jgraham

"People take things other than standards positions as endorsement" has certainly been a real problem historically. If Blink in particular has improved its process here so that this doesn't happen...

Requiring an exact size means that non-reftests have the same problem on mobile that reftests have i.e. you either can't run them or have to run in some very nonstandard...

I think at least some of those are things where the spec _does_ define the testing API. But I agree that tests which depend on undefined browser internal APIs aren't...

This feels extremely sketchy to me in terms of the incentive structure it creates. In particular having entire tests show up as FAIL if one subtest is failing feels misleading,...

A concern I have here is that the original motivation is specifically "we want to track a target at the test level", and I worry that's going to cause problems....

> We do want to track our implementation progress against a set of tests over a long period of time, but that doesn’t mean there’s no room for nuance in...

> Tackle this another time. Yes, sorry I didn't mean to imply that this issue should block on that. > Well, recently also Ladybird It looks like they're literally just...

Right, it doesn't really know how to find different binaries for different Firefox channels, and I don't really know how we'd solve that. We could make `WARNING Supplied channel doesn't...

@dependabot rebase

@dependabot rebase