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

Re: 960612: ncgen

>Organization: PNNL
>Keywords: 199606120126.AA17210

Hi Connor,

> I'm working at PNNL on the Arm Program and have been trying to
> recunstruct a cdf file with ncgen which was converted to text using
> ncdump.  It appears these two programs are not compatable.  I continue to
> get a line too long message when going from txt to cdf.  I've even tried
> to pipe the output from ncdump directly to ncgen.  Any suggestions on how
> to fix this problem.

It's possible there are outputs from ncdump that would cause problems for
ncgen, but I didn't think line length was a source of any such problems,
because ncgen doesn't read in a line at a time (it only reads in a token at
a time with a lex-constructed tokenizer).

This may be a bug that no one has encountered before, but to diagnose it
and provide a fix, I'll need an example of an ncdump output for which ncgen
produces a line too long message.  I can find no reference to any such
message in the ncgen sources, although there is a message about a "string
too long" that occurs if a string longer than 5000 characters is
encountered.  The limit on string sizes that ncgen can handle is an
arbitrary limit set in ncgen.h that can be changed by raising it and
recompiling ncgen.

It's also possible that the "line too long" problem is caused by a bug in
the lex or yacc program used to generate ncgen on your platform, but to
determine that I'd still need to have an example that demonstrated the
problem and to know on which machine and operating system your ncgen was
built.

So, can you make available an ncdump output that demonstrates the problem,
either via e-mail if it's short, or via FTP if it's large?  Thanks!

--Russ

______________________________________________________________________________

Russ Rew                                           UCAR Unidata Program
address@hidden                              http://www.unidata.ucar.edu