Re: Geomview is available --- 3D object viewer

[slevy]

> From rthomson at dsd.es.com Tue Jun 23 18:52:17 1992
> To: slevy at geom
> Subject: Re: Geomview is available --- 3D object viewer 
> Reply-To: rthomson at dsd.es.com
> Date: Tue, 23 Jun 92 17:51:24 CDT
> From: Rich Thomson <rthomson at dsd.es.com>
> 
> 
> >   We haven't considered it very seriously.
> 
> I'd be interested in hearing the reasons why you're not considering
> it.  Simply compare the number of 8-bit pseudocolor X displays versus the
> number of SGI machines and it seems a very convincing argument to
> increase the user base dramatically.

 
> >   We're definitely doing a NeXT port and possibly one for Sun XGL.
> 
> Hmm.  I'm curious to know why you're going for the NeXT box over PEX.

The NeXT is a nicely integrated system with a great user interface toolkit!
It also has 3-D support built in, using the Renderman library.  This last
is under NeXT 3.0, not out yet, but it should be reasonably soon.
(We're developing using a beta version.)  NeXTs seem to be in moderately wide
use in the mathematical community, which is our main target, though perhaps
they're still not as widely used as Suns.

> If you're going to go for Sun's Xgl, then you might as well do a PEX
> port -- the concepts and terminology is almost virtually the same (not
> surprising since Sun's Xgl is lifted from their knowledge of
> SunPHIGS).

We started on the XGL port a while back, before X11R5 came out.  Most of the
low-level code for XGL is done.  Performance for anything other than wireframe
is disappointing, though, which is part of the reason a generic X/PEX port is
less appealing than it might be.  The NeXT actually does a better job than
a Sparcstation IPC with GX card, when we're using shaded surfaces.

> >   Want to try to convince us?
> 
> I'm game.  What does it take to convince you? :-)
> 
> 						-- Rich

So tell me about PEX.  We don't have any systems with PEX-equipped X servers
at the moment, nor PEX client libraries, though I understand there's something
in X11R5.  If we were to write for a "generic" PEX implementation, as opposed
to something from a particular vendor, what features could we depend on?
Is there a well-defined library interface so that we could expect to
write portable code?

   Stuart