Mikael,

I think this would make a very good topic - there is a lot of
material buried on this newserver, and it would be great for
someone to have the opportunity to form a solid proposal.
Whether or not your proposal is adopted, it would make a very
good point of discussion and reference, especially if you're
studious about your reasoning, assumptions and the pros and
cons.

The only thing I would suggest is thinking carefully about
the backwards compatibility criteria.  I think the user
community could tolerate something resembling the current
syntax - assuming that there are reasonable benefits.
On the one hand, backwards compatibility helps narrow the
possible solutions, but on the other - limit the benefits.

One other thing - consider the "serialisability" in terms
of 3rd party applications reading, writing and manipulating
these data structures.  This is one weakness in the current
setup that could bear some serious thought.

And finally - extensibility.  The POV team are opposed to 
a binary plug-in architecture on the basis of portability.
The ability to create a module without having to edit any
code in the core of POV would be a very nice feature.
Can we think of POV as a platform proving services to 
independent modules, rather than relying on a central parser?
What are the implications for the scene description syntax?

Plenty of juicy ideas to tackle... Have fun!

Nigel

Disclaimer - I speak for myself, not the POV team.
 
--
Nigel Stewart (nig### [at] nigelscom)
Research Student, Software Developer, Tokyo Dweller
Ignorance + Internet = 1984

Post a reply to this message