[evlatests] Software time March 29

[Barry Clark]

All in all, a pretty constructive day.

However, a puzzle.  Last week at X band, I set all the delays pretty reasonably 
at X band for IFs A & B.  The EVLA and VLA delays today were systematically
offset by about 30 ns in IF A.  When the file was restarted, the difference
went away.  (Before the delay finding file was started, we had a few 
minutes of running with only the EVLA antennas.  Maybe some initialization
problem with the correlator controller?)

Set delays for all four IFs at X band.  (Improvement in delay solver
helped.)  Data taken for delays at C band, but not yet entered.  Many
are pretty small anyway.  Delay solver and delay display software still
need improvement/debugging, respectively.
 
Bug found in the course of this - removing an antenna from the Executor
apparently does not change the baseline list in the correlator.
 
After a new executor takes control, it sometimes takes many minutes 
(more than 10) before vispipe and idcaf get synchronized again.  This 
may be made better if I take care to always send stuff to the correlator 
at the same time of the 10 second cycle (but again, this is a bit 
hit or miss anyhow, so that appearance may be coincidental).  Starting 
a new script within the same executor does, somehow, not seem to  
require the lengthy resynchronize interval. 
 
About half an hour data taken on Venus, starting about 21:07.  Somebody
might have a look at it and see how reasonable it is.  The header stuff
will be highly unreasonable - idcaf hasn't a clue about what the  
Executor is up to.