@0a734d30 @c5389096 First, we have rewritten our pytest support, so unless you're in the lucky 25% (for now) in the experiment, you should be getting it shortly.
Second, we don't have a specific way to override test discovery or execution to the point of swapping out an entirely different command, but you can probably fool it all by making `pytest` hide what you want to do.
Third, if you just want a pass/fail sort of thing, Christopher is right that tasks may work for you.
@f29919fc @c5389096 I generally dislike pytest and and try to avoid it for my own projects, so I want something that doesn’t depend on it. I’d love an all-singing all-dancing fully immersive interactive debugging experience but I’ll happily settle for keyboard shortcuts for the 3 basic functions of “run contextually relevant tests” and “next/previous error from most recent test run”
@0a734d30 @f29919fc @c5389096 what is your reason for disliking pytest - genuinely curious
@6730d36f @f29919fc @c5389096 Too much magic at a distance. Practically, magic parameter name matching as a modularity mechanism will never not be extremely jarring to me. Aesthetically, byte-code rewriting to change the semantics of a little-used piece of dedicated syntax seems bananas. It’s been a problem exactly twice in my entire career so it’s not like I am tripping over bytecode modification bugs all the time but it seems like a wildly unnecessary risk even if it’s vanishingly small.
@0a734d30 I can completely understand your position. I’m currently trying to demystify some of the magic in a project I inherited
@6730d36f I used to really advocate against it, but at this point I am resigned to just being a weirdo on this point. It’s fine, the abstractions work okay once you’ve gotten used to them, but I will probably not choose it voluntarily unless I specifically know that colleagues in this context specifically want to use it
@0a734d30 @6730d36f I’m a weirdo with you @0a734d30. I 110% agree about the magic action at a distance and the name matching. But what alternative do you use?
@4c2fcda1 @6730d36f Usually, Twisted's `trial` test runner, but this advice may not generalize :)
(Seriously though come maintain trial with me we could make it so much better)