[Geowanking] Re: GML and Multiple Instances of Geometry

[stephen white]

On 25/06/2007, at 1:37 PM, Anselm Hook wrote:
> What I want is a shared grammar that captures a sizable chunk of
> reality.  Beyond multiple instancing I would like to also express
> geometry parametrically and I'd like to express linear constraints,

I thought about that problem for a long time, and concluded that it  
was not a grammar problem. You can represent any relations like that  
in a programming language which is embedded into the viewer.

So the grammar you might want is "Python" or "Ruby" or some other  
form of language that is eval()'able at run-time (assuming the viewer  
is in C or something that is actually fast enough to handle the system).

The grammar needs to hand over the parameters concerned to a language  
stub that can then calculate any such relationships between elements  
in the model. The language is Turing-complete, whereas grammar isn't.

I'm posting too much to the list, so I'm going to go quiet again.

--
   steve at adam.com.au