Re: visadForm(?).save

• To: "Campbell, Garrett" <Campbell@xxxxxxxxxxxxxxxxxx>
• Subject: Re: visadForm(?).save
• From: Dave Glowacki <dglo@xxxxxxxxxxxxxxxxxx>
• Date: Wed, 24 Apr 2002 10:17:51 -0500

"Campbell, Garrett" wrote:
> I was recently bitten by a problem with
> visadform().save(file,dataobject)
> 
> I saved some data, Updated visad.jar and then I could not read the
> dataobjects back in because the underlying data object class was
> changed.
> 
> There is an other version: visadform(1).save(file,dataobject).
> This saves something called a bianary object.  The documentation does
> not realy explain what that means.  Is this designed so that I could
> read back files even if the underlying data object class had changed?

Yes, the VisAD binary format should be able to read back any VisAD
Data object built using standard VisAD classes, even if those
classes change.

A few classes are serialized, but they're things like
CoordinateSystems, non-standard Delaunay triangulation classes or
user-extended versions of various VisAD type classes, so they're
things that only programmers hacking on VisAD internals will
encounter.  Even if the system encounters a problem with
de-serializing these objects, it *should* continue trying to
load the file, rather than giving up.

The VisAD binary format is documented at
http://www.ssec.wisc.edu/~dglo/binary_file_format.html

I've added this URL to the VisADForm javadocs.  This should show up
tomorrow on the public pages at http://www.ssec.wisc.edu/~dglo/docs/visad/

• References:
  • visadForm(?).save
    • From: Campbell, Garrett