Re: "proxy" data objects in visad.data.netcdf.in

• To: visad-list@xxxxxxxxxxxxx
• Subject: Re: "proxy" data objects in visad.data.netcdf.in
• From: "William L. Hibbard" <WHIBBARD@xxxxxxxxxxxxx>
• Date: Sat, 04 Apr 1998 07:19 CST

Steve,
 
> One question, though.  Where should the decision to use FileFlatFields
> be made?  In the Form's?  In the Repository?  By the user?  The
> java.lang.Runtime.getMemoryAdvice() method, which indicates the level of
> memory usage, might be applicable.
 
In the design you and I and others worked out last October, Forms
construct Data objects so Forms decide whether to use FileFlatFields.
Forms also pass CacheStrategy objects to FileFlatField constructors,
which also lets Forms decide how to manage data migration between
memory and disk.
 
We could add methods to Form and / or Repository to pass this decision
up to the application.  Letting applications control these decisions
is nice, but would put a burden on all Forms to support FileFlatFields.
 
Of course, as more Forms are written, applications may have some
control over these decision just by their choice of Forms.  I think
for the time being, we should leave it as is.
 
Note that you are free to write Forms and CacheStrategys that use
the java.lang.Runtime.getMemoryAdvice() method to decide how to
manage data migration.
 
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