gurry commented on issue #6858:
I wonder what tests we can write for this since any faulty input will be caught much before we reach the egraph pass by the earlier verifier invocations.
afonso360 commented on issue #6858:
Huh, I'm not sure how we would test this. Maybe @jameysharp has an idea?
jameysharp commented on issue #6858:
Yeah, I don't think we can write sensible tests for running the verifier after specific passes. It wouldn't be too hard to add an invalid egraph rule that only matches when some flag is set, and only turn that flag on in a test, but I don't think it's worth doing that.
Mostly I think we'll notice whether this is working whenever we write new incorrect egraph rules, which I'm sure will happen often enough. :sweat_smile:
gurry commented on issue #6858:
That makes sense @jameysharp. Thanks.
Last updated: Dec 23 2024 at 12:05 UTC