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

    raulraja

    06/09/2019, 7:03 PM
    @pakoito we discussed with Samuel that we may be able to keep Try if we just change it’s invoke to be 
    suspend
  • r

    raulraja

    06/09/2019, 7:03 PM
    that way people can still use Try like they do today when they are inside 
    suspend
    if they want to since that still requires a runner or IO and is still pure.
  • p

    pakoito

    06/09/2019, 7:05 PM
    is it worth it?
  • p

    pakoito

    06/09/2019, 7:05 PM
    I mean, we need to redo Try to support suspend both in creating and running it
  • r

    raulraja

    06/09/2019, 7:05 PM
    nope
  • r

    raulraja

    06/09/2019, 7:05 PM
    it’s not the lambda
  • r

    raulraja

    06/09/2019, 7:05 PM
    it’s the actual function
  • p

    pakoito

    06/09/2019, 7:06 PM
    and what we’ve seen the most demand of is choosing your own error type
  • r

    raulraja

    06/09/2019, 7:06 PM
    suspend fun invoke
    requires no changes to Try and makes it pure
  • r

    raulraja

    06/09/2019, 7:07 PM
    I prefer to remove it but many people in Android use it inside 
    suspend
    which they start on presenters and then they use Try in between to avoid exceptions
  • r

    raulraja

    06/09/2019, 7:07 PM
    that is still pure
  • r

    raulraja

    06/09/2019, 7:07 PM
    but not with the current signatures of 
    invoke
  • p

    pakoito

    06/09/2019, 7:07 PM
    we can still do that on Either and kill Try
  • p

    pakoito

    06/09/2019, 7:07 PM
    it’s redundant
  • r

    raulraja

    06/09/2019, 7:07 PM
    the issue with that is inference
  • r

    raulraja

    06/09/2019, 7:08 PM
    Either is partially applied so accessing it’s monad instances requires partial application if they use the companion unless we build in specialized methods for the 
    Try
    versions fixed to Throwable
  • r

    raulraja

    06/09/2019, 7:08 PM
    IMO Either is fine as is and Try is a lesser evil if suspended and we should discourage people from using it until it dies on its own
  • r

    raulraja

    06/09/2019, 7:09 PM
    because really they should not use Either either
  • r

    raulraja

    06/09/2019, 7:09 PM
    They should use Bifunctor IO when that is available for stuff that throws exceptions
  • p

    pakoito

    06/09/2019, 7:09 PM
    unless we build in specialized methods for the 
    Try
    versions fixed to Throwable
    that’s my point. There are only a couple of constructors, and only one would be problematic
  • p

    pakoito

    06/09/2019, 7:09 PM
    just infers Throwable just fine, as the others do. catch can be specialized for Throwable like on the PR
  • p

    pakoito

    06/09/2019, 7:10 PM
    and that’s about it
  • r

    raulraja

    06/09/2019, 7:10 PM
    I mean instances etc. 
    Either.monad<Throwable>()
    is not the same as 
    Try.monad()
  • r

    raulraja

    06/09/2019, 7:10 PM
    in terms of syntax
  • p

    pakoito

    06/09/2019, 7:10 PM
    in what way
  • r

    raulraja

    06/09/2019, 7:10 PM
    not semantics
  • r

    raulraja

    06/09/2019, 7:11 PM
    unless we provide 
    Either.monadThrow()
  • p

    pakoito

    06/09/2019, 7:11 PM
    Either.monadThrow()
  • p

    pakoito

    06/09/2019, 7:11 PM
    which we do
  • p

    pakoito

    06/09/2019, 7:11 PM
    same ideas lol
1...969798...115Latest