Zulip Chat Archive

Stream: lean4

Topic: tracing typeclasses?


view this post on Zulip Jason Gross (Mar 12 2021 at 19:57):

Google suggests set_option trace.class_instances true, but that's not a thing in Lean4. What's the equivalent?

view this post on Zulip Leonardo de Moura (Mar 12 2021 at 20:32):

We have set_option trace.Meta.synthInstance true but it is too verbose. As part of https://github.com/leanprover/lean4/issues/337, we want better error messages that explain why some typeclass resolution failed.

view this post on Zulip Jason Gross (Mar 12 2021 at 20:40):

Thanks! That (plus set_option pp.all true) helped me figure out why my coercion wasn't working: I had picked the wrong universe for the output sort when looking for CoeTC _ (Sort _). (Though I guess here in general I should be using coeSort, not coe.)


Last updated: May 07 2021 at 13:21 UTC