Closed pi8027 closed 2 years ago
~In the latter case, it rather seems to fail to recognize that ratr
is a ring morphism :/~ Indeed, F
should be a numFieldType
.
Goal forall (F : fieldType), ratr 2%:~R = 2%:~R :> F.
Proof.
move=> F.
Fail field. (* Not a valid ring equation. *)
I happened to see another performance issue in the example above, which has been fixed in #29. The performance issue of morphism detection is not easy to fix. I attempted to do that by implementing a restricted version of canonical structure inference in Elpi and it does not improve anything.
I confirm that this performance issue has disappeared.