Concrete#, SpecialPred, and CtEvidence invariants

Alexis King lexi.lambda at
Thu Dec 9 18:33:20 UTC 2021

On Wed, Dec 8, 2021 at 2:20 PM Richard Eisenberg <lists at> wrote:

> Does this help?

Yes, it does, and what you said makes sense.

Really, the intent of the invariant isn’t “zonkedness” *per se*, but merely
that the type of the evidence should be “at least as zonked” as ctev_pred
is. But that is a little vague and handwavy, so I’ll go the route you
suggest and just reword things a little to explain.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <>

More information about the ghc-devs mailing list