[cfe-commits] r165361 - in /cfe/trunk/include/clang/Basic: DiagnosticGroups.td DiagnosticSemaKinds.td

[Ted Kremenek]

On Oct 8, 2012, at 3:49 PM, Chandler Carruth <chandlerc at google.com> wrote:

> On Fri, Oct 5, 2012 at 10:25 PM, Ted Kremenek <kremenek at apple.com> wrote:
> Author: kremenek
> Date: Sat Oct  6 00:25:43 2012
> New Revision: 165361
> 
> URL: http://llvm.org/viewvc/llvm-project?rev=165361&view=rev
> Log:
> Place warn_impcast_different_enum_types in a new warning group, -Wenum-conversion, that is a subgroup of -Wconversion.
> 
> Ted, it doesn't look like this patch actually added -Wenum-conversion as a subgroup of -Wconversion, and we had regressions from code that used -Wno-conversion and started to get -Wenum-conversion warnings.

*slaps head*

THANK YOU for noticing this and fixing it.

> 
> I've fixed this in r165440, could you take a look and make sure that looks right?

Looks right to me.

> Also, a test case for -Wno-conversion turning it off might not go amiss…

We definitely should have a test case for this, as that matches what people want.

> I wonder if we could generated test cases for all of the set logic we currently have and then just have nice simple diffs when we change it.... Maybe not worth it.

It's definitely a nice idea.  Even if we cannot generate test cases around code that triggers a warning, we should be able to use a tool like diagtool to regression test whether or not a -W flag (or a combination of -W flags) can turn a warning on/off.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/cfe-commits/attachments/20121008/9aa10f18/attachment.html>