subscript ranges

• To: netcdfgroup@xxxxxxxxxxxxxxxx
• Subject: subscript ranges
• From: B. Langdon <langdon@xxxxxxxxxxxxxx>
• Date: Mon, 4 Oct 93 14:25:46 -0700

Is there some existing practice that I could consider adopting for
specifying the lower and upper subscript limits of a dimension?

netCDF says all subscript ranges are 1:n for Fortran, 0:n-1 for C.
However, several languages allow one to declare, e.g. Dow_Jones(1982:1987).
--------------------
I have been using attributes, e.g. for field Ex declared "real ex(0:50,0:90)":

...
dimensions:
        xg = 51, yg = 91;       // logically, (0:50), (0:90)
...
variables:
        float Ex(yg, xg) ;      // subscript order per C, not fortran.
                Ex:dims = 0, 50, 0, 90 ;  // indicates (0:50,0:90)
...

If I could put an attribute on dimension xg, I could use add_offset, say.
--------------------
Anyway, we are soon to adopt some convention in new software that will be
used by other people.  This seems a good time to ask what convention other
people have cobbled together for this functionality.
----------------------------------------------------------------------
        Bruce Langdon  L-472                   langdon@xxxxxxxx
        Physics Department                     Fax: (510) 423-9969
        Lawrence Livermore National Laboratory langdon@xxxxxxxxxxxx
        Livermore, CA 94550                 (510) 422-5444