[OpenLayers-Dev] Patches Need Tests

[Christopher Schmidt]

Recently, we've been getting, and applying a large number of patches
from the community. These patches have helped to solve dozens of bugs
that might have otherwise gone undiscovered in the OpenLayers code base. 

However, our testing of these changes has fallen by the wayside. In the
past, I took the time out to write tests for most of the patches that I
committed -- anything that could be tested programatically. Recently, I
haven't taken this time out: patches have been committed which have no
tests, and this is the kind of thing that leads to regressions. 

I'd like to see more patches accompanied by tests. Understandably, not
all problems can be easily tested in an automated way. An example of
this is the init tiles issue, #480. I was unable to come up with an easy
way to write an automated test for this, so instead, I wrote a simple
HTML page which demonstrated the problem, with a description of what
should be seen and what shouldn't:

  http://openlayers.org/dev/tests/grid_inittiles.html

I understand that there are cases where this is the case, and I would be
much happier accepting a patch if it had even a non-automated test that
described the problem it was solving, and how to reproduce it, and what
it should look like if it has passed or failed. 

I'm speaking solely for myself here, but I'm leaning towards pushing
back on patch writers to write some kind of tests that go along with
patches where possible. An example of this is ticket #496. In that case,
I was able to write tests:

  http://trac.openlayers.org/changeset/2237

In the future, I'd like to see patch authors include these tests where
possible.

Is there anything that can be done to make the test-writing process
simpler? Do we need more docs on the test writing stuff, or is this just
something people hadn't been thinking about?

Regards,
-- 
Christopher Schmidt
MetaCarta