Re: [netcdfgroup] Strange new ncgen behavior

• To: netcdfgroup@xxxxxxxxxxxxxxxx
• Subject: Re: [netcdfgroup] Strange new ncgen behavior
• From: Dennis Heimbigner <dennis.heimbigner@xxxxxxxxx>
• Date: Sun, 08 Aug 2010 11:49:28 -0600

Some time ago, Ken Knapp (below) pointed out a bug in ncgen.
I have since done a complete rewrite of the way
character datalists are handled in ncgen.
This rewrite is now in the nightly snapshot.
=Dennis Heimbigner
 Unidata



Ken Knapp wrote:

Something that wasn't happening with netcdf3.6 is now happening with 4.1.1

The test.cdl is

netcdf test {
dimensions:
    isochar = 19 ;
    ncenters = UNLIMITED ; // (2 currently)

variables:
    char center(ncenters,isochar) ;

// global attributes:
        :history = "testing" ;

data:

 center =
  "atcf",
  "hurdat_atl" ;

}

using
$ ncgen test.cdl -o test.nc
$ ncdump test.nc
 > The result only shows "atcf" for variable /center/

However, if I change "atcf" to "atcf_____" to equal length as "hurdat_atl", then the resulting test.nc is correct (both atcf____ and hurdat_atl are listed in the variable output. However, the converse is not true: if I shorten "hurdat_atl" to "hurd" (same length as "atcf") then I still only get "atcf" for the center variable.

This seems like a bug or at least a new feature of ncgen.

Your thoughts?
-Ken


------------------------------------------------------------------------

_______________________________________________
netcdfgroup mailing list
netcdfgroup@xxxxxxxxxxxxxxxx

For list information or to unsubscribe, visit: http://www.unidata.ucar.edu/mailing_lists/

• References:
  • [netcdfgroup] Strange new ncgen behavior
    • From: Ken Knapp