I am lucky in that my customer sites do the testing for me.

We have a test schedule which is worked through and
hopefully all aspects of the system are tested. As bugs are
identified, they are logged, and a bug report is generated
for the version under test. I then process the bug report,
identify and clear each problem, and pick up any 'off track'
items that may become involved. I then produce a set of
notes for each new build, covering the previous bug report
and flagging any additional changes.

Testing off the track of the test schedule is also carried
out, and where problems are identified, the test schedule is
expanded.

Wish list items then get included as 'bugs' and get
processed in the same way.

I keep a copy of the source of each build ( zipped ) and if
a problem is identified that has appeared later, the version
at which the bug arose can be identified, and Beyond Compare
provides a fast analysis of the changes.

I don't know how one could test a system other than using
this approach. Could software actually do anything to help?

--
Lester Caine
-----------------------------
L.S.Caine Electronic Services