[daip] Fittp/Fitld impact on indxr

[Eric Greisen]

Arnaud Collioud wrote:
> Hi Eric,
> 
> 
>> There are very many things wrong with your AN table.
>> Two keywords, NUMORB and NOPCAL, have very wrong values.  When the AN table is copied by AN knowledgeable subroutines, these insanely large values will make columns containing an insanely large number of values. 
> 
> Should I put zero for these two parameters ?

NOPCAL is normally 2 although 0 at this point is okay.  NUMORB is the 
number of orbiting telescopes in this AN file.  I suspect that this is 0.

> 
>> Other things are wrong also - many antennas are located at the center of the Earth, you have new names for many antennas so almost none were recognized by a subroutine that knows 58 station identifications,
> 
> I did not notice this. It concerns AIRA, BADARY, HN-VLBA, HOBART26, KASHIM34, PIETOWN, SC-VLBA, WESTFORD, WETTZELL, YEBES40M. 
> 
> It is strange that extensively used stations like SC-VLBA, HN-VLBA or PIETOWN were not recognized. Is this due to the station name used in this session ?
> Can the list of known stations be changed or extended easily ?
SC and SC_VLBA are recognized but SAINCROI, STCROIX, StCroix, etc etc 
are not.  Computers recognize strings literally not by mental 
association.  The association algorithm would be to look for SC anywhere 
in the string but are there no stations with an SC in the name 
somewhere?   If correlators just make up names for the stations and 
change them all the time, we are not able to list them all.  I can add 
some more but you will note that PIETOWN is not in the pattern of all 
the other VLBA names - somebody is just making things up for the fun of it.

>> and the thing that causes the full collapse is that there are short integer valued columns (I suspect).  I should look at FITLD to see if it can fix that on reading a file...
> 
>> 31DEC11 has code to change short-integer columns into long-integer ones on input.  
> 
> Ok.
> 
>> Use 31DEC11 FITLD and I suspect your INDXR failure will be over. 
> 
> The INDXR error occurs precisely when using 31DEC11 FITLD. 31DEC07 INDXR runs smoothly but I have an error after that with ANTA (that is why I wanted to test the process with 31DEC11):
> 
> VANOIS> ANTAB1: Task ANTAB  (release of 31DEC07) begins
> VANOIS> ANTAB1: NXREAD: INCREASE PARAMETER MAXNX
> VANOIS> ANTAB1: Purports to die of UNNATURAL causes
> VANOIS> ANTAB1: vanoise      31DEC07 TST: Cpu=       0.0  Real=       0
> 
>
Your INDXR problem has changed from catastrophic failure due to an 
illegal format to something manageable.  You need to run VLBAFIX to take 
care of your extreme subarray condition.  Then INDXR will work.

Eric Greisen