[cfe-dev] SemaCXX/using-directive.cpp

Douglas Gregor dgregor at apple.com
Tue Apr 7 09:33:51 PDT 2009


On Apr 6, 2009, at 9:29 PM, Chris Lattner wrote:

> This test has been failing for me for a long time, with:
>
> Notes expected but not seen:
>   Line 97: forward declaration
> Notes seen but not expected:
>   Line 89: forward declaration of 'struct OneTag::X'
>
> Should the test just be changed, or is this a real problem?

There's a real problem here. When we hit an ambiguity in type-name  
lookup, we emit an error and then return one of the answers (for  
recovery). The problem, however, is that which answer we return isn't  
stable from one machine to the next.  The solution will be to make  
Action::getTypeName return a TypeResult, so that we can return an  
error after emitting the ambiguity diagnostic.

I've XFAIL'd the test, although that'll cause complains on other  
systems :)

	- Doug



More information about the cfe-dev mailing list