[LLVMdev] Possible LiveInterval Bug

[Evan Cheng]

On Jan 30, 2008, at 3:53 PM, David Greene wrote:

> On Wednesday 30 January 2008 16:22, Evan Cheng wrote:
>
>>> I was thinking along the same lines.  But does this work for the
>>> cases where
>>> IP is _not_ end()?  Is it true that we always want to look one back
>>> from IP?
>>
>> I think so. That's the intention and it's safe because IP != begin().
>
> Right.  But doesn't that imply we've had some messed up intervals  
> being
> created for quite some time, even onces where IP was _not_ end()?   
> How the
> heck did this ever work?  :-/

This routine is rarely called. I am surprised the bug hasn't surfaced  
before. But that's how it is...

Thanks,

Evan

>
>
>                                   -Dave
> _______________________________________________
> LLVM Developers mailing list
> LLVMdev at cs.uiuc.edu         http://llvm.cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev