Re: gep-1

[Murray Cumming <murrayc usa net>]

On Fri, 2002-08-16 at 21:13, Sander Vesik wrote:
> On 16 Aug 2002, Michael Meeks wrote:
> 
> > > Sounds like you need to write your own GEP. (-:
> > 
> > 	Absolutely; it seems having C++ stuff more integrated into the 'core'
> > would be nice - but then again, can they make committments to API and
> > ABI stability going forward - and do they understand the ramifications
> > of that ;-)
> > 
> > 	My feeling would be that you can't on the 1 hand ask to be in the core
> > and on the other not be ABI/API stable.
> > 
> 
> I think this should start with defining what ABI guarantees we will be
> giving for the c++ bindings.

We will give _no_ ABI guarantees until we hit stable. Just like a C
library, a C++ API should not be frozen before people are even using it.

> If you want to give C style guarantees,
> you will have to wire down the comiler and toolchain versions. 

It's quite simple at the moment because there is no C++ compilers seem
to share an ABI standard, unlike C compilers. Therefore, you can use
whatever compiler you like as long as you don't use more than one.
 
> It would IMHO also be nice if the adoption into the core of any language
> bindings was staggered and happened one layer at a time.

That is a separate issue. 

-- 
Murray Cumming
murrayc usa net
www.murrayc.com