Re: [cxx-abi-dev] trivial __dynamic_cast fails?

[David Baraff <deb@xxxxxxxxx>]

Begin forwarded message:

> From: David Abrahams <dave@xxxxxxxxxxxxxxxxxxxx>
> Date: March 1, 2004 4:36:28 AM PST
> To: David Baraff <deb@xxxxxxxxx>
> Cc: Daveed Vandevoorde <daveed@xxxxxxx>, cxx-abi-dev@xxxxxxxxxxxxxxxx, 
> Daniel Wallin <dalwan01@xxxxxxxxxxxxxx>
> Subject: Re: [cxx-abi-dev] trivial __dynamic_cast fails?
>
> David Baraff <deb@xxxxxxxxx> writes:
>
> > >
> > David Abrahams wrote:
> > > I think it's probably more widely neededthan you'd guess, at the
> > > boundary between statically and dynamically-typed systems.  I doubt
> > > anyone else has had the insanity to try to simulate it in conforming
> > > code,
> > I did it in standards-conforming code, so either doing it made me
> > insane, or I am insane for doing it...
> Didn't you say that you were computing byte offsets for Base <-->
> Derived translations?
No, I instantiate a function templated on Base and Derived which does 
the cast.
( I was thinking of caching the offset after the first time I found it. 
 That would be non-portable I guess.)
>  If these classes can be non-POD, that's not
> portable according to the standard.
>
> Incidentally, my system stores those offsets as an optimization for
> all platforms that can handle it (which so far is everything) but if
> you want to be strict about portable conformance you have to walk
> through the cast graph.
I do, so I end up calling the synthesized cast functions several times 
in a single cast.


> --
> Dave Abrahams
> Boost Consulting
> www.boost-consulting.com