[cfe-commits] [PATCH] Abstract v-table generation into the CGCXXABI interface

[John McCall]

On Jun 28, 2012, at 4:08 PM, Timur Iskhodzhanov wrote:
> FTR, this patch has regressed:
> http://llvm.org/bugs/show_bug.cgi?id=13231
> 
> Now I can't build non-trivial C++ code anymore.
> [I had local patches to work OK on some files and was planning to send
> a few things for review but now I'm literally stuck at ~r158985]

There are going to be growing pains like this — I do understand your use
case of "I'd just like as much to work as possible, even if it's not using the
right logic for MSVC at all", but as long as we have multiple people working
on the ABI, that's not really supportable.

I think Charles is working on patches to actually implement this, though,
so it might not be too long.  In the meantime, I think you should either
bite the bullet and propose a patch to add a -cc1 flag to opt in to
inconsistent ABI generation (understanding that that means you'll get
miscompiled from time to time), or just locally make the Microsoft
implementation call the Itanium implementation.

John.