From: Ron Jeffries <ronjeffries@...>
Date: 2001-12-17T04:35:42+09:00
Subject: [ruby-talk:28704] Re: John Roth dolt (  Re: A challenge to proponents of Unit Testing. )

On Sun, 16 Dec 2001 17:22 +0000 (GMT Standard Time),
brangdon@cix.co.uk (Dave Harris) wrote:

>He was citing 100% as an exit criteria. I've not read his book, though. I 
>get the impression he is achieving it by, well, testing. Writing some 
>tests, then measuring the coverage, and if it is less than 100% writing 
>more tests and repeating. In other words, his methodology does not 
>naturally produce 100% coverage first time; he only gets that after using 
>feedback from the coverage tool.

It's certainly one way to get 100%. I suspect that in most projects,
100% would be miraculously high. I would also suspect that without
serious code inspection, 100% is almost impossible to get, since you
would have to cause exceptions that are perhaps literally impossible,
but have to be coded for, as in Java.

When I'm working incrementally, with the exception of that exception
thing, I seem to get really close to 100% kind of "automatically",
with the added advantage that the test-first mode is guiding the
interface, the testability, and -- of course -- the moment-by-moment
correctness of the program.

Anyway, 100% isn't THE thing, but it's a good thing.

Ronald E Jeffries
http://www.XProgramming.com
http://www.objectmentor.com