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

Re: Compound data types and Vlens

Roy Mendelssohn <address@hidden> writes:

> Hi Guys:
>
> Do you have an example of a compound datatype  including one or more
> Vlen's?  seeing the code and the file would be a big help.

Look in nc_test4/cdm_sea_soundings.c for an example.

>
> At first I couldn't see how you were going to do the offsets for the
> Vlens, but can I assume all that is actually stored in the structure
> is the nc_vlen_t structure so that it is a fixed size?

This is my understanding as well.

> Years ago I wrote a fortran 77 interface of my own to necdf2 and I
> figured out how to do some of the stuff.  It is slowly coming back to
> me how I did it, and usually simpler is better.  So to do Vlen's I
> just really need to pas you two integers, the size and the value of
> the location of the start of the data, which can be gotten by the
> "loc" function in fortran.  I haven't tried writing yet but reading
> is working,
>

I am also struggling with how to use these new features in fortran!
Sounds like you are well ahead of me on this issue.

At the moment I am working feverishly on a paper that I am presenting
at the AMS in three weeks. (And swearing, over and over again, that I
will never again submit an abstract for a paper that is not written
yet!)

When I return from the AMS I will be back on the fortran problem
again.

After rereading all of your very informative emails on the topic I am
starting to wonder if we should use some features of Fortran 2003, if
implemented in at least one compiler, and then count on all the other
fortran compilers to catch up with fortran 2003 eventually.

They'll have to, one day! That's the way the parade is headed and they
don't have much choice but to march along.

Thanks for all the help!

Ed
--
Ed Hartnett  -- address@hidden