I completely agree that having this code coverage tools is a great step forward and that we should aim for 80% or more coverage. However, I also agree with Ted that it’s not always effective or efficient to strive for a fixed number. I think that we should start with reporting the numbers and making them part of the review to gain some more experience on what is feasible and attainable in the AsterixDB code before we decide on accept/reject numbers.
My 2c, Till On 20 May 2016, at 18:50, abdullah alamoudi wrote: > Thanks to Mike Blow, we now have reports of our test coverage for the code > base. > Testing is essential in controlling code quality and our current coverage > could use serious improvement. > > Now that we have the tools to get coverage information, I suggest we > establish a minimum required test coverage for all changes. We can start by > enforcing this on new classes while not reducing the coverage for existing > classes. > > I suggest we set the minimum coverage to 80%. What is nice about this is > that this is an objective measure and no one can claim he's being treated > unfairly. > > Code that is not tested is a broken code. Thoughts?
