Nuclear Python?

[achrist at easystreet.com]

"John J. Lee" wrote:
> 
> Mind you, if you've ever attempted to read some of the huge Fortran
> codes that get written by scientists, you'll understand why it might
> not always be a good idea to touch them in order to make them easily
> wrap-able!-)
> 

Python is noted for its poor machine efficiency vs Fortran, etc. 
However, these huge Fortran codes are noted for their poor efficiency
when run all weekend on a supercomputer with bad code or incorrect
parameters.  Wrapping some UI, preferably GUI with charts and graphs,
around them  to see if they are running at all reasonably is a big
improvement in efficiency.  A large proportion of the runs intended
to crunch numbers until a week from Tuesday give away their deficiency
of  correctness or relevance within a few minutes after they start,
if  they are monitored astutely. 

This is a humongousish increase of  efficiency -- cancelling a multi-day
run that was going to be a water-haul. 


Al