Charlie Zender - 2002-01-30

Henry Butowsky wrote:

> Hi Charlie,
>  Attached is the log file produced when i ran "ncap -v -c -S ncap.in in.nc
> foo.nc "
> Also attached is an ncdump of the partial netCDF file produced.

Yes, I was going to work on this tonite, it's the last part of getting
the LHS working. The script runs fine when you omit the -v and -c,
i.e.,

ncap -S ncap.in in.nc foo.nc: works
ncap -c -S ncap.in in.nc foo.nc: fails
ncap -v -c -S ncap.in in.nc: fails
ncap -v -S ncap.in in.nc: fails

The failures seem to occur for slightly different reasons,
I do not understand all of them but yes, some are definitely
because the LHS dimensions are not yet picked up in initial_scan.
Much of this is reflected in the ncap portion of the TODO

> For ncap_initial_scan() to work we need to figure out how to output the
> dimensions in the subscripts ! ( So that we can extract only the dimensions
> needed) . For now users can use the -c option to output all of them.

My initial thought is every time that LHS casting is noticed in the
initial scan, those LHS dimensions should be appended to a separate
list of LHS dimensions. These LHS dimensions will then be defined
in the output file, if they havn't been already, in a separate call right after dmn_dfn(). Does this sound reasonable?

> I'm using texi2dvi as part of textinfo-3.12h-2.   The conversion to dvi
> format isn't working correctly - Shall have a proper look tomorrow.

I think it may require texi2dvi v. 4.0+ which I used to have installed
but disappeared when I "upgraded" to RedHat 7.2.