code example in API comment for spread' is broken

[jwaldmann]

https://hackage.haskell.org/package/tidal-1.9.5/docs/Sound-Tidal-UI.html#v:spread-39-

GHCi, version 9.10.1: https://www.haskell.org/ghc/  :? for help
ghci> import Sound.Tidal.Context 
ghci> :set -XOverloadedStrings 
ghci> :t spread' slow "2 4%3" $ sound "ho ho:2 ho:3 hc"
<interactive>:1:14: error: [GHC-39999]
    • No instance for ‘Enumerable (Pattern Time)’
        arising from the literal ‘"2 4%3"’
    • In the second argument of ‘spread'’, namely ‘"2 4%3"’
      In the first argument of ‘($)’, namely ‘spread' slow "2 4%3"’
      In the expression: spread' slow "2 4%3" $ sound "ho ho:2 ho:3 hc"

this version is type-correct (inserting pure)

spread' slow (pure "2 4%3") $ sound "ho ho:2 ho:3 hc"

is that how it's supposed to be used? (I guess not, since implementation then does pure "..." >>= _ where the monadic-bind is useless)

then perhaps this?

spread' slow (fmap pure "2 4%3") $ sound "ho ho:2 ho:3 hc"

NB: to catch documentation bugs like this, it would be good to have a method of evaluating examples (like https://github.com/sol/doctest#readme) (at least, type-checking)

[bgold-cosmos]

It looks like the docs are out of date there. spread' is meant to be used with functions that don't take "patternified" arguments. So

spread' _slow "2 4%3" $ sound "ho ho ho hc"

evaluates correctly. But a lot of Tidal functions were updated so that their "vanilla" version already accepts pattern arguments, which makes spread' mostly redundant in those cases.

[yaxu]

Yes when spread' was written, the signature of slow would be

slow :: Time -> Pattern a -> Pattern a

now of course it's

slow :: Pattern Time -> Pattern a -> Pattern a

and patterns of numbers are numbers so slow 2 still works. The unpatternified function _slow exists but only for internal use to try to be a bit more efficient in cases where the value is never a pattern.

As you say spread' is now redundant. I think we should probably just remove it.