I think almost everyone but Tim missed the main point.

Although it does matter what the standards say...and I've quoted Posix &
SUSV2 a lot on these lists...I use 1003.1b as a wristpad...the important
point is that absolutely everyone else seems to define this.

I just did a quick check on Solaris, IRIX, SunOS, FreeBSD, Digital Unix,
an older IRIX, an older DUh, and I even tried VAX Ultrix and NetBSD/i386.

Every last one of them defined "unix".

It doesn't matter how many nice abstract arguments are valid for driving
on the left side of the road. If everyone else drives on the right side
that's what you have to do. (If you are in the UK or Japan or somewhere
else where this is reversed, OK, YKWIM.)

Please note, everyone has made perfectly valid points, but I don't see how
any of them matter. The decision has been made and it's just too bad if the
standards did not codify the existing practice.

The arguments about name space pollution, for example, are unbelievably
bogus as applied to this case. Sure, it's bad in general. Sure, it's
against the standards, but in _this case_ the program _has_ to be OK in
the face of a unix define, because everyone does it. "Think." Every program
everywhere must be able to handle a unix define. So why not do it?

I'm definitely planning on defining unix for port-alpha, and I'm the one
you have to convince if you don't like this. I'm listening.

--Ross