[Met_help] binary for linux

[John Halley Gotway]

Theo,

Thanks for filling out the registration information.  Let me respond to your request with a few pieces of information.

First, the current release of MET is version 1.1, which was released on 7/11/2008.  However, the next release of MET, version 2.0, is expected to be out sometime during the first week of April.  We've
finished the development on it and are currently doing testing on it.  I'd suggest waiting to run version 2.0 when it become available.

Second, I understand your frustration in setting up your environment to get WRF to compile and run correctly.  I assist in the WRF testing, and our group is tasked to test WRF using many different
platform/compiler/configuration combinations.  So I feel your pain.  However, MET really is a much less complex set of tools than WRF. Since it doesn't require great amounts of processing, it's all
single threaded.  Therefore, there's no need to compile with tools to handle distributed memory - like MPICH or OpenMPI - making the installation process more simple.

Third, while we do our best to test as well as we can, there will inevitably be bugs that show up.  If you were to be running pre-built binaries, you'd have no way of updating your code to fix the bugs.

You're the first user to request pre-built binaries for MET, so we currently don't have any available set up for that.  We do our development work primarily using Debian Linux - Etch release.  While
we could make pre-built binaries available for it, we don't have access to many other platforms to make other types of pre-built binary packages available.  I really have no idea if what was built on
Debian Linux - Etch release would run on Fedora 10.

So here's what I recommend:
(1) Unless you're in a really big rush, wait for the release of METv2.0 in the next few weeks.  We'll email you when it's released.
(2) When METv2.0 is available, I'd really recommend installing it locally yourself with whatever assistance you need from us at met_help at ucar.edu.
(3) If you'd like to try the pre-built binary route instead, I could provide you with what was built on Debian Linux - Etch release, and you could try it out.  But you wouldn't be able to access any
bug fixes that way.

As I mentioned, I've found compiling MET to be less complex and quicker than compiling WRF.  Here are some compiling instructions we have as part of our online tutorial:
http://www.dtcenter.org/met/users/support/online_tutorial/compilation/index.php

You'd need to compile NetCDF (3 or 4 commands), GNU Scientific Library (3 or 4 commands), and BUFRLIB (3 commands).  And then to compile MET, you'd need to set 3 or 4 things in the top-level Makefile,
and type one command to build it all.  We're happy to provide as much assistance as you need to get it working well on your system.

John Halley Gotway
johnhg at ucar.edu

Theo Carter wrote:
> Hi,
> 
> I wonder if you are able to help me or point me in the right direction, I am
> an operational forecaster at the Dubai Int. Airport, and have been tasked
> with evaluating the output from our WRF ARW model (which was built and
> supplied by a private firm..). I would like, if possible, to skip the how to
> build a package on linux bit and try and obtain a binary of the MET tools
> program that has all the dependancies statically linked already in an
> executable which I can then install and run on Fedora 10 or whatever. Who
> can I speak to about this? I am sure someone has already built it, or would
> find it a very short job to do so.
> 
> Then I can rather learn how to use the program itself, which appears
> non-trivial.
> 
> I have read through the user's manual and have had a preliminary go at
> building the necessary other parts, and it looks like a dependancy "hell"
> program to build. (I  have built WRF in NMM and in ARW before on linux - so
> I know how much time is spend just on getting the various dependancies and
> compiler problems sorted).
> 
> Thanks kindly,
> Theo Carter
> 
> I apologize if this sounds too moany - I have just spent so much time on
> building and getting WRFrunning, that I am almost gunshy at trying to build
> this one now.. But it has to be done one way or the other.
> 
> 
> 
> ------------------------------------------------------------------------
> 
> _______________________________________________
> Met_help mailing list
> Met_help at mailman.ucar.edu
> http://mailman.ucar.edu/mailman/listinfo/met_help