[daip] CUTOFF in CCEDT

[Eric Greisen]

Claire Chandler writes:
 > Has the meaning of CUTOFF in CCEDT been changed?  It used to be that 
 > leaving CUTOFF at 0.0 removed all negative clean components (and indeed 
 > the help file says this is what it does) but I've run PRTCC after CCEDT 
 > and found negative components remaining unless I set CUTOFF to some very 
 > small number.
 > 

The change was inadvertant and I am taking it out.  You could even set
CUTOFF to a small negative number and therefor allow weak negatives
but disallow deep negatives.  It will be fixed around 10 pm tonight.

I did overhaul the task in many other ways:

11903.  August 6, 2005          CCEDT                  Eric
        Corrected numerous problems:
        1. Made CCBOX be in a common so that its dimensions could
        change should the automatic splitting be done.
        2. Fixed history to tell the boxes correctly in both arc sec
        and pixels.  Fixed formats to fit.
        3. Put tests in to make sure that the number of boxes and the
        number of output files are the same when writing multiple
        output files.
        4. NCCBOX as a negative bvalue could be used in the past to
        create data in an object.  Error fixed by going to the common.
        5. Changed to set CCBOX in arc sec to look like like BLC/TRC
        in those units and corrected the test for component inside a
        box.  The old test assumed boxes in the correct order and axis
        increments < 0 in X and > 0 in Y.
        6. Changed code to do all the work of summing fluxes only when
        CUTOFF is being used.
        7. Some counters assumed that there were <= 10 boxes despite
        the parameter MAXBOX used elsewhere.
        8. The automatic algorithm changed the size of search boxes by
        too much for each minor iteration.  Limited the change to
        10%.  This gets more realistic boxes.
        9. Changed the automatic algorithm to set CCBOX and NCCBOX -
        so that the history file could contain the correct boxes
        rather than some adverb values that were ignored.
        10. Changed the automatic algorithm to do only one pass -
        previously it would do overlapping boxes - such as a large one
        with a smaller one taken out of the middle!
        Moved nowhere.

Eric