[cfe-commits] r86962 - /cfe/trunk/tools/clang-cc/clang-cc.cpp

[Douglas Gregor]

On Nov 12, 2009, at 7:39 AM, Daniel Dunbar wrote:

> On Thu, Nov 12, 2009 at 5:39 AM, Sebastian Redl
> <sebastian.redl at getdesigned.at> wrote:
>> Daniel Dunbar wrote:
>>>
>>> Author: ddunbar
>>> Date: Thu Nov 12 00:48:24 2009
>>> New Revision: 86962
>>>
>>> URL: http://llvm.org/viewvc/llvm-project?rev=86962&view=rev
>>> Log:
>>> clang-cc: -fixit is actually option, not an action, although its  
>>> use with
>>> non
>>> -fsyntax-only is probably untested.
>>>
>>
>> Also, misleading and not meaningful. Some warnings fixit the code to
>> something else semantically than the current code, e.g. i & 1 == 0  
>> is fixed
>> to (i & 1) == 0. If I do CodeGen+Fixit, I get the old semantics in  
>> the
>> object code even while changing the source.
>>
>> So perhaps we should turn -fixit into an action instead of relying  
>> on users
>> not to do anything stupid with it.
>
> Good point. Doug, was there a specific reason to allow this outside
> -fsyntax-only?

Nope, I just didn't think about it.

> For now I'll just disable fixits with code generation, from the
> clang-cc view its still really an option not a action. Clients (and
> the driver) can always enforce a stronger policy.

That makes sense to me! Thanks Sebastian and Daniel.

	- Doug