Murphy's Law rules, period.
Advances in software methodology in recent years
include the notion that one should write your test-cases FIRST, before writing
the actual functioning code, after which you should write a test harness
arranged to make you see the tests fail, so that you know that failure can be
reported at all. It's actually not a new idea (I think I first heard of it
in the late 70s). It just took a while to catch on. One of the
reported advantages of this approach is that you define your interfaces more
carefully - interfaces being where systems (including aerospace systems) most
often break down or go awry.
Of course, with software, you have the advantage that
you can re-run your tests effortlessly with each build - you can even automate
the process, so that no build is reported as complete unless all tests have
passed. Testing and re-testing costs serious money when you're developing
exotic hardware, and it costs a lot more money when the test conditions are
extreme and hard to reproduce. Still, a "test-first/test-often" approach
to building spaceworthy hardware might be worth investigating, even if the added
costs forced some compromises in design and in performance.
-michael turner
|
- Murphy's Law Rules Outer Space LARRY KLAES
- Michael Turner