Re: AC_REQUIRE problems

[Dan Manthey]

On Sat, 12 Feb 2005, Stepan Kasal wrote:
>
> OK, back to reality: let's just suppose functions for this work.
>

So, do functions work portably?

It seems that it should be possible for each function foo to define a
shell variable $fn_foo, and that these could get set prior to the body of
configure.  If some early tests show that functions work, we set each
$fn_foo to foo and define the function foo.  If functions don't work,
$fn_foo could be some hairy thing that behaves like a function.  I may be
a bigot in this matter, but I think it's okay for that hairy work around
to be monumentally slow; that's just a cost to the user whose shell
doesn't have functions.

Of course, if we decide that functions "are portable", we can skip the
step of designing the hairy workaround.

Anyway, I'd be willing to start looking into actually implementing the
idea of AC macros expanding to function calls (or hairy work-arounds
thereof).  Any further comments on the matter?

(I'd be doing this on my own time, not my work's so, it could take a
while, though.)

-Dan