RE: DOQ

["Crispen, Bob" <Robert.Crispen@HSV.Boeing.com>]

> Volker Coors[SMTP:coors@igd.fhg.de] sez:
> 
> Bob, I also think its a good idea to present the work in a paper to
> publish the results!  
> Of course, some work has to be done before. Let's start with that
> DoubleCoordinate. Looks quite important.
> 
I think it's a reasonable thing to do as well, but
after the paper is published I think we need to
do some selling, politicking or what have you.

I think this is very doable, btw.  There will be
some annoyance in building the demo, having
to do with squeezing data into an unnatural
form and getting it back out again, but I don't
think this is a show stopper, and I'll certainly
be glad to pitch in.

I also agree with Lee Iverson that there are
at minimum some serious difficulties with
representing entities in single-precision float.
Whether VRML is technically adequate to
address a significant number of problems of
the geo community is, at this point unknown.
For instance, a pair of 32-bit floating point
numbers for each coordinate (with one being
the bias or displacement and the other being
the value) may turn out to be a reasonable way
to proceed.

Mitigating our difficulties is the range of
importance that (in my perception) the VRML
community gives this work.  That is, I think
the *low* end of the range most VRML folks
would give geo applications is "important".
No one I know of feels it's unimportant.

And as usual Don Brutzman has prepared a
brilliant list of things we can begin to solve.

Bob Crispen
bob.crispen@boeing.com