[Met_help] Sharing a comment
John Halley Gotway
johnhg at rap.ucar.edu
Tue Jul 15 10:14:56 MDT 2008
Hello David,
Thanks for your comments.
I have good news for you. While MET was developed primarily on Debian Linux, it has been successfully built and run on several different Linux distributions. For example, we've run it on Redhat
Linux and CentOS, both over in the MMM Lab of NCAR. We've even had success building and running most of MET on a Mac. I'd actually be surprised if you were NOT able to build and run MET on a
relatively common Linux distribution. And with version 1.1, you should now be able to build MET on an IBM using the IBM compilers.
Building and running on Redhat should be no problem at all. I'll look to see how we should update the documentation to reflect that.
Regarding compilers, MET was developed primarily using the GNU compilers - which are very common and freely available. On Linux, it can also be compiled with the PGI compilers - since so many WRF
users use PGI.
Regarding the library dependencies, I agree, it is a bit of a hassle. We require NetCDF, BUFRLIB, GNU Scientific Library, and F2C or G2C (which is a pretty standard library). We've tried to balance
the hassle of requiring external libraries against having to reinvent the wheel. We have considered implementing a stripped down version of BUFRLIB internally and therefore removing that dependency.
But that would be down the road a ways.
And as for configuring MET, that's an area we need to work on as well. Currently, we have users edit a Makefile by hand before building. There are 5 or 6 things to set before building. In future
releases we hope to set up an interactive configuration tool that's a little more "slick".
If your group is interested in beginning to use MET, you should be able to do so now. If you run into any issues or problems during the build process, please feel free to write met_help at ucar.edu with
questions. We'd be happy to help.
Good luck,
John Halley-Gotway
johnhg at ucar.edu
David Novak wrote:
> Hello,
>
> First, I wanted to commend the group for the development of MET. It will
> be a tremendous resource. However, I wanted to share a general comment
> from one of our Science and Operations Officers regarding MET use in NWS
> Weather Forecast Offices (WFOs):
>
> "While this package has potential, it does not run on the forms of Linux
> commonly used in WFOs or WFO systems that run the WRF. It runs in
> Debian Linux which is old, and meets none of todays security protocols,
> not the Enterprise Redhat Linux we use. This package also requires
> alot of configuration and compiling . Numerous libraries, Fortran
> compilers and the like must be obtained, linked and installed. These
> may not be present on most Linux machines"
>
> It sounds like there are two requests:
> -compiling for Enterprise Redhat Linux
> -more user-friendly installation/configuration process
>
> I'm not sure if version 1.1 addresses these concerns, but perhaps these
> issues can be considered in future development.
>
> Thanks,
> David Novak
> NWS Eastern Region Headquarters
> Scientific Services Division
> Bohemia, New York, 11716
> _______________________________________________
> Met_help mailing list
> Met_help at mailman.ucar.edu
> http://mailman.ucar.edu/mailman/listinfo/met_help
More information about the Met_help
mailing list