Should GHC.Parser.Lexer.P always be eta-expanded?
The PState
parameter of the P
monad seems like another case for the one-shot monad trick.
Although I find surprisingly few bindings that need eta-expansion in the Core GHC.Parser.Lexer
.
Hmm, maybe we shouldn't eta-expand PState
, because there might be some legit Alex Actions that profit from sharing? It's hard to tell, let's try and measure.