https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102775
Sam James <sjames at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |sjames at gcc dot gnu.org
--- Comment #4 from Sam James <sjames at gcc dot gnu.org> ---
(In reply to Harald van Dijk from comment #3)
> Here is another example:
>
> $ cat test.cc
> struct S {
> int32_t i;
> } s;
> int main() {
> return s.i;
> }
> $ g++ -c test.cc
> test.cc:2:3: error: ‘int32_t’ does not name a type
> 2 | int32_t i;
> | ^~~~~~~
> test.cc:1:1: note: ‘int32_t’ is defined in header ‘<cstdint>’; this is
> probably fixable by adding ‘#include <cstdint>’
> +++ |+#include <cstdint>
I don't see how this is an example because it's demonstrating a fixit being
emitted. The bug is requesting that in more cases.
> 1 | struct S {
> test.cc: In function ‘int main()’:
> test.cc:5:12: error: ‘struct S’ has no member named ‘i’
> 5 | return s.i;
> | ^
>
> The second diagnostic is useless here, ideally GCC would continue after the
> first error as if S had been defined with a valid definition of the i field.
>
I think you're requesting something different to this bug.
> (Incidentally, the fix-it here also looks wrong. <cstdint> is not required
> to make types available in the global namespace. It should either suggest
> <cstdint> with std::int32_t, or <stdint.h>.)
There's another bug for that I can't find right now.
