The classic data model does not have a primitive string type with variable-length strings, only arrays of type char.

• Do not use char type for 8-bit numeric data, use byte type instead.
• Use an attribute in preference to a variable to store a single string.
• Use a multidimensional char array for arrays of strings. In netCDF-3, all of these strings must be the same length.
• Various conventions are used for writing variable length string data in netCDF-3:
  1. Fortran: pad with blanks and never terminate with a zero byte.
  2. C: pad with zeros and always terminate with a zero byte.
• To accommodate these variable-length string conventions when reading, trim zeros and blanks from the end of the char array and if in C, add a zero byte terminator if needed.
• A common error in using the netCDF-3 C API is assuming char attribute values are null-terminated.

2008 Unidata NetCDF Workshop for Developers and Data Providers > Best Practices