[daip] Re: FREQID question

[Eric Greisen]

Mark Reid writes:
 > Eric,
 > 
 > 	It seems to me that the best overall solution
 > to this problem (which by the way has come up before)
 > is to make sure that tasks allow "FREQID=-1".
 > CLCAL, for example, resets FREQID=+1 if you set it
 > to -1.  I don't know why it does this.  If all AIPS
 > tasks allow FREQID=-1, and interpret this to mean
 > that one does not select by FREQID, then AIPS will be
 > much more flexible (and internally consistent).
 > 

Yes and then calibration at 1.3 GHz will be applied to data at 1.7
GHz.  The use of FREQID is to keep separate data that a priori should
be kept separate and I still believe that its presence in AIPS is the
worst design mistake we ever made.  The VLBA correlator makes the
choice of FREQID not our software so it is too late for use to undo
that except by writing a separate task to convert all FREQIDs to 1.
That could be done but I would worry about losing the info needed to
get correct velocities in the end.  The source table has lots of
generally ignored info in it - if each source is observed at only one
freqid, then one might be able to retain enough info to manage the
velocities...

To change each task to allow FREQID=-1 to mean ignore FREQID is to
invite all sorts of disasters - we would have to keep all cal, flag,
etc tables functions of freqid from 0 (any) to to max freqid (pretty
high number - 30 or so at least).  That would eat up a lot of logic,
core etc.  If you arrange for NRAO to hire more programmers there
might even be manpower to do something like this - but I am uncertain
that we would want to do it anyway.

The real solution is to observe enough calibrators that you do not
need what you are suggesting.  To assume that the delay, gain, phase,
and/or bandpass calibration applies in narrow bands to another narrow
band 100 kHz away is to trust our instruments more than you should.

Eric