[C++-sig] Patch implementing void * support

[David Abrahams]

"Niall Douglas" <s_sourceforge at nedprod.com> writes:

> On 11 Feb 2006 at 16:51, David Abrahams wrote:
>
>> > The attached applies cleanly to Boost CVS 06-02-09-0100 and compiles 
>> > and passes the test suite on MSVC7.1 and GCC 4.0.2.
>> 
>> Well, I finally got it cleaned up and working with older compilers,
>> but now the 1.34 feature freeze is in effect.  I am waiting for the
>> release manager's permission to check in on the trunk.
>
> Thank you.
>
> Are you happy with the way I used a trampoline type to pass a void * 
> through BPL? 

Not really.  If you look at my revised patch you'll see that it's gone.

> It seemed a bit nasty to me. If you ever do a v3 of BPL, you may
> want to pass everything around by pointer though I suppose this
> could inhibit optimisations.

Not sure what you're getting at, here.  My only thought was that when
we do langbinding, we should probably treat pointer types separately
from their pointees so that opaque pointers are less of a hack.  It
means a few more type converter registration records, but that seems
like a small price to pay.

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