[daip] A bit more on FXPOL

[Jim Ulvestad]

Hi Ryan,

It looked to me like Eric implied, but didn't actually say,
that there are some of the cases where we need FXPOL to
work where it MUST write a new file, due to (among other
things) the structure of the polarization data/tables.
In order to make the task general enough to handle almost
all the cases that can get thrown at us, the data handling
had to be done in such a way as to accommodate this
requirement.

We did have an internal complaint or two about doing things
this way, but the alternative was to have two significantly
different pieces of code, depending on which of the cases
of polarization re-labeling was done; since the person
working on the code had limited time remaining at NRAO, this
was not an option.  Even if it had been, I probably would have
argued against it for reasons of simplicity.

I'm not sure this explanation will help you feel better if
you don't have enough disk space, and Eric will probably 
tell me that I got something wrong (!), but at least it
might convince you that the consequences of writing a
new file were known and thought about before going in the
direction we did.

Best,

Jim Ulvestad