??? 12/11/06 18:17 Read: times |
#129299 - I know of no testing that can prove that Responding to: ???'s previous message |
I know of no testing that can prove that something works.
Testing can, of course, show that something does not work, but you do not really need testing for that. Just try this mental experiment: You have created something that 'works', but will fail if interrupt b happens while interrupt a is executing lines 47 thru 49 and interrupt a happened while the main was executing lines 333 thru 334 in module y. NOW, write a testing procedure that will catch such a bug. I had to find such a bug once (not my code), the report was "a random one of 10.000 units fail once a month" Erik |