2018-04-30 11:20 GMT+00:00 Jan Matèrne (jhm) <apa...@materne.de>:

> For that single point I think parametrization is 'good'.
> If I count right, it impacted three test classes.
> Why change all the others?
>
> Starting a discussion before such big changes would also be helpful.
>

Just to pick a nit, there are 9 parameterized test classes.
I could easily add another 10-20 classes.
But, I'd rather put trivial changes into a separate commit
(I remember being told off for mixing in whitespace changes on previous
occasions...)

Anyway, I also recall that we had a discussion about unit tests which ended
up in decision on one detail:
any setup details (eg special properties) that can break tests and are
controlled by end user must be asserted rather than assumed.
The latter should be used for OS- and/or runtime-specific properties.

Gintas

Reply via email to