[CIG-MC] python parser issue

[Craig O'Neill]

Hi Walter (at least I think you still answer these!)
I've been trying to install the new version of CitcomS on our new cluster.
I had a few issues which I might need some direction on.
First, setting PHDF5_HOME as an environmental variable is not enabling
Citcom to find the hdf installation (I believe its still looking  in
/usr/lib etc). Is this a known issue?
Second, I am having problems with the post processing. In order to
post-process using autocombine.py, I need the pidXXX.cfg file, and a
mpirun.nodes machine file. I've followed the instructions in section
3.5.2.1 about including a machine.cfg file, which works and produces a
mpirun.nodes file.
However, I can't process this.  The error I get is:

 ../../visual/autocombine.py pid13613.cfg mpirun.nodes1 100
Traceback (most recent call last):
  File "../../visual/autocombine.py", line 80, in <module>
    parser.read(inputfile)
  File "/home/coneill/CitcomS-3.1.1/visual/parser.py", line 143, in read
    self.__read(fp, filename)
  File "/home/coneill/CitcomS-3.1.1/visual/parser.py", line 240, in __read
    raise e
parser.ParsingError: File contains parsing errors: mpirun.nodes1
	[line 12]: 'n012\n'

It seems the issue, upon looking at the parser.py file, is that it expects
a "x=y" format, but the machinefile has a single entry format (one node
listed per line, no "="), which makes it impossible to parse using this
routine.

Am I missing something obvious here? Has anyone else had problems parsing
these files? Surely everyone else is processing their data so I imagine
I'm doing something wrong, I just can't see what.

Many thanks

Craig



-- 
Dr Craig O'Neill
GEMOC, Department of Earth & Planetary Science
Macquarie University, NSW 2109
AUSTRALIA