There's an obscure feature of Test::Harness and Test.pm I stumbled on
recently. You can declare that certain tests are supposed to fail,
they represent tests on features which haven't been implemented yet or
bugs which have yet to be fixed. The syntax is alittle weird, but
that can still be fixed as its undocumented. Basically, it looks (or
it will look) like this:
1..5
ok 1
ok 2
not ok 3 # TODO fix the hydrospanner
ok 4
ok 5
And Test::Harness will determine that a successful test.
Now, the first thing you do when you get a bug report is confirm that
you actually have the bug. Currently this is done by providing a bit
of test code, but it would be much better if instead a test patch was
provided (or generated later).
We then have an executable TODO list. This creates some interesting
possibilities. One is posting a weekly nag to p5p noting what's
outstanding. Another is often a bug dissappears as part of another
fix, but goes unnoticed. This would not happen if there's a TODO test
in place.
There's not a whole lot of technology involved in getting this going.
The only real necessities are mostly social, and a way to associated a
given individual test with a bug.
