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

[netCDF #KGN-827851]: libnetcdff ??

> Dear Ed:
> 
> address@hidden> wrote:
> 
> > >
> > > Dear Sirs:
> > >
> > > With netCDF Version 4.1.x, I am having a _lot_ of problems
> > > from the environmental community's use of the Models-3 I/O API
> > > data access library, which is built on a number of lower-level
> > > platforms including netCDF and PVM3.  This library has been in
> > > extensive use in the environmental modeling community since
> > > 1993, and is the official data access library for the EPA's
> > > emissions and Eulerian air quality models.
> > >
> > > Suddenly, all of the model-building and tool-building
> > > facilities are breaking:  whereas previous versions of
> > > libnetcdf.a stood alone, now new libraries, including
> > > particularly libnetcdff.a, are required -- making it
> > > impossible to construct a Makefile that works simultaneously
> > > for netCDF v3.x and v4.x.
> > >
> > > At least with the other extra libraries (HDF, etc., it is
> > > possible to configure the netCDF build so as to turn off
> > > these features.  How can my community configure it to
> > > construct a _single_ C-and-Fortran "libnetcdf.a" ??
> > >
> > > Sincerely,
> > >
> > >
> > > Carlie J. Coats, Jr., Ph.D.
> > >
> > > EPA/BAMS/MCNC Models-3 I/O API Systems Architect
> > >
> > > Chief Systems Architect                     address@hidden
> > > Environmental Modeling Center                   address@hidden
> > > Baron Advanced Meteorological Systems, LLC.
> > >
> > > 1021 Main Campus Drive, Suite 300                     campux box 7268
> > > Raleigh, NC 27606           phone: (919)515-0762 / fax: (919)515-0767
> > >
> > >
> >
> > Howdy Carlie!
> >
> > I have a hard time understanding why it is so difficult to modify Makefiles
> > to use -lnetcdff instead of -lnetcdf. What am I missing here?
> >
> > Thanks,
> >
> > Ed
> >
> > Ticket Details
> > ===================
> > Ticket ID: KGN-827851
> > Department: Support netCDF
> > Priority: Normal
> > Status: Closed
> >
> ^^^^^^^^^
> 
> It's a horde of users building models from source.  They don't
> know how to modify Makefiles; they have a mix of different
> netCVDF versions, and with your having changed the default
> netCDF build-behavior -- without notice and without proper
> documentation -- it is impossible for me to _give_ them Makefiles
> that will work with both (3.6.x, 4.x.y) netCDF versions.
> 
> Googling, I notice that the WRF met-model community has
> similar problems (though a more sophisticated user-base).
> 
> If nothing else, given that the default behavior was to build
> unified libraries, so that building separate libraries required
> an affirmative --enable-separate-fortran" configure-flag,
> changing this default behavior <STRONG>without any usable
> notice</STRONG> is extremely hostile to the user community.
> (Googling, the only clear notice I can find is on Florida State
> U's high performance computing web-site!)
> 
> You may be paid to deal with user issues, but I am not -- I'm an
> _unpaid_ volunteer maintaining an open-source data access
> library for a host of EPA, State, regional and local environmental
> applications, and you are gratuitously stealing my time, forcing me
> to spend it dealing with naive users who don't (won't) understand
> what is going on.
> 
> And you just mark the problem "closed" without any discussion ????
> 
> --
> 
> Carlie J. Coats, Jr., Ph.D.    address@hidden
> Chief Systems Architect
> Baron Advanced Meteorological Systems, LLC
> 920 Main Campus Drive, Suite 101
> Raleigh, NC 27606
> phone: (919)424-4444 / fax: (919)424-4401
> 
> 

Charlie,

The support system marks every issue "closed" as soon as I reply. That's 
because in many cases my response solves the issue, and people don't reply. In 
other cases people have already solved the issue, and never reply. So closing 
each issue as we reply is just a way to prevent our support system to be 
clogged with a thousand open issues at once. When and if a user replies, the 
issue is automatically reopened. As you can see, what we are doing here is 
discussing the issue, and attempting to find a good solution for all users.

I am sorry that you don't feel you were adequately notified about the separate 
fortran library issue. I agree, I did not announce this issue prominently 
enough. I have made a note to prepare send an email to the netCDF mailing list 
with a better explanation of this change.

You are not alone in the problems that you are having, so I am trying to come 
up with some reasonable solution that will work for everyone.

Thanks,

Ed

Ticket Details
===================
Ticket ID: KGN-827851
Department: Support netCDF
Priority: Normal
Status: Closed