Re: rmicdf problem

> From: address@hidden (Tom Glaess)
> To: address@hidden
> Subject: rmicdf problem

Hi Tom,

>       I've been using Joe Sirott's rmicdf implementation and I found a
> problem. I told him about it and he told me you guys are taking over the
> software. With that in mind, I thought I'd tell you what I found.
> 
> The software seems to have a problem with scalar (dimensionless) variables. In
> the file netcdf/JiNcSource.java, the code allocates memory for an array
> of Dimension classes (line 300). However, if the variable has no dimensions,
> this code will not do anything. On line 306, the code tries to access the
> first member of that dimension array. With no memory allocated, this line
> produces an exception and the code halts. I've experienced no other problems
> with the software and I'm looking forward to using a great deal. Our database
> produces netCDF files as output and our users are clammering for more web
> apps to look at the data. A Java interface to netCDF will help us
> immensely.

Thanks for reporting the bug.  We'll put in a fix.

> Oh yeah, what are your plans for the software?

We plan to enhance and support it.  The version we release will have
package and method name and access changes, however, so be prepared for
such changes before you write a lot of code to the current interfaces.

Joe is here visiting today and I've just gone through a walk-through, so
I have a good idea of what we will be doing next:

  - Adding an attribute class to the netcdf package
  - Adding create and write methods to the netcdf package
  - Adapting the rmicdf classes to Sun's Alpha 2 release of RMI (almost
    done)
  - Changing the access of many fields and member functions from public
    to private or "phriendly"
  - Adding more documentation for javadoc
  - ...

I can't say when we'll have a version to release, but I'm hoping it won't
be too long.

--Russ