Subject: Re: future NetBSD cpp predefines...
To: Ted Lemon <email@example.com>
From: Chris G Demetriou <Chris_G_Demetriou@LAGAVULIN.PDL.CS.CMU.EDU>
Date: 12/22/1994 01:25:42
> It might be worthwhile also to #define __NetBSD__ to some value - say,
> the major version times 100 plus the minor version.
in a word: no.
for several reasons:
(1) something like this already exists. (see <sys/param.h>)
(2) it unnecessarily burdens the compiler configuration
process. There's no reason that, if on day N the
'name' of -current is "1.0A", and we decide that
on day N+1 that it should be "1.0B" that the compiler
configuration goop should care.
(3) it means that your compiler is strictly limited to
at-most-one-release-name between recompiles.
Especially if people have their own locally-compiled
versions of gcc, or whatever, this is _NOT_ appropriate.
> Gcc currently does something similar with __GNUC__ - its value is
> currently 2. __GNUC_MINOR__ is 6. There's no predefine to
> differentiate between 2.6.0 and 2.6.2, though, which I think is a bug.
that's OK for the compiler to do: the compiler version is inherently
bound to the compiler's source and binary code. the os version is