On Jan 31, 2013, at 11:59 AM, Summers Pittman <supittma@redhat.com> wrote:

On 01/31/2013 12:12 PM, Kris Borchers wrote:
I want to throw this out to the list for feedback. Something we have been doing for a while now with the jQuery project is to require a unit test(s) for any PR or change committed. This has worked very well in two ways. First, it provides a built in way to see the issue being fixed/implemented within the PR. That way, the reviewer doesn't have to build their own test(s) to see if the issue being addressed is actually fixed. Second, it helps prevent regressions down the road since more of the code is covered by tests so you know if some change you think is unrelated breaks something that fixed days, weeks, years ago.

I would like to suggest we make this policy for at least the JS library (since that is the one I review most often) but I believe this policy would be useful across the entire project. Let me know what you think.
On the one hand I really REALLY appreciate good testing.  If your code is easy to write tests for it is  (in theory) easy to use.  Otoh I REALLY hate writing tests for obvious implementations/impossible scenarios just to get coverage.
I agree. We should not write tests just for coverage but instead to test actual bug fixes (not simple typos or minor things where the solution was obvious) or new features.

(ex In Java land obvious is getter/setters and an impossible scenario would be catching UnsupportedEncodingException for String.getBytes("UTF-8").)
Thanks,
Kris
_______________________________________________
aerogear-dev mailing list
aerogear-dev@lists.jboss.org
https://lists.jboss.org/mailman/listinfo/aerogear-dev

_______________________________________________
aerogear-dev mailing list
aerogear-dev@lists.jboss.org
https://lists.jboss.org/mailman/listinfo/aerogear-dev