[daip] possible lpcal bug

[Eric Greisen]

Cormac Reynolds wrote:
> hi,
> 
> I think I've found a bug in LPCAL. I'm using it with a model of the
> source so have set NMAPS=1 and NCOMP=200 to select just the first 200
> clean components of that model. However, If you set 'NCOMP' to a value
> greater than 10, it aborts with the message:
> 
> unctio> LPCAL1: TOO MANY MODELS ( >  10)
> unctio> LPCAL1: Purports to die of UNNATURAL causes
> 
> Looking at the code, it looks like there is some confusion between the
> number of models (i.e. fields) and the number of clean components per
> model.
> 
> I can work round the problem by first running CCEDT to select the
> clean components that I want, and then setting NCOMP=0 and the program
> runs fine, but it seems like that extra step of running CCEDT should
> not be necessary.

LPCAL was written by a student and has no one in the group at present 
who understands its insides.  It appears to me that the student has used 
NCOMP for the number of CC file versions to be used separately inside 
the task.  That concept is very deep in the task because it loops over 
NMODEL everywhere.  The parameters that actually control which CC 
components are used by UVMSUB are not set in the task, so all are used.
What confuses me even more is that an array that is dimensioned 
(2,MAXMOD,MAXVAS) in an upper routine and handled that way a lot, 
becomes suddenly (MAXMOD,NVIS) in a lower level routine.  The complex 
values are now lost somehow???

The whole thing looks strange - if you can get it to work somehow I 
think you should be grateful.  I will look at it more when and if I get 
time.  I have no data to test it with however.

Eric Greisen