Double Checking the Regression Testing Suite

Ian Monroe ian.monroe at gmail.com
Mon Mar 23 19:24:16 UTC 2009


2009/3/23 N Randolph <yoshikira at gmail.com>:
> It's me again. This will be the last e-mail for a good long while. Due to
> lacking creativity, I am going to go down the Regression Test Suite avenue.
> Just to make sure I don't propose something unrealistic:
> Would a person be able to create test code for every 'component' (I'll
> define that later) of amaroK and then create a program that would easily run
> any combination of the test code within the summer? Oh, and also implement
> somewhere in there (some research is needed here) a means of comparing test
> results. The main question is if this could be done for all of amaroK in a
> summer...... roughly.....
> As for the size of the individual test 'components', how detailed could the
> test suite _reasonably_ get? For example, the smallest component would be
> one test code that encompasses anything under/related_to "using external
> media devices", or perhaps the smallest test component would be just testing
> a class that is used in working with media devices..... or somewhere in
> between. This is referring to the idea of being able to choose what you
> test.
> I hope the questions aren't too confusing. These are just fine, nit-picky
> stuff.
> -Nelson

Hey Nelson,
I think most tests would be at the class level. Maybe some at the system level.

Qt has a testing app for models that would probably be nice to
incorporate as well.

Anyways mostly the task is putting in the infrastructure, and yea,
making a substional amount of tests. We wouldn't expect you to make
them all though.

Ian



More information about the Amarok mailing list