[netCDF #YBG-623807]: Bug report vs. Netcdf 3.6.3: Compiler warning.

> NetCDF team,
> 
> This may be insignificant.  This is at the compiler warning level
> only.  All tests from make check passed.  However, this seems to be
> a newly added compiler check that you might want to know about.
> 
> Requested bug report items:
> 
> 1. VERSION:  -n 3.6.3
> 2. Complete output from "make" etc.:  See two attached log files.
> 3 & 4: Omitted, not requested per criteria on the problem reporting
> web page.
> 
> .---------------------------------
> Problem description:
> 
> This may be a new warning message from the latest version of
> gfortran, so you may be seeing more of this.  While building Netcdf
> 3.6.3 from source, I noticed the following warning near the
> beginning of make.  There are two instances in the same source file;
> this is one of them.  See the top of the attached make check log file.
> 
> netcdf_text_variables.f90:284.93:
> Included at netcdf.f90:51:
> 
> numDims  ) = (/ 1, (product(localCount(:counter)), counter =
> 1, numDims - 1) /)
> Warning: DO loop at (1) will be executed zero times
> 
> I used gfortran 4.5.0 which is cutting edge.  I have noticed that
> gfortran developers are increasing error checking over time, so this
> may be a result of some newly added compiler check.
> 
> A similar build on Mac PPC with gfortran 4.4.0 did NOT incur these
> warnings.  I did NOT check any Netcdf version 4.x builds.

Thanks, this turned out to be a harmless error, but changing the code
fixes the warning, so it should not occur in the future after this
change gets into the next snapshot release.

--Russ

Russ Rew                                         UCAR Unidata Program
address@hidden                     http://www.unidata.ucar.edu



Ticket Details
===================
Ticket ID: YBG-623807
Department: Support netCDF
Priority: Critical
Status: Closed