Rumor about AT&T Validation

[Carl S. Gutekunst]

In article <2846 at brl-smoke.ARPA> ron at BRL.ARPA (Ron Natalie) writes:
>Well, one point is that fstat doesn't work right on systems that use
>sockets to support the PIPE system call.

Um, not necessarilly. Pyramid OSx implements SV pipes using sockets, and the
SV fstat() works perfectly. (Believe me, I know; the 4.2bsd fstat() doesn't
distingush between pipes and sockets, so I had a BSD program calling the SV
fstat() so it could determine which was which....)

I really don't believe the rumor that AT&T will reject machines that have
sockets *anywhere*, but I *DO* believe the validation suite will check for
correctness on fstat().

Waving a red flag in front of the bull again, :-) I'd bet that Sun's Unix 3.2
does the fstat() correctly for SV, too, and they also have pipes on sockets.

<csg>