[daip] AIPS problem with VLBA RFI post-processing

[Jim Ogle]

The VLBA RFI tests are post-processed using a set of scripts that automates running the data through AIPS. The scripts were created in 2014 and has been updated slightly in the years since to deal with OS and Python changes but for the most part it's unsupported software. 

Earlier this year the operators running the scripts reported seeing an "illegal instruction" error in the output but the scripts still ran and produced the normal output.  An example of that error:

DADEVS.PL: This program is untested under Perl version 5.016
sh: line 1: 199903 Illegal instruction     aips notv pr=1 > LI2403AR.6UW-aipslog.txt < LI2403AR.6UW Creating the AIPS run file "LI2403BR.6UW"

Then in April the operators reported that the scripts were unable to run, returning numerous 'file not found' errors. This proved to be due to an update to the ops-trunk environment Walter Brisken had made which didn't propagate fully. Once that was addressed the scripts appear to run but the final ones spit out a bunch of "no data found" errors which appear to be due to the first scripts not producing anything. 

Walter attempted some changes to update the scripts to Python3 to see if that would fix the problem but this did not appear to have any effect. He revered the scripts back to their prior version and wondered if the issue might be with changes in AIPS.  His suggestion was to have someone manually run AIPS based on what the scripts do and see where it fails, but only one operator in VLBA has any AIPS experience and they have not had time to try this so no further attempts at processing have been made since mid-April. 

The current versions of the scripts are located in /users/vlbsoft/software/OPS-TRUNK/bin/ and are configredrfi.py, create_dirs.py, dygraph-combined.js, generate_webpages.py, mystyle.css, parse_listr_scans.py, plot_total_power.py, redrfi, and write_aips_script.py. An info file written by the original creator describing the scripts can be found at /home/elbow2/rfi/redrfi/rfi-plots-info.tex (there's also a PDF version of the info file there.) That directory also contains early versions of the scripts themselves.

The operators run the scripts logged onto swc000 as difx in the ops-trunk environment ('ot'). The most recent attempt, which failed, was run in /home/elbow2/rfi/li2403/ while a prior successful attempt for the same band can be found in /home/elbow2/rfi/li2402/ . The li2403 directory contains an extra _pycache_ subdirectory, this was created by Walter's attempt to update the scripts to Python3 so it's not seen in the other directory. 

Jim