On Mar 25, 2009, at 20:44, James Hawkins <[email protected]> wrote:
On Wed, Mar 25, 2009 at 11:50 AM, Paul Vriens
<[email protected]> wrote:
James Hawkins wrote:
On Wed, Mar 25, 2009 at 8:25 AM, Paul Vriens <[email protected]
>
wrote:
Hi,
This fixes bug 17843 but I'm not sure it's a 100% correct. James
didn't
change
this just for the fun of it.
If you're unsure of the correct fix, you should write a test case
that
fails without your patch and succeeds with your patch.
AJ already put in a fix (similar to mine).
I don't want to sound negative but you added loads of test cases
that still
didn't prevent this regression.
That is pretty negative. Your comment implies that the failure of the
current test suite to be comprehensive enough means we shouldn't add
more tests, which is incorrect. No test suite is ever complete. Any
change to the code that can be tested, should be tested, and there are
very few places that can't be tested with enough ingenuity.
That's probably the downside of todo_wine.
With my last patch in the series (which needs to be reworked and more
tests added for), there were only two todo_wine's left, and they are
unrelated to this problem.
--
James Hawkins
Sorry if that came on too strong. I'm all for more tests. My point was
that todo_wine can be a 'dangerage' thing sometimes (not specifically
in this case).
Maybe we should even make it mandatory that fixes for regressions
should be accompanied with tests, although I doubt that will be
feasible.
Paul
