>> It is severely, deeply, horrendously obnoxious for the library
>> function [setreuid] to print to stderr!

> i agree that it's ugly to do so.  however, i'm the person who
> perpetrated the "crime," in this case, and i think i'll explain why:

> The current version of setreuid(), which was beaten on so that the
> setuid(), etc., calls conform to the POSIX_SAVED_IDS proposed mod to
> POSIX's def'n for those functions, does _not_ work correctly.

While we're on the subject, does anyone care to say why netbsd seems to
consider POSIX worth conformance just for the sake of conformance?
That is, why are people breaking traditional usage simply to make
something work the way POSIX says it should?  I was disappointed and a
bit upset to see that netbsd had deprecated (syscall interface) or
dropped (stty(1) output) the Berkeley tty interface.  I was
disappointed to see that netbsd had actually implemented the POSIX
session insanity instead of just doing process groups right.  And now
setreuid is broken for no visible reason but to make some other call
conform to POSIX, though why letting setreuid do what it's supposed to
do - set the real uid - could be felt to break POSIX conformance for
some other call I can't see.

					der Mouse

			    mouse@collatz.mcrcim.mcgill.edu

------------------------------------------------------------------------------