🚀 go-pugleaf

In article <1992Aug19.155052.21659@dsd.es.com> cannon@ziggy.es.com (Gary Cannon) writes:
| 
| Since I haven't seen the spec it is hard to know from this description
| exactly how this all works but from your description it sounds a lot like 
| the PHIGS SOFA's that are supported in PEX or the PHIGS Triangle Set that is 
| likely to be supported in PEX 6.0. Could you explain the difference?

You might want to check out the OpenGL manual pages.  They're available
as compressed Shell archives of PostScript files, by anonymous ftp from
sgi.com, in the files sgi/OpenGL/{gl,glu,glx}.shar.Z .

I used the tetrahedral mesh as an example because it's common in FEA
applications and isn't supported as a primitive by either PEX or
OpenGL.  (One reason it isn't supported directly is that it doesn't
define a "surface" in a way that's conventional for most graphics
packages.)

But the argument isn't specific to tetrahedral meshes.  As long as an
application's problem domain calls for some data structure that doesn't
correspond *precisely* to one in the PHIGS or PEX-based API, then some
reformatting or data duplication is necessary.  You could choose to
reformat tetrahedral meshes into SOFAS, Triangle Sets, Fill Areas,
Triangle Strips, or perhaps other primitives, but you would always be
paying a reformatting cost.

In contrast, the OpenGL API works with arbitrary application data
structures.  Since the relevant graphics data can always be extracted
without having to interpret the remainder of the data structure, no
reformatting or duplication is needed.  This also makes life easier for
the application developer, since he has greater freedom to tailor his
algorithms and data structures to match his problem domain, rather than
the graphics API.

Allen