Re: 19990715: "Numeric conversion not representable"

Hi John,

>Date: Fri, 16 Jul 1999 11:50:59 -0400
>From: address@hidden (John Sheldon)
>Organization: Geophysical Fluid Dynamics Laboratory, Princeton University
>To: Steve Emmerson <address@hidden>
>Subject: Re: 19990715: "Numeric conversion not representable"
>Keywords: 199907152254.QAA09656
In the above message, you wrote:

> Hi again Steve-
> 
> Hope you all are hanging in there OK with Glenn gone. I know it's
> gotta be hard....

It is.  We all miss him.

> I think you overestimate our awareness when you cite our "reliance"
> on a (hitherto unknown) unspecified behavior :-)   We just noticed that
> it worked! :-)  I'm not sure that non-documentation of how netCDF
> handled conversions is something most users would pick up on....

You didn't rely on it, just your program.  :-)  Programs invariably
contain dependencies of which the programmer is unaware.

> > In version 3 of the netCDF interface, we tightened-up the specification
> > by, among other things, specifying what will happen in certain
> > value-conversion situations.  
> 
> This is a "Good Thing"! - don't be defensive.  It just needs to
> be communicated to users, that's all.

We tried to communicate it in the "Type Conversion" section of "Appendix D
- NetCDF 2 FORTRAN Transition Guide" of the "NetCDF User's Guide for
FORTRAN".  Sometimes with succeed; sometimes we fail.  I think it's
impossible to write a single document that will highlight important
changes for all categories of users.

> Now that we know about it, we can certainly address it.  In the past, 
> we were quite happy to be "in awe" of netCDF's robustness in handling
> these things (ie, ignorance was bliss :-).

Tell me about it.  (Sigh)  :-)

> > Even though we should know better, all programmers with any experience
> > have written code that relys on unspecified behavior- (I've done it more
> > times than I care to remember).  If there's anything we can do to make
> > the update easier, please let us know.
> 
> No sweat.

OK.  Keep in touch.

--------
Steve Emmerson   <http://www.unidata.ucar.edu>