From: Waider (waider at domain waider.ie)
Date: Tue 25 Jun 2002 - 10:10:31 IST
Philip Reynolds wrote:
> I don't understand why you think complying to POSIX standards some
> of the time, when it suits, is a good idea. Fair enough if you think
> POSIX compliance is an unworthy goal of UNIX and it's variants
> nowadays, but some of us differ.
I think it is an unworthy goal if POSIX compliance means the inclusion
of a pair of binaries that replicate existing functionality for (again,
hearsay) the gratification of a possibly long-dead Unix version. I have
not come across these executables on any of the systems I've worked
with, and would expect people writing portable shell scripts to use the
"ln" and "rm" commands to do what "link" and "unlink" are purporting to
do. Prior to Padraig's mail, I didn't even know "link" and "unlink" were
/supposed/ to exist, whereas I learned about "ln" and "rm" in my first
two days with a Unix system.
While I think that following standards is generally a good thing (I'm an
engineer, after all, and appreciate the value of knowing your
environment conforms to some known level), slavishly following
badly-written ones in areas like this isn't doing anyone any good except
perhaps providing gratification to someone that they can truthfully tick
the box marked "POSIX" on their feature sheet. I really think this sort
of thing should be left on the release notes as an area of non-compliance.
FWIW, some other random part of my hearsay is that parts of POSIX
concerning the shell are actually impossible to comply with because
they're mutually exclusive. I don't find this surprising, because I know
that the ICCCM spec (part of X) /does/ have this problem.
In summary, standards-following is good, but blindly following a
badly-written standard is bad, especially if it gains you nothing more
than a check-box.
-- waider at domain waider.ie / Yes, it /is/ very personal of me
This archive was generated by hypermail 2.1.6 : Thu 06 Feb 2003 - 13:17:29 GMT