Re: Why go outside the java type system.

------- Forwarded Message

CC: visad-list@xxxxxxxxxxxxx
 
Glenn,
 
This is a perennial and interesting question.  The developers guide
says that MathType is not a synonym for class, since many different
classes may be used for different approximations to the same MathType
(e.g., a FieldImpl and a FlatField may have the same MathType).
Further, Data classes may differentiate based on data source (e.g.,
the data form adapters define various new Data sub-classes, and
remote Data also define some new Data sub-classes, and these may
all have the same MathType).
 
Furthermore, the same Data class may correspond to different MathTypes.
For example, the RealTuple class corresponds to the MathTypes
(latitude, longitude, altitude), (ir_radiance, vis_radiance), (time),
and infinitely many others.
 
Probably the main reason for having MathTypes is for the information
they carry, including text names and distinct identities for
different RealTypes, default Units for RealTypes, default
CoordinateSystems for RealTupleTypes, and default Sets for RealTypes
and RealTupleTypes.  The distinct identities for different RealTypes
plays a fundamental role in the VisAD visualization architecture,
providing a handle for saying things like time is mapped to the X
axis and temperature to the Y axis.  If we used class for MathType,
it would be hard to give distinct identities to different RealTypes.
 
It has recently occurred to me to say that MathTypes are a lot like
data base schemas.  In fact, the differences between MathTypes and
data base schemas can be traced to the distinction that data bases
assume they model discrete things whereas VisAD models continuous
things.
 
I hope this clarifies the issue.  As I said in my answer to John's
question, one of the most pressing needs is VisAD extension packages
that hide the complexity of the data model for users in specific
disciplines or with specific tasks.
 
Regards,
Bill
 
- ----------------------------------------------------------
Bill Hibbard, SSEC, 1225 W. Dayton St., Madison, WI  53706
whibbard@xxxxxxxxxxxxx  608-263-4427  fax: 608-263-6738
http://www.ssec.wisc.edu/~billh/vis.html
- ----------------------------------------------------------
Bill Hibbard, SSEC, 1225 W. Dayton St., Madison, WI  53706
whibbard@xxxxxxxxxxxxx  608-263-4427  fax: 608-263-6738
http://www.ssec.wisc.edu/~billh/vis.html

------- End of Forwarded Message