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

20010212: variable attributes problems (cont.)

Edwin,

> From: "Edwin R Wolfe Jr." <address@hidden>
> Subject: Re: 20010209: variable attributes problems 
> Organization: University of Michigan
> Keywords: 200102091837.f19IbPL07744 netCDF variable attributes

The above message contained the following:

> Thanks -- this helps.  I've found, by experiment that
> 1.2f
> 1s
> 1b
> 
> also work (note no <.> in the last 2). I can't seem to find this in the 
> documentation -- is there a document I'm missing?

The specification for CDL data constants may be found in secion 10.3 of
the "NetCDF-3 User's Guide for Fortran 77", which is available at

    <http://www.unidata.ucar.edu/packages/netcdf/docs.html>

> A couple more questions, if I may:
> 
> 1)
> I find that using 1s begets
> 
>       iret = nf_put_att_int(ncid, ms_time_id, 'missing_value', NF_INT2,
>      11, int2val)
> 
> which seems an inconsistency in that I'd expect the put routine to be
> put_att_int2 ??

The routine "nf_put_att_int" is used to write values that are stored as
INTEGER-s in memory.  The "1" value in the above is such an integer.

The routine "nf_put_att_int2" would be used to write values that are
stored as INTEGER*2-s in memory.

You need to understand the distinction between a external netCDF type
(e.g. NF_INT2) and the internal, in-memory representational type (e.g.
INTEGER).  In general, they needn't be the same.

> 2)
> using 1s in the data section of the .cdl doesn't seem to similarly change
> the code in the writerec subroutine of the fortran code generated. How can
> I accomplish this ?

I'm afraid that I don't understand the question.  Please explain.

> 3)
> when defining character strings, such as
> char ut_date(nvec, date_len);
> data:
> ut_date = "1999000" ;
> 
> I get
> 
>       iret = nf_put_vara_text(ncid, ut_date_id, ut_date_start, ut_date_c
     1ount, '1999000')
> 
> instead of using the variable name like it does for anything other than
> chars?  Is there a way to get ncgen to use the variable name instead of
> the actual text?

I'm afraid that I don't understand the question.  Please explain.

> 4)
> I find, in order to get any put_var code in the generated writerec
> subroutine, I need to specify data for each variable.  Is there a way to
> specify, in the .cdl, the equivalent of
> 
> lat / 75 * -1.f / other than to actually enter -1 75 times?

Sorry.  There are no such syntactic short-cuts in CDL.

Regards,
Steve Emmerson   <http://www.unidata.ucar.edu>