[llvm-commits] [llvm-gcc-4.2] r110043 - /llvm-gcc-4.2/trunk/gcc/objc/objc-act.c

[Stuart Hastings]

On Aug 2, 2010, at 3:44 PM, Dale Johannesen wrote:

>>>> 
>>>> Hrm.  Well, I guess I could make the test break itself when compiled with clang.  You're suggesting
>>>> 
>>>> #if __clang__
>>>>        BOGUS XLERB HERE;
>>>> #endif
>>>> 
>>>> ?
>>>> 
>>>> However, it wont' be the same diagnostic as llvm-gcc, so I can't have the test check for the (llvm-gcc only) diagnostic I want.
> 
> #if __clang__
> #error text close enough to gcc's error message to fool grep
> #endif
> 
> It's a real hack, though.  IMO improving the mechanism is a better way forward.  Different compilers won't report the same error the same way, and the mechanism should understand that.

Wow.  That's ...  "ingenious."  (Not the first word I thought of.)  Regardless of my initial reaction, that's pretty diabolical, Dale.  I'm impressed!  :-)  In fact, I'm going to do just that (thank you for the suggestion!).  The diagnostic is already wired into the test, inside a "CHECK:" line, so I'm not making it any less portable.

FWIW, Daniel says he regards all the LLVM Frontend* tests as LLVM-GCC-specific, and is quite unconcerned about making Clang pass them.

stuart