Thanks!

On May 2, 2011, at 12:00 PM, Chandler Carruth wrote:

> On Mon, May 2, 2011 at 11:33 AM, Argyrios Kyrtzidis <[email protected]> 
> wrote:
> 
> On Apr 29, 2011, at 7:48 PM, Richard T wrote:
> 
> > Moved the logic for checking undefined behavior and its warnings into 
> > CheckCompatibleReinterpretCast().  Also, cleaned up the test case a little.
> 
> Thanks! Submitted in r130703 with a tweak; I changed
> 
> +  if (Diags.getDiagnosticLevel(DiagID, SourceLocation()) ==
> +          Diagnostic::Ignored) {
> 
> to
> 
> +  if (Diags.getDiagnosticLevel(DiagID, Range.getBegin()) ==
> +          Diagnostic::Ignored) {
> 
> FYI, it's important to call Diags.getDiagnosticLevel with the location where 
> you intend to emit the warning; in the presence of diagnostic pragmas it is 
> possible that a small section in the source has the warning enabled while it 
> is normally disabled.
> 
> Chandler, if you can evaluate the new warning that'd be great.
> 
> Only two warnings in LLVM+Clang. Fixed one in r130707 as it looked clearly 
> wrong (although perhaps benign).
> 
> The other is in lib/Object, I'll chat about that on IRC to see if that code 
> should be changed as well. 
> 

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

Reply via email to