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

Re: Use of patch. Where are ncstat, etc.?



> Organization: CSIRO Division of Atmospheric Research
> Keywords: 199404210917.AA04686

Harvey,

> On Thu, 7 Apr 1994, Russ Rew wrote in reply to my report of apparent
> bug in ncgen:
> 
> > This was actually a bug in ncdump that was fixed by a patch file posted to
> > netcdfgroup last year, and available from 
> > 
> >     ftp://ftp.unidata.ucar.edu/pub/netcdf/2.3.2-patch2
> > 
> > Hence you must be using an unpatched netcdf 2.3.2 distribution, to which
> > patches 1 and 2 from the above ftp directory should be applied.
> > 
> > __________________________________________________________________________
> 
> I have now done this and it fixed the bug.
> 
> But it is not stated in installation procedure that patches need to be
> applied.  It really would be better if netcdf.tar.Z contained the latest
> (patched) version.  Perhaps another digit is needed on the version number
> (e.g. 2.3.2.2) to indicate minor patches.
> 
> The latest patch (2.3.2-patch2) is dated Wed Jun 23 13:27:24 1993, so the
> code is very stable now.  There may well be good reasons for using
> patches, but surely not for 10 months!

Yes, it's been a lot longer than we intended for a new release.  Furthermore
there have been some additional minor bug fixes (especially to the C++
interface) that should either get into a new patch or a new release.  I
favor the latter, but our procedures call for testing each new release on
numerous platforms, fixing or documenting all the problems that result, and
restesting with the new fixes until we reach convergence.  This has taken at
least two weeks and more typically a month for past releases, especially
when lots of changes and additions have accumulated.  We've been working
flat out on a different project (the Unidata LDM) that currently has higher
priority.  I hope to get back to netCDF work in late May or June and create
a new release then.

In the mean time, I think I'll accept your suggestion and apply the two
patches to the current distribution.  I'll leave the patches available but
change the documentation at the top to indicate that they've already been
applied to the distribution.

> Another minor query re ncstat & ncextr.  I had to do a full installation
> because I had deleted the old source code.  I was surprised that ncstat &
> ncextr were excluded.  I checked the directory date on these & I installed
> them on Nov 17 1992.  They were obviously included at that time.  What
> happened to them?

They were never included with the netCDF release, but rather were part of a
separate "netCDF operators" package that we've set aside until we can get
development resources to continue work on it.  It's still available as 

    ftp://ftp.unidata.ucar.edu/pub/ncopers/ncopers.tar.Z 

> Thanks Russ for the help with the ncdump bug.  I really do appreciate the
> excellent software & service you provide.  We are using netCDF to store
> all our GCM and ocean model results & it working very well.

Glad to hear it, and thanks for the useful feedback and bug reports.

> Harvey Davies,                              Home: +61 3 772 5199
> CSIRO Division of Atmospheric Research,     Work: +61 3 586 7574
> Private Bag No. 1, Mordialloc,               Fax: +61 3 586 7600
> Victoria 3195,  Australia                 E-mail: address@hidden

__________________________________________________________________________
                      
Russ Rew                                              UCAR Unidata Program
address@hidden                                        P.O. Box 3000
(303)497-8645                                 Boulder, Colorado 80307-3000