[cfe-commits] MS pure virtual call ABI

[Chandler Carruth]

I agree -- my response was to his first paragraph, not to the MSDN article.
The added information since my post also helps clarify much. =]


On Wed, Jun 27, 2012 at 6:50 PM, John McCall <rjmccall at apple.com> wrote:

> On Jun 26, 2012, at 11:28 PM, Chandler Carruth wrote:
> > I'm sorry, but for now I would ask you to not submit patches for
> unpublished or undocumented ABI features.
> >
> > There are some very limited places where we can take patches for such
> features, but it's not yet clear to me that this is one. I'll talk to some
> others about this particular instance.
> >
> > We're still working out some aspects of the Windows support strategy for
> Clang, but it would help a lot to focus efforts on fixing correctness
> issues with the published (on MSDN) ABI specifications.
>
> Chandler, this seems pretty well documented to me — the MSDN page that
> João linked (http://msdn.microsoft.com/en-us/library/ff798096.aspx)
> clearly specifies that we have to ensure that _purecall is called when a
> pure virtual call is made, and Timur demonstrated that the observable
> effect is indistinguishable from that call happening immediately.
>
> Given that requirement, we could implement it in any of several different
> ways, but putting &_purecall in the appropriate vtable slot is the
> obviously best option.
>
> John.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/cfe-commits/attachments/20120627/63705bb3/attachment.html>