Usenet 1994 January/usenetsourcesnewsgroupsinfomagicjanuary1994.iso/sources/std_unix/volume.16/text0019.txt

Internet Message Format | 1989-08-11 | 1.8 KB

From: [e-mail] (rja)

( The original article and Guy's reply were in comp.std.misc, but this
  is really more a comp.std.unix article so followups are directed there.)
[ The above note is from rja.  Please send followups to one newsgroup or
the other rather than crossposting.  -jsq ]

(unclear who wrote this originally)
% I understand that the semantics of new-file creation are also in conflict.
% The SVID says that the gid of the new file is the effective gid of the
% process, but POSIX says that it gets copied from the parent directory.
% (Actually it's optional in POSIX, but FIPS requires this behavior, and AT&T
% is committed to making SVR4 comply with FIPS as well as POSIX, right?)

In article <[e-mail]>, [e-mail] (Guy Harris) writes:
> S5R4 will probably support both, in the fashion that SunOS 4.0 does - if the
> set-GID bit is set on a directory, files newly created in that directory
> will get their GID from the directory, and if it's clear, they'll get it
> from the effective GID.  One presumes they'll update the SVID to
> describe this; otherwise, while FIPS-compliant, their system may not be
> SVID-compliant....

I'm sure Guy knows what he is talking about.  In fact I'll be very surprised
if the next release of the SVID isn't 100% compliant with the NIST FIPS
variant of the POSIX (really IEEE 1003) definition.  I understand that
there are some conflicts between ANSI C and the 1003.1 definitions, but
it isn't clear to me where they disagree (I don't have current versions
of either document yet).   I suspect that the SVID will resolve those
conflicts according to the NIST FIPS as well.

It would be nice if someone at AT&T or UNIX International could
authoritatively conform or deny these guesses.

  [e-mail]

Volume-Number: Volume 16, Number 17