[daip] Feature request for FITLD for VLBA data

[Eric Greisen]

On 03/30/2018 12:55 PM, Walter Brisken wrote:
> 
> Hi AIPS,
> 
> I have learned of an interaction of a previous version of DiFX with 
> FITLD that yields troublesome processing within AIPS.  I believe a 
> simple, and very save, change to FITLD could seamlessly correct for this.
> 
> The problem is that the IDI FITS file contains multiple FREQIDs which 
> are in fact identical.  Having separate FREQIDs precludes proper 
> operation of a number of tasks in AIPS and makes calibration nearly 
> impossible.  A reasonably small example file is
> 
>   /home/parallax/VLBA/BZ065B/VLBA_BZ065B_bz065bpass1_BIN0_SRC0_0_170303T211055.idifits
> 
> which has an AIPS friendly sym link
> 
>   /home/parallax/VLBA/BZ065B/GEO1.FITS
> 
> Running FITLD with default adverbs (except for DATAIN) yields the 
> abridged output shown below.
> 
> The proposed fix is for AIPS to renumber all equivalent FREQIDs to a 
> common FREQID.  I think the hard part is already done in FITLD: 
> identification of the matching frequency IDs.  So perhaps when a file is 
> being loaded, a mapping from FITS-FREQID to AIPS-FREQID could be 
> established, and then the data and relevant tables could be remapped 
> on-the-fly during read.  This could be optional behavior if so as to 
> allow the existing behavior (maybe there is a legitimate use of this).
> 
You do not want the existing behavior which numbers FQ 2 as 1, FQ 3 as 2
etc to FQ 17 as 16.  The logic to detect duplicates worked but the logic 
to set up the correction for it failed badly.  I have fixed it and added 
better messaging.  I will patch 31DEC17 for this.

Eric