[cxx-abi-dev] Deleted virtual functions

[Lawrence Crowl]

On 6/1/09, Dennis Handly <dhandly at cup.hp.com> wrote:
> From: Lawrence Crowl <crowl at google.com>
> > I have been keeping an eye on the issues, and trying to avoid
> > making matters worse.  However, there are no good solutions
> > without language support.  In particular, the language needs
> > to recognize a layer of scoping between the translation unit
> > and the program.
>
> Yes, that was part of it.
>
> > I think we did our users a disservice by abandoning the work.
>
> Don't blame me.  ;-)  I was saying we could do most of it but we
> would have to work harder in the runtime.

I'm not blaming anyone, just expressing frustration.

> > > Future directions in the standard should be thought out for
> > > the ABI.
> >
> > I don't follow this comment.  Can you clarify your intent?
>
> The issues that Sean and Mark brought up, whether the type_infos (and
> dynamic_casts) are the same or not.  Based on their visibility to a
> dynamic loader.
>
> I don't know if the questions we raised were in your URL?
> http://www.open-std.org/JTC1/SC22/WG21/docs/papers/2007/n2425.html

In reviewing the paper, not explicitly.  The intent was that
type_infos for protected or private types be distinct between
libraries.  Anything else would lead to unreliable combination
of libraries.

We are a bit off topic here, though.

-- 
Lawrence Crowl