[Haskell-cafe] Parallel interruptible computations

[Chris Wong]

> "Event" is a small DSL that I interpret in the back end.
> It allows me to write nice expressions about events. Say you have functions
> to create text fields and buttons on the GUI:
>
> -- Create an event binded to a text field, with the first argument as a
> title.
> -- Once validated, the event returns the content of the text field.
> inputText :: String -> Event String
>
> -- Create an event binded to a button, with the first argument as a title.
> inputButton :: String -> Event ()
>
>
> You could then express nice combinations:
>
> -- using Applicative: create a form with two fields
> data NameSurname = NameSurname String String
> form1 :: Event NameSurname
> form1 = NameSurname <$> onInputText "Name:" <*> onInputText "Surname:"
>
> -- using Alternative: create two buttons, first button clicked returns
> False, the second True
> form2 :: Event Boolean
> form2 = True <$ inputButton "click here for True" <|> False <$ inputButton
> "click here for False"
>
> -- using Monad: create the two buttons of form2, if "True" button is
> clicked, then a text field appears asking for a name.
> form3 :: Event String
> form3 = do
>    myBool <- form2
>    if myBool then onInputText "Name:"
>    else return "No name"

Oh, I see now. So we need to reflect on the structure of the
computation, to figure out what widgets to show. That makes sense.

Chris

> But I am lacking a way to express the situation where I have a bunch of
> events, which can be cancelled as soon as a result can be calculated:
> ShortcutEvents :: [Event a] -> ([Maybe a] -> Maybe b) -> Event b
>
> You can think of it as a generalization of the "or" shortcut, where the
> evaluation is cut short if the first argument evaluates to True.
>
>
> With your type:
> type Event a = Maybe (BaseEvent a)
> Could you express the combinations above??
>
>