Add printer combinators dual to parser combinators #71
Conversation
# `Divisible` Combinators * `(>*)` * `(*<)` # `Decidable` Combinators * `optionalD` * `manyD` * `many1D` * `sepByD` * `sepBy1D`
`(>*<)`, `(>*)`, and `(*<)` are `Divisible` operators, not requiring a `Decidable` constraint
| manyD p = choose (mayhaps . uncons) conquered (many1D p) | ||
|
|
||
| -- | One or more. | ||
| many1D :: Decidable f => f a -> f (a,[a]) |
There was a problem hiding this comment.
It's possibly more ergonomic to work with a NonEmpty a than an (a, [a])?
There was a problem hiding this comment.
Maybe, but I don't really think NonEmpty is much better (and it has an evil IsList instance) than the pair and this matches better with divide. I could be convinced otherwise.
There was a problem hiding this comment.
I don't know either; just wanted to flag it.
There was a problem hiding this comment.
If we used NonEmpty a in many1D then you'd have to pay the cost of conjugating by the isomorphism NonEmpty a <-> (a,[a]) in manyD (at least if we do the naive thing).
There was a problem hiding this comment.
I do like NonEmpty here, even if there is a little bit of overhead.
There was a problem hiding this comment.
Ok, I added a commit to use NonEmpty. I think the cost is n-1 calls conjugating by the isomorphism for a list of length n. Happy to make any other changes to fixity or naming.
There was a problem hiding this comment.
Or I could just write a helper function and not be naive. Ok, now the cost is just 1 call conjugating by the isomorphism.
| (*<) :: Divisible f => f a -> f () -> f a | ||
| p *< after = (,()) >$< p >*< after | ||
|
|
||
| infixr 5 *< |
There was a problem hiding this comment.
The @gwils talk uses infixr 4. Not saying that one is correct, but is there a reason?
There was a problem hiding this comment.
I haven't seen this talk yet. My reasoning was by analogy. The Applicative operators are all infixl 4.
infixl 4 <*>
infixl 4 <*
infixl 4 *>The Divisible operator so far is infixr 5 >*< so I gave >* and *< the same fixity. I'm not sure that's "correct".
There was a problem hiding this comment.
Looks like I switched the signatures of *< and >* from that talk too, oops.
simplify definitions for `(>*)` and `(*<)`
I think |
I also think that Tom Ellis's |
DivisibleCombinators(>*)(*<)DecidableCombinatorsoptionalDmanyDmany1DsepByDsepBy1D