Index: [thread] [date] [subject] [author]
  From: G. Gregory <ggregory@bitbox.co.uk>
  To  : ggi-develop@eskimo.com
  Date: Fri, 19 Feb 1999 15:29:34 +0000 (GMT Standard Time)

Re: libGGI3d -- ready to code?

On Fri, 19 Feb 1999, Tristan Wibberley wrote:

> 
> 
> On Fri, 19 Feb 1999, Aaron Van Couwenberghe wrote:
> 
> > CORBA for such low-level interfaces destroys all possibility of efficiency
> > and reasonable size.
> > 
> > CORBA for their higher-level wrappers (Mesa-GGI et al, possibly even a
> > directx clone) is feasible in many cases. However, usually these APIs are
> > not well suited to being networked and need rethinking anyway during
> > CORBAfication.
> > 
> > Believe me, taking an API and making a straight dupe of it on an IDL
> > interface isn't the right way to do things. You sacrifice an incredible
> > amount of speed for flexibility, on each layer that you're using CORBA.
> 
> He didn't say CORBA, he said IDL.
> 
> It would be good to use an IDL compiler to produce a C interface, then
> create other language bindings using the C interface for their
> implementation. This keeps things consistent and has *no* overhead.
> 
> The problem is finding a suitable IDL compiler that doesn't mandate the
> use of CORBA. It might also be a problem coming up with an interface
> that's suitable for IDL.

Yup, Ill do a trawl over the weekend and see if I can find some useful
stuff. Ill see if I can get time to look at current LibGGI3d interfaces as
well.

I wouldnt suggest that any level of GGI be implemented using CORBA, leave
that to higher levels such as Berlin.

Graeme

PS Due to my work changing the rules Ill be posting from second address on
my sig from now on.

-- 
Graeme Gregory                          | "god is a number you cannot
ggregory@bitbox.co.uk                   | count to, you are posthuman
dp@khazad-dum.freeserve.co.uk           | and hardwired" - M. Manson

Index: [thread] [date] [subject] [author]