Closed conal closed 8 years ago
Would it be more valuable to match partial applications of datacons? We could rewrite callDataConT to do that.
I don't know. I first assumed that it worked for non-saturated as well, and so I added code to exclude those cases (for my use), only to find that callDataConT
already did so.
But then I also handle non-saturated data constructor applications by eta-expanding and re-trying my transformation on the new lambda body (recursively).
as does the underlying
exprIsConApp_maybe
from GHC.