Re: Strawman comments

• To: netcdfgroup@xxxxxxxxxxxxxxxx
• Subject: Re: Strawman comments
• From: gabor@xxxxxxxxxxxxxxxxxxxxxx
• Date: Tue, 13 Oct 1992 11:30:49 CDT

> Tim Holt has made numerous comments about the "conventions.info" and
> the brief variable names used (like T, Tmin, etc.).  Personally, I
> think the variable names should be short and that extra information be
> put into attributes.  I like the idea of naming variables the same way
> they are used in the dynamics equations, or using well known acronyms.
> Nothing will frustrate a programmer more that having to type
> unnecessarily long names that add little to the understanding of what
> the variable is.  And if you are trying to deal with netCDF files in a
> generic way, you can grab temperature without worrying about whether
> it is "temperature_from_ship_type_457b2" or
> "temperature_from_groundstation", you just grab "T".
> 
> Information on how the data was derived, limitations, etc., more
> naturally belong in an attribute than in a variable name.  Certainly
> it is important information to keep with the data, but the variable
> name is the wrong place for it.

I feel both side right, let'  make a  compromise.   In addition, let's
consider a rather technical aspect: Partial or whole content of NetCDF
files are often post-processed in environments like IDL where we would
like to use the same variable names as  in the imported  NetCDF files.
Unfortunately, environments   often impose a  limit  on  the length of
varaiable names.  And  what is worse, the compiler  may   truncate the
names by itself: over a certain limit limit variable names will not be
distinguishable for the compiler.  Therefore, the user himself/herself
must truncate the long variable names before the compiler would do so.
The user must set up  a "home-convention" which  may or not follow the
convention we have been  coming up  here.   On the other hand, I would
like to  avoid  short abbreviated   names   like we  use   in  fortran
subroutine libraries (brrr, it is cool even think of it).  I know, all
this dos not affect the generality and content of information captured
INSIDE NetCDF files, however, it may turn out  to be inconvenient. (At
least, it already did for me.) I would like to see not-too-long names,
that are nicely  readable, informative, and  not too long for the most
programming environments.

Gabor Fichtinger
Scientific Visualization Group,         |campus: BRC.CMS.1-154, 78700 
Center for High Performance Computing   |email : gabor@xxxxxxxxxxxxxxx
The University of Texas System          |Ph/1  : (512) 471 2409
Balcones Research Center, 1.154 CMS     |Ph/2  : 1-800-262-2472/2409
10100 Burnet Road, Austin, TX,78758-4497|Fax   : (512) 471 2445 
                              __o
             __o             -\<,
            -\<,  __________O / O
___________O / O