Looking for feedback on an upcoming winnow
API change
#2
Labels
A-parsing
Area: Related to the parsing user input
C-question
Category: Further information is requested
C-upstream
Category: Issue for a dependency, will be fixed upstream
When developing
winnow
, I found that certain behavior had a dramatic performance impact (e.g. usingLocated
). My theory on the root cause was that the slowdown is because of how much bigger the return type of parsers gets. winnow-rs/winnow#268 tries to change the API to allow people to reduce the size of their return types. In that PR, I've made comments on the code that highlight what the impact of this change is.This is a major change going from a pure-functional API to more imperative (while still allowing a mostly declarative use). It allows for a new class of bugs but prevents a different class.
What I'm wanting to understand is what users of
winnow
think of this API, ranging from "its good" to "if the gains are there" to "I'd dropwinnow
.The text was updated successfully, but these errors were encountered: