[McIDAS #PSX-470067]: mcidas-x 2009 compile on Ubuntu question

Hi Tom,

Sorry to stop replying yesterday.  A friend was visiting and it was time
to hobnob and BBQ :-)

re:
> Tom.. see below. Thanks!
re: which virtualization package
> I am using VirtualBox, but I have exported to a vmdk file, which I have
> been able to run in VMware in the recent past, but I think I had to
> generate the vmx file.

OK, interesting.  I don't think I need to run your virtual machine...
I cranked up my instance of Ubuntu Ultimate 1.4 (old), and reviewed
the McIDAS build there.  I ran into the same problem that you did
when: execution of the script 'tclcomp' would result in an error
about mismatched quote marks.  My workaround was:

- edit ~mcidas/mcidas2009/src/tclcomp

  change:

  #!/bin/sh

  to:

  #!/bin/bash

- restart the build:

  make

What is interesting is that 'sh' on a lot, if not most, Linux systems is a link
to 'bash'.  Why changing the shell used by 'tclcomp' works is a mystery to me.
I guess that when one runs /bin/sh, there are some defaults used that are not
quite right.

Please try editing 'tclcomp' like I note above and see if your build goes 
through
to completion this time.

Comment:  I did _not_ run into the problem with syslog under Ubuntu Ultimate 
1.4.

re:
> Mac may still be an option to get away from dual boot, but see my
> comment about NAWIPS-II below.

OK. I like Macs.  I just wish that Apple would take the bug that is affecting
the LDM more seriously.  Steve has spent a LOT of time trying to get them
to find and fix their problem, but with no success so far.


re: trend in the Unidata community for Linux users to migrate to CentOS.
This may be a compelling move since AWIPS-II is being developed under RedHat
Enterprise, and CentOS is the free distribution of RH Enterprise.

> OK, that's news to me. Perhaps I should be moving in that direction instead.

If you are going to embrace AWIPS-II/NAWIPS-II, then you should start
thinking about outfitting your lab with Linux machines running RedHat
Enterprise or CentOS.  Also, the machines will need to be well configured
with RAM (probably 4 GB or more), and will need to have a high quality
video card with at least 256 MB of RAM and that supports OpenGL 1.2.

NB:  the hardware/software needed for AWIPS-II is still a moving target,
so it would be wise to wait a bit before starting to purchase new equipment!

re:
> I still don't have a sense of the timetable on NAWIPS-II. Is this
> something expected to be widely available to the Unidata community
> within the next 18 months?

I think so, but the schedule for AWIPS-II deployment/acceptance in the
weather service has been slipping routinely.  We (the UPC) are taking
part in the evaluation program for the test builds (drops).  We (Michael
James) have been successful in running build 10 on a 32-bit CentOS
machine.  Comments we hear during our monthly telecons indicate that
64-bit support should be available at some point, but it is not
fully supported now.  We believe that support for 64-bit _must_ occur
especially given the need for access to lots of RAM (just like the
IDV).

re:
> If yes, then perhaps I need to revisit this
> whole topic with our IT people and try to get them to stick with dual
> boot systems. They are short staffed and use a significant amount of
> time dealing with these dual boot systems. If it is further in the
> future, then I may be less concerned about this issue.

The entire issue of what AWIPS-II will run on eventually is very murky.
One of the BIG problems is that it currently uses compiled C code that
is "wrapped" by Java.  This means that a distribution will need to
have the C code compiled for the target platform (i.e., 32-bit will
be different from 64-bit; object modules for Linux will be different
from object modules for Solaris and MacOS; etc.).  Since the weather
service forecast offices currently use 32-bit RedHat Enterprise Linux
machines, the initial development of AWIPS-II is aimed at that platform.
The Unidata community is _not_ the primary users that the weather service
has to worry about, so their planning does not encompass different versions
of Linux, Solaris, MacOS, or Windows.  This may change over time, but
I would not bet a lab's worth of machines on it.  Suffice it to say that
the safe bet at the moment is to use 32-bit RedHat Enterprise or CentOS
since it is RH Enterprise that is free.
 
re:
> Before we bother with the vmdk file, I'd like to see your response to
> the NAWIPS-II question. That may change my thinking entirely about how
> we proceed. Perhaps I should drop the virtualization approach.

If my comments above are not sufficient, we could discuss the state of
affairs in a phone call.  Please let me know if you would like this,
but also please remember that hardware configurations have not been
set by the AWIPS-II developers (Raytheon) yet.  We are trying to anticipate
what will likely happen, but we do not want to give premature, incorrect
information to the community since some might go out and start purchasing
new equipment that could turn out to be wrong for the job.

For my own information, please give the 'sh' -> 'bash' change in 'tclcomp'
a try and let me know how the rest of the build goes.  Thanks!

Cheers,

Tom
--
****************************************************************************
Unidata User Support                                    UCAR Unidata Program
(303) 497-8642                                                 P.O. Box 3000
address@hidden                                   Boulder, CO 80307
----------------------------------------------------------------------------
Unidata HomePage                       http://www.unidata.ucar.edu
****************************************************************************


Ticket Details
===================
Ticket ID: PSX-470067
Department: Support McIDAS
Priority: Normal
Status: Closed