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

Re: New Client Reply - [Support !KOM-880811]: chunksize, NC_STRING questions

Hi Charlie,

> It looks like ncdump/ncgen use simple double quotes for NC_STRING.
> How does one disambiguate between NC_CHAR and NC_STRING?
> For variables it is not a problem because the type of the
> variable is specified in the metadata before the data section.
> This is not true for attributes.
> The type has to be indicated by some sort of sentinel, e.g.,
> 1.0f, 1ll, 1us.
> Aren't double quotes, alone, ambiguous, in that they could
> represent both NC_CHAR and NC_STRING?
> 
> We have (temporarily, perhaps) addressed this ambiguity
> in ncap2 by adding an "s" after the quotes to indicate NC_STRING, e.g.,
> 
> global@Conventions_NC_CHAR = "CF-1.0"; // NC_CHAR
> global@Conventions_NC_STRING = "CF-1.0"s; // NC_STRING
> var_foo=1.0;
> var_foo@units="meters"s; // NC_STRING
> var_foo@units2="meters"; // NC_CHAR

This is a good question.  I think we need to distinguish char and string
attributes to make sure a classic model netCDF-4 file doesn't become a
netCDF-4 file inadvertently on a round trip from netCDF to CDL to netCDF
via ncdump and ncgen.

I'm at a meeting in D.C. all day today and tomorrow, so I may not get to
this until the end of the week, but in the limited time I've had to
think about it, I think you're right, that we have to fix this
ambiguity while preserving as much backward compatibility as practical.
Thanks for bringing this to our attention.

--Russ