Re: help with netcdf file, please

Blanca,

> I followed your instruction but I don't have positive results.  The error mes
> sage is the same.
>
> If it still doesn't work, see if you can make the modified CDL file
> available to me and I'll try to fix whatever bug in ncgen is causing the
> problem.
>
> Hummmm, the file is big (104.3 MB),  maybe I can find some WEB place to put i
> t for you.
>
> I really need help because my timeline is very short to obtain any goal for m
> y thesis work.  Thanks a lot.
>
>
>  Blanca Elvira Oviedo T.
> Universidad Nacional de Colombia

If I gave you a short C or Fortran program that read a copy of the
original netCDF file, added 4 to each value of the temp variable, and
wrote the resulting netCDF file back out (in the same file as was opened
for reading), would you be able to compile and run it to solve your
problem?  That might be faster than trying to determine why ncgen
doesn't seem to work on your DataModified.cdl file.

I could send such a C, Fortran-77, or Fortran-90 program today if I knew
you could use it, but it would only be useful if you can

  1. compile the program with a C, F77, or F90 compiler, which means
     using the right flags to tell the compiler where to look for the
     netCDF "include" header file that defines the functions and
     constants needed in a netCDF program
  2. link the program (using the compiler in step 1) with an installed
     netCDF library to produce an executable program
  3. copy the original netCDF file to a new file
  4. run the program resulting from step 2 on the file copy produced in
     step 3

If there's some reason you can't do these steps (for example, the netCDF
library is not installed on your computer), then it makes sense to see
what the problem is with ncgen.

Perhaps you could mail me a copy of the DataModified.cdl file that has
only one of the 24 times, although I realize it might be hard to edit
the 13 variable values to include only the first time.  Nevertheless, it
would provide a smaller file to try to reproduce the ncgen problem ...

--Russ