1 does require that dev_t is an arithmetric type, which
HPA> means that a 64-bit dev_t would require that Linux permits "long
HPA> long" in the offical Linux APIs for 32-bit machines. Since "long
HPA> long" is a GCC-ism, it seems to me Linus has been avoiding making
HPA> it mandatory in user space. There are a few more issues; a 32-bit
HPA> dev_t would maintain the alignment of struct stat, which would
HPA> make backward compatibility easier to implement.
It could be done by making long 64 bit on all linux machines. I would
absolutely _love_ that, but I doubt it would be universally liked
among linux developers. But maybe as an option?
It would not break Standard C. It would not break POSIX. It would
break a lot of programs. *sigh*
Benny
-- Finger for PGP Public Key WWW: <A HREF="http://www.daimi.aau.dk/~amorsen/">Benny's homepage</A>