make versionitis? (OpenBSD)

[Zack Weinberg]

On Thu, Nov 09, 2000 at 12:34:04PM +0100, Marc Espie wrote:
> In article < FE465D8F724E3F4F811D067203A214AE11AB00@red-msg-08.redmond.corp.microsoft.com > you write:
> >Before I go off and try to figure out a fix (or workaround), what are the
> >rules of
> >engagement w.r.t. make versions?
> >
> >Specifically, recently the line 
> >   GCC_WARN_FLAGS= $(LOOSE_WARN) $($(@D)-warn)
> >was introduced.
> 
> The standard is crystal-clear: embedded variables like what you're trying
> to do are NOT standard at all.

The only use of this trick is to prune warning flags for non-C front
ends.  I don't think this is particularly valuable, but the non-C
front end maintainers insisted.

I don't have access to POSIX.1b.  SUS2
( http://www.opengroup.org/onlinepubs/007908799/xcu/make.html ) says
nothing either way.  I'm not surprised to find it not universally
supported.  However, I think "crystal clear" is a bit excessive.

> This is not even a question of `fixing' OpenBSD make.
> What does OpenBSD make 1.6 means, btw ? There is NO OVERALL VERSION NUMBER
> in make.

People make these things up.  My copy of pmake, which I *think* came
from NetBSD, is labelled "1.45".  This version is said to have been
released around May 1999.  It supports embedded variables.

zw