[daip] Two curiosities from 1 Software correlator FITS file

[Eric Greisen]

Walter Brisken wrote:
> 
> 
> On Wed, 7 May 2008, Eric Greisen wrote:
> 
>> Walter Brisken wrote:
>>>
>>> This is not just plain concattenation -- the FITS builder allows 
>>> general changes of antenna IDs and freq IDs, source IDs, ... and 
>>> seems to do the right thing.
>>>
>>> For antennas that have otherwise no IM table entries for certain time 
>>> ranges should I make IM table rows with zero delay, or is it OK to 
>>> not include them?
>>
>> 1. GREAT!  The VLBA users wanted to just concatenate and that does not 
>> work. FITLD goes to great lengths but does want an EOF between groups.
> 
> It also puts HALF polar data in the right order (no FXPOL neede), sorts 
> data in time, calibration transfer data is in proper tables (no tbmrg 
> needed) and doesn't end up with unintentional subarrays, so overall it 
> should be gentler on the users.
> 
>> 2. I guess one can tolerate the warning messages.  The right thing is 
>> done - a zero is entered for the GEODLY and the other actions on the 
>> CL table values are done despite the "error".
> 
> Could this message be suppressed at some level (i.e., not printing every 
> instance, but perhaps one per source or so, or not printing it if there 
> in fact is no corresponding vis data) or are these changes more trouble 
> than they are worth?

I have found a logic error in FITLD - it set antup for the new antennas 
of a sample and then decided it had to write out for the previous source 
- hence the first record of source 7 had antenna 11 in it and so 11 was 
required for the first file's last scan.  I moved the antup setting to 
after the CL writing.  Note that this also meant that an antenna could 
be missed from a scan if it occurred only in the first record of that scan.

Eric