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

Re: change to program gennet.f

> Organization: NOAA/FSL
> Keywords: 199502031746.AA23223

Hi Barry,

> Hi. Long time no communicate with. Anyway, I was "net surfing"
> and ran into my program gennet.f and discovered an error in it.
> It also brings up a question. When I wrote that routine, the fortran
> interface returned the dimensions for variables in C order, i.e., in
> my routine I had to account for this by looping thru the dimensions
> in reverse order. About 3 or 4 months ago, a local user of gennet.f
> told me that the program was writing out variable declarations
> in the reverse order; I concluded that since the original version
> worked correctly, a change to the netcDF software must have
> occurred. So, I fixed my routine here locally but never ftped it
> to you. Below you'll have the code that accounts for what I think
> was a software change. Is it possible to be notified of such future
> software changes, or do I need to occasionally login to the unidata
> access (via mosaic or whatever) and check myself???? Anyway,
> I thought it was time to update gennet.f. 

There has been no such change in the netCDF software.  The order of
dimension IDs returned by NCVINQ has always been the same as the order in
which the dimensions were passed in to NCVDEF, which is the reverse of the
order for the C interface.  Hence the data in the file has the first FORTRAN
dimension varying fastest, but the last C dimension varying fastest, as
users of the two interfaces would expect.  Neither the documentation nor
code for this feature of the Fortran interface has changed since the first
release.

We would never intentionally make a change to the documented interface that
was not backwards-compatible, and certainly not even consider such a change
without announcing our intentions on the netcdfgroup mailing list.  There is
no need to check here for what's changed: if you are on the netcdfgroup
mailing list, you will should receive all announcements of new versions and
additions to the interface.

I have no explanation for why your program requires changes to work with the
current version of netCDF.  There has been no change to the Fortran
interface since about April 1993, and that was only to add new functions to
the interface that weren't there before.

Nevertheless, I've updated the copy in our netcdf/contrib/ directory to the
new version you appended to your message.

______________________________________________________________________________

Russ Rew                                                UCAR Unidata Program
address@hidden                                          P.O. Box 3000
http://www.unidata.ucar.edu/                          Boulder, CO 80307-3000
______________________________________________________________________________