>On Fri, 8 Mar 1996, Peter Galbavy wrote:
>> I would like to see *BSD move towards trashing the __P() macros,
>> declaring functions ANSI style and generally making sure that the code
>> (at least under GNU C) compiles without warning or "areas of doubt".

>For what little it is worth, I wholeheartedly aggree.

>Peter
[Peter Svensson, that was.]

Shall we have a conspiracy of people named Peter who are sick to death of
pre-ANSI compatability?

Would anyone like to offer an example of a machine that:
1.  Can physically run NetBSD in principle.
2.  Does not have an ANSI compiler.
3.  Cannot build gcc.

The last is the killer; there is no way we will *ever* be able to support a
machine that doesn't support gcc, without a major rewrite of, at a bare
minimum, the C compiler.  Given that, we can *assume* an ANSI-like compiler
underlying the bootstrap effort; without gcc, we're long-dead anyway.

Can anyone offer an example of such a platform?  If not, can anyone offer
an example of how pre-ANSI compatability is useful, *given the existance of
gcc*?  A platform that our code should run on that gcc doesn't, for instance?

-s