This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: Attribute questions
- To: Jason Merrill <jason_merrill at redhat dot com>
- Subject: Re: Attribute questions
- From: "Joseph S. Myers" <jsm28 at cam dot ac dot uk>
- Date: Mon, 2 Jul 2001 18:01:32 +0100 (BST)
- cc: Mark Mitchell <mark at codesourcery dot com>, "gcc at gcc dot gnu dot org" <gcc at gcc dot gnu dot org>
On 2 Jul 2001, Jason Merrill wrote:
> Apparently so. I can't remember what I was thinking; it seems clear now
> that prefix attributes should be treated syntactially like
> declspecs/cv-quals, and bind to the left.
>
> > I must suspect that my documentation rather than the C++ front end is
> > correct here, and implement what I've documented for C.
>
> Please.
OK, if reverting your patch (revision 1.240 of cp/decl.c) leaves the
compiler passing its testsuite, I suggest you start by doing that, since I
won't have the full implementation for both C and C++ (making attributes
such as "format" work inside declarators, etc.) done immediately.
In some places there is a kludge that if a function attribute is applied
to a function pointer, then it applies to the pointer target type instead.
I'll probably make this more systematic (for attributes which only make
sense on functions, not data). If MS-compatibility (or compatibility with
legacy GCC-compiled code which doesn't expect a tight formal spec for
where attributes go and just puts them anywhere that happened to work)
turns out to need any more such kludges for cases such as here - I
couldn't find documentation for what MS think the formal syntax of __cdecl
should be - then we can consider adding them later.
--
Joseph S. Myers
jsm28@cam.ac.uk