may_alias attribute and type identity (PR c++/34935)

[Mark Mitchell]

Richard Guenther wrote:

> I don't think that may_alias is an attribute that needs to be mangled, as its
> semantics affect code generation but not interfacing.  For example:
> 
> void foo(int *x __attribute__((may_alias));
> 
> and
> 
> void foo(int *x);
> 
> are not distinguishable at the point of the caller -- but only affect code
> generation in foo() itself.

But only one direction of conversion makes sense implicitly: the usual 
contra-variance applies.  And, S<int *> and S<int *x 
__attribute__((may_alias))> are different types, and therefore must be 
mangled differently; you certainly can't assume silent conversion 
between them.

-- 
Mark Mitchell
CodeSourcery
mark@codesourcery.com
(650) 331-3385 x713