Jerry James <james(a)xemacs.org> writes:
1) The DECLARE_DOESNT_RETURN_GCC_ATTRIBUTE_SYNTAX_SUCKS macro was
created and so named because, at the time, you couldn't have
more than 1 __attribute__ declaration per function. That was
changed in gcc 2.6.0, so there's not much point in complaining
about it anymore.
Agreed.
2) Defining "register" to nothing was done by Steve Baur.
I've never understood the rationale behind that. A compiler is
certainly free to ignore the `register' declarations, which I
understand many of them do. Although I don't use `register' when
writing new code, I don't see a good reason to actively nullify them
where they do exist.
3) It would be nice to have access to the __attribute__
definitions
from files in lib-src, but lisp.h in its entirety should not be
included there.
Again, agreed.