O'pain Software Foundation: (2) Why is it better than AT&T?

[Richard A. O'Keefe]

In article <5085 at nsc.nsc.com>, glennw at nsc.nsc.com (Glenn Weinberg) writes:
> 2) ABI's
> 	Now, you say, "well, why doesn't everyone just sign up for an ABI,
> 	then?"  The answer is simple: because AT&T wouldn't let them.
> 	AT&T alone decided which vendors it would sign ABI agreements
> 	with.

It is important to be clear about "vendors of WHAT?"
SUN are doing a SPARC ABI.
Motorola (not SUN) are doing a 68020 ABI and an 88000 ABI.
If Intel aren't doing an 80386 ABI I'll be much surprised (there
already is a de facto definition).
The relevant vendors here are the companies that make the chips.
These companies can listen to anybody _they_ want to.
You can easily get a copy of the Motorola ABI drafts and tell them
what you think.

HP and Apollo are ok as far as their 68020-based machines are concerned;
as far as their proprietary machines are concerned ABIs *would* be an
issue except that no-one else is making UNIX boxes out of thosse
machines, so they have de facto total control over "ABI" definitions
for their machines anyway.  IBM don't need an ABI for the RT PC.
If anyone were to say what a 370 ABI ought to look like, I think we'd
be better off with Amdahl deciding that.

> 	Many people have commented on the fact that OSF will be dominated
> 	by IBM and DEC.  This shows a clear lack of understanding of
> 	how OSF will be structured.
Never mind the structure, who pays the bills?  Somehow I don't see IBM
letting Apollo tell them what to do.  The giveaway as far as I am concerned
is that
> 	Yes, it has been stated that AIX will be the base software for the
> 	OSF version of Unix.
I've used AIX on an RT.  Rhymes with "aches", and my word, that's appropriate.

>	So there will be an
> 	advantage to being a member of the OSF, but anyone can join the OSF,
> 	unlike the ABI club, where AT&T has exclusive control over the
> 	membership.

Serious question: just _how_ is it guaranteed that anyone can join the OSF?