On 05/05/11 10:24, Eric Blake wrote:
>> +In C99, @code{assert} can be applied to any scalar expression.
>> > +In C89, the argument to @code{assert} is of type @code{int}.
> Is it worth trying to fix that in gnulib, as an independent fix? Are
> there any implementations where assert(0x100000000LL) fails instead of
> succeeds, because the conversion from long long back to int lost the
> non-zero bit?
A more interesting case is assert (P) where P is a pointer.
I've written code like that and never had a problem.
(But then, how would I know? :-)
Another case is assert (0.5). In C89 this must
report an assertion failure, but on all practical
implementations I know of, it succeeds. So to some extent
that C89 requirement never reflected reality, and the C99
change adjusted the standard to match reality better.
I wouldn't object to fixing 'assert' so that we're sure that
it has C99 semantics, but it's pretty low priority as the
problem, I expect, is only theoretical. Also, making the
fixed 'assert' report failures in the same way as the original
might prove to be a challenge.
