Hello John!

Finally moved this to SemaType and added more tests for pointers to
function appearing within other declarations.
Could you, please, look at this?

On 10 April 2013 22:40, John McCall <[email protected]> wrote:

> On Apr 8, 2013, at 3:06 PM, Alexander Zinenko <[email protected]> wrote:
> > Sure, this code should belong to SemaDeclAttr. It required several not
> very elegant changes to make it work correctly in some places.
> > I also removed 'if'-ed dyn_cast to FieldDecl used just to get type since
> it is inherited from ValueDecl and should be covered in the previous 'if'.
>
> Mmm.  We generally handle this as a type attribute, not a decl attribute,
> so I think you probably need to handle it in SemaType.  This will let you
> handle things like variables declared as pointers to __thiscall functions.
>  It should just be a matter of looking at the immediate declarator context
> when applying the attribute (and inelegantly repeating some small amount of
> analysis about CurContext and 'static' and so forth).
>
> John.
>

Attachment: callconv.patch
Description: Binary data

_______________________________________________
cfe-commits mailing list
[email protected]
http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits

Reply via email to