Re: netCDF OarS conventions

Tim Holt (holtt@oce.orst.edu)
Mon, 28 Sep 92 10:29:08 PDT

Ken & netCDF Group...

Thank's for the reply.  I guess lately I've been reading too much Guy
Kawasaki and my latest horiscope says...

    "...you're blessed with a disquieting talent for asking questions no
    one else has the nerve to.  You wouldn't be afraid to don a giant
	chicken suit if it helped you prove a point."

I think what has made me a bit off lately is confusing DIMENSIONS in a
data/file definition sence, with POSITIONS in space and time -- ie.,
the theoretical v.s. the practical.  What it comes down to for the
average tech/PI with raw data is this -- "I want to make a graph of time vs
temperature", "I want to plot the tracklines from the cruise", or "Where
were we when we took water sample 38, and what was the flow-through temp
and conductivity?"

One thing I can see a good data file format must manage is this:  every 
platform will have some position at a given time.  This position may be 
fixed or variable, and probably defined by some high-precision device 
like GPS.  This position will be global for the platform.  Global in the
sence you use it in programming.  Many instruments will inherit this 
global position, or will derive a position by some offset from the 
global position.  And to make it more complex, this offset may be 
variable, or it may be fixed. 

So, how do you manage global and local coordinates for data samples in a
clean, simple form?  To take a word from OOP, INHERITANCE is a good
possibility.

The big catch though, is making something that works.

-Tim