[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: 971002: netcdf 3 and GMT

This archive contains answers to questions sent to Unidata support through mid-2025. Note that the archive is no longer being updated. We provide the archive for reference; many of the answers presented here remain technically correct, even if somewhat outdated. For the most up-to-date information on the use of NSF Unidata software and data services, please consult the Software Documentation first.

  • Subject: Re: 971002: netcdf 3 and GMT
  • Date: Sat, 04 Oct 1997 18:20:25 -0600
>To: Russ Rew <address@hidden>
>From: Paul Wessel <address@hidden>
>Subject: netcdf 3 and GMT
>Organization: University of Hawaii
>Keywords: 199710022207.QAA22322 netCDF 3

Paul,

> >Yes, the netCDF-3 interface tries to prevent reading data as text that
> >was written as numbers, so unless the external type of the "title"
> >attribute is NC_CHAR, you will get this error message.
> 
> Great, that was it.  The old code that made the database had NC_BYTE.
> rewriting the database with NC_CHAR fixed it.
> 
> THe only inconvenience I have now is that there are no #defines for NC_INT
> etc in netcdf.h, only a typedef enumeration.  However, all the examples use 
> NC_INT
> and NC_SHORT etc.  I had to put in #define NC_INT 4 to make things compile.
> I must be missing something .

I don't understand this.  The enum declaration in netcdf.h defines
NC_INT as 4:

    /*
     *  The netcdf external data types
     */
    typedef enum {
            NC_BYTE =   1,      /* signed 1 byte integer */
            NC_CHAR =   2,      /* ISO/ASCII character */
            NC_SHORT =  3,      /* signed 2 byte integer */
            NC_INT =    4,      /* signed 4 byte integer */
            NC_FLOAT =  5,      /* single precision floating point number */
            NC_DOUBLE = 6       /* double precision floating point number */
    } nc_type;

which is a better way to define it than using #define (because it can be
type-checked by the compiler), but you shouldn't need to #define NC_INT
in addition to this.  All the nc_test code uses these types and doesn't
require any #defines.  Maybe the assumption that you need to #define
these is due to a misleading compiler message.  Can you send us a
small example that doesn't seem to work with just the enumerated typedef
above that including netcdf.h defines?

--Russ