chore: do not perform type analysis in tests #7852
Conversation
There was a problem hiding this comment.
I am in favour of this change, since I find it superfluous to make such assertions, but I think we need to wait for @keradus' opinion. I can only guess what was the reasoning and I think it's more about ensuring that our internal classes provide interface instead of being concrete-only classes. Yeah, most probably removing implements X would lead to other failures or inability to test things properly, but I guess that these are "architectural" assertions, and I won't merge this change on my own.
There's always an option to introduce tools like Arkitect or other, to ensure such design constraints are matched.
| 'weird' | ||
| ); | ||
|
|
||
| self::assertInstanceOf(\InvalidArgumentException::class, $exception); |
There was a problem hiding this comment.
no clue why we have tests like those.
I believe they cause no harm, but also no huge benefit behind them.
IE, we rely that result of function foo(): Date is Date without making utest for result type
keradus
changed the title
|
Thanks @kubawerlos 🍻! |
Related to the currently failing CI on the main branch.