https://kotlinlang.org logo
Channels
Powered by
#arrow-contributors
  • e

    Eugenio

    07/21/2018, 8:29 PM
    isn't the instance always the "leaf", where parameters are fixed?
  • p

    pakoito

    07/21/2018, 8:29 PM
    we should cache the objects for those without parameter for sure, there’s a ticket somewhere
  • p

    pakoito

    07/21/2018, 8:29 PM
    no, for example many Semigroups are transitive
  • p

    pakoito

    07/21/2018, 8:30 PM
    because they depend on aggregating internal values
  • p

    pakoito

    07/21/2018, 8:30 PM
    (if you’re just interested in aggregate by container, there’s SemigroupK)
  • e

    Eugenio

    07/21/2018, 8:31 PM
    hmmmm example?
  • p

    pakoito

    07/21/2018, 8:31 PM
  • p

    pakoito

    07/21/2018, 8:32 PM
  • p

    pakoito

    07/21/2018, 8:32 PM
  • e

    Eugenio

    07/21/2018, 8:33 PM
    oh I see what you mean
  • e

    Eugenio

    07/21/2018, 8:33 PM
    then I'll just do a magic trick
  • e

    Eugenio

    07/21/2018, 8:33 PM
    (coff coff type erasure coff coff)
  • e

    Eugenio

    07/21/2018, 8:33 PM
    if only there was a way to access the 
    DefaultImpl
    of interfaces from Kotlin...
  • p

    pakoito

    07/21/2018, 8:34 PM
    hahahah what’s going on that evil mind of yours
  • e

    Eugenio

    07/21/2018, 8:35 PM
    oh you'll see
  • e

    Eugenio

    07/21/2018, 8:35 PM
    evil laugh
  • e

    Eugenio

    07/21/2018, 8:36 PM
    essentially I'm trying to remove the need to have 
    extension
    when calling methods coming from type classes
  • e

    Eugenio

    07/21/2018, 8:36 PM
    and instead have them directly accessible on the normal instances
  • e

    Eugenio

    07/21/2018, 8:37 PM
    oh and also have a standardised way to access a type class that doesn't force the extended type to have a companion object (I'm looking at you, 
    Boolean
    😒)
  • p

    pakoito

    07/21/2018, 8:39 PM
    neat! If you could calculate all typeclasses at compile time in a transitive way all the way down, we’ve basically fixed typeclasses because we can just write a dictionary with O(1) access
  • e

    Eugenio

    07/21/2018, 8:39 PM
    that
  • e

    Eugenio

    07/21/2018, 8:39 PM
    but as extensions
  • e

    Eugenio

    07/21/2018, 8:39 PM
    since kotlin is kind enough to allow us to disambiguate statically on the generic types ❤️
  • p

    pakoito

    07/21/2018, 8:40 PM
    is there any way of scoping it for replacement? it’s not a requirement really, as even the KEEP87 specifies single implementation per typeclass per datatype
  • p

    pakoito

    07/21/2018, 8:41 PM
    okay, I’ll wait for the solution first lol
  • e

    Eugenio

    07/21/2018, 8:41 PM
    yes, you just import one from a different package
  • e

    Eugenio

    07/21/2018, 8:41 PM
    or access it directly and create a scope for it, like any other extension
  • p

    pakoito

    07/21/2018, 8:41 PM
    👍
  • e

    Eugenio

    07/21/2018, 8:42 PM
    also, while you're here, which syntax do you hate the most like more for getting the instance of the type class manually?
    Copy code
    +Eq<Boolean>()
-Eq<Boolean>()
!Eq<Boolean>()
Eq..Boolean::class
Eq..of<Boolean>()
  • e

    Eugenio

    07/21/2018, 8:45 PM
    lol now that I think about it, I can even do
    Copy code
    Eq + Boolean::class
1...444546...115Latest