This file summarizes how to use the 'regression.py' script for testing
CASA.

The working directory can be chosen for convenience as all the
*_regression.py scripts in $AIPSROOT/code/xmlcasa/scripts are copied
into the working directory, before execution commences, when you run
'casapy -c $AIPSROOT/code/xmlcasa/scripts/regression.py'.

The file 'regression.py' will need to be modified if the data for some
tests is not present in $AIPSROOT/data.  You will need to set
DO_theTestName=False for all tests with missing datasets.

After regression.py terminates, a summary of regression test results
(test name, pass/fail, runtime, runtime/expected runtime) is left
within the working directory in an output file named
'REGRESSION'+datestamp+'.log'.

To add another test to the 'regression.py' script, first make sure
that the new test script defines startTime, endTime and regstate.  For
example, the 'imagetest_regression.py' script contains:

	import time
	startTime = time.time()
	regstate = False
	imagetest()              # calls the actual test script
	endTime = time.time()
	regstate = True

At the front of the 'regression.py' script, add a line to indicate
that your new test script should be ran.  You need something like:

###############################
# Regression 
# 
###############################
[...]
DO_IMAGETEST            = True
###############################

Now modify scriptlist and ratedict using an existing test, such as
IMAGETEST or NGC5921, as an example of what is needed.

If your source test script appears in scriptlist (all upper case
letters as 'IMAGETEST', for instance) and DO_IMAGETEST=True, this
causes the regression script to try to execute a file called
'imagetest_regression.py'.  It uses startTime and endTime to determine
the script running time which is compared to the expected value
recorded in ratedict.  For instance, ratedict has the key value pair
'IMAGEPOLTEST':41.  Note that the expected runtime will be
machine/load dependent.