On 17/09/2007, Nicholas Clark <nick ccl4.org> wrote:
> If we've got it right, then all the functions in
mathoms.c exist only for
> binary compatibility. Given that everything has to be
recompiled for 5.10,
> is it right to make the default for 5.10 NO_MATHOMS?
>
> I can see a danger that we'd break any source that
references Perl API
> functions by their full name, rather than using the
short name (which is
> now a macro).
The main problem with that is that nobody smokes with
-DNO_MATHOMS, so
maybe that default should have been turned on a bit earlier,
to catch
potential problems.
|