CVS update: samba/source
rsharpe at ns.aus.com
Wed Sep 11 16:23:00 GMT 2002
On Wed, 11 Sep 2002, REILLY,JOHN R (HP-Roseville,ex1) wrote:
> Richard -
> I'd be interested in your experiences, how you drove your unit tests,
> what coverage target you hoped to achieve, etc. I don't believe the
> specific numbers are as important doing the exercise causes. I think
> several things will be driven off this effort:
We had no target; we were just checking the coverage of our tests.
The tests we ran were a set of internal tests, plus smbtorture (those that
would run) plus some tests of mine.
> 1. Further populate Tim's pyunit tests (i.e. use coverage data to "test
> the tests")
> 2. Targeted code review (i.e. Why didn't we hit this code? Why did we
> hit this code so much harder than we expected? etc.)
> 3. Drive improved Samba quality (yada yada Mom and apple pie :-)
> Since my focus is printing, naturally, that is where I think the above
> will initially apply. A longer term goal 'might' be to produce some
> numbers automatically at the build farm.
I think that is a fine goal. I was warning against the view that high code
coverage is goodness and low code coverage is badness.
Richard Sharpe, rsharpe at ns.aus.com, rsharpe at samba.org,
sharpe at ethereal.com
More information about the samba-technical