> >though.
>
> + if (((TREE_CODE (t) == ENUMERAL_TYPE && COMPLETE_TYPE_P (t))
> + || TREE_CODE (t) == INTEGER_TYPE
> + || TREE_CODE (t) == BOOLEAN_TYPE
> + || TREE_CODE (t) == REAL_TYPE
> + || TREE_CODE (t) == FIXED_POINT_TYPE)
> + && (TYPE_MAX_VALUE (t) != TYPE_MAX_VALUE (tv)
> + || TYPE_MAX_VALUE (t) != TYPE_MAX_VALUE (tv)))
> + {
> + error ("type variant has different TYPE_MIN_VALUE");
> + debug_tree (tv);
> + return false;
> + }
>
> The second || check of TYPE_MAX_VALUE should probably be something else,
> maybe TYPE_MIN_VALUE ?
>
> Why don't we warn about such useless || where both hands are identical? :)
Thanks for both corrections! Something I crept in at last moment when
restructuring the code
(I want to re-use the variant checking for ODR varaints and also in limited
form for canonical
types - combined with existing code to verify canonical type sanity in C++ FE)
Yep, warning would be cute, though it may be bit hard to interpret when the
tests are produced
by different maco expansions/inlines.
Honza
>
> Thanks,
> >
> >Honza
>
