Hacker News new | ask | show | jobs
by dmitriid 2273 days ago
I simply love it how every single Haskell framework I've seen that has "ease of development" and "it's easy" and similar immediately devolves from this (from Concur's docs [1]):

  hello = do
    button [onClick] [text "Say Hello"]
    text "Hello Sailor!"
into this:

  inputWidget = input [(Changed <<< unsafeTargetValue) <$> onChange, Focused <$ onFocus]
or this:

  inputWidget st = input [st {focusCount = st.focusCount+1} <$ onFocus
                         , ((\s -> st {currentText = s}) <<< unsafeTargetValue) <$> onChange]
for even the slightest modifications (which are not even complex in this case)

  These instances, allow a natural handling of 
  return values using <$>, <*>, <$, monadic 
  do-notation, etc
Right...

[1] https://github.com/ajnsit/concur-documentation/blob/master/R...

3 comments

The second example can be rewritten as such:

    inputWidget = let
        changeAction = do
          v <- onChange
          return (Changed (unsafeTargetValue v))
        focusAction = do
          _ <- onFocus
          return Focused
      in input [changeAction, focusAction]
and likewise for the third:

    inputWidget st = let
        focusAction = do
          _ <- onFocus
          return (st {focusCount = st.focusCount+1})
        changeAction = do
          v <- onChange
          return (st {s = unsafeTargetValue v})
      in input [focusAction, changeAction]
(I didn't compile this, so hopefully there's no major mistakes here)

Believe it or not, many Haskell programmers prefer brevity when it comes to programs like this. They're honestly not that unreadable once you learn the meaning of a few infix operators.

It's not clear to me what you're trying to show? Are you saying the later examples look more complex?
My guess: The latter examples look unfamiliar (looks like Haskell) while the first example looks familiar to the general eye (looks like any PL)

The things being compared don't even have the same purpose so idt comparing them even makes sense /shrug

Of course they do. Of course they are.

The simplest code change turns the code in to a monadic/functor soup.

... and if you program in Haskell that's exactly what you want. Programming with these abstractions is common and well supported in Haskell.
Given what the examples are doing (describing inputs) I don't see how you solve it ergonomically without higher kinded types, which in turn leads to functors etc.
I love it too, because it is a standard and flexible way to program in Haskell!