[llvm-commits] [llvm] r45221 - /llvm/trunk/lib/Target/PowerPC/PPCTargetAsmInfo.cpp

[Dale Johannesen]

On Dec 19, 2007, at 5:16 PM, Chris Lattner wrote:
> On Dec 19, 2007, at 5:14 PM, Dale Johannesen wrote:
>>>> However, there's more.  To take a concrete example, I'll discuss g+
>>>> +.dg/eh/ctor1.C.
>>>
>>> Ah, ok!  To be concrete, I meant the llvm nightly tester program
>>> suite, not the dejagnu suite.  Is there any way to make *it* work on
>>> ppc?
>>
>> Not sure what you mean.  The unwinding library is not an issue in  
>> that
>> case.
>
> Ok.  My concern is that I want the nightly tester testing the most
> useful thing possible, ideally what users will get when they use
> "llvm-gcc foo.cpp; ./a.out".   There are a couple of EH-using tests
> in the llvm-test suite, I'd like for them to "start passing" :)

Some will.:)

>> The EH will work (modulo bugs) if you pass -enable-eh; however,
>> the top-level Makefile currently sets -enable-correct-eh-support.
>
> What is the difference these days?

I believe correct-eh uses sjlj, but haven't really looked at it.  I'm  
pretty sure it's not
something we want to worry about going forward.

> Should the top-level makefile be changed?


You're asking me?  IMO it's appropriate to change it for targets where  
-enable-eh
has been implemented, and not for others.