Time.since relies on removed Signal.count

[jwmerrill]

https://github.com/elm-lang/core/blob/93bcbcde7a72581d01f0d40e8881b02e530d3c2a/src/Native/Time.js#L52

[jwmerrill]

Here's my crack at a pure Elm implementation without count, but I haven't been able to test this yet because of some other issues getting running in the new world:

since : Time -> Signal a -> Signal Bool
since t s =
  let
    start = map (always 1) s
    stop = map (always -1) (delay t s)
    delaydiff = foldp (+) 0 (merge start stop)
  in
    map ((/=) 0) delaydiff

• Edit, changed (==) to (/=) to get parity correct.

[jwmerrill]

Here's my crack at a pure Elm implementation without count, but I haven't been able to test this yet because of some other issues getting running in the new world:

since : Time -> Signal a -> Signal Bool
since t s =
  let
    start = map (always 1) s
    stop = map (always -1) (delay t s)
    delaydiff = foldp (+) 0 (merge start stop)
  in
    map ((/=) 0) delaydiff

• Edit, changed (==) to (/=) to get parity correct.

[evancz]

This fix looks good to me. Can you confirm @jcollard?

[evancz]

This fix looks good to me. Can you confirm @jcollard?

[jwmerrill]

I made some comments on why I don't yet understand how to test this at the end of this e-mail thread: https://groups.google.com/d/msg/elm-discuss/Bs4nsLVZAXM/1drQHhqYccMJ

[jwmerrill]

I made some comments on why I don't yet understand how to test this at the end of this e-mail thread: https://groups.google.com/d/msg/elm-discuss/Bs4nsLVZAXM/1drQHhqYccMJ

[jcollard]

This implementation is close. At the end you should be checking (/=). I think the following version is more readable:

since : Time -> Signal a -> Signal Bool
since t s =
  let
    start = Signal.map (always True) s
    stop = Signal.map (always False) (delay t s)
  in Signal.foldp (\ n _ -> n) False (Signal.merge start stop)

Also, I'll make a reply to your message on how I go about testing changes so everyone can see.

[jcollard]

This implementation is close. At the end you should be checking (/=). I think the following version is more readable:

since : Time -> Signal a -> Signal Bool
since t s =
  let
    start = Signal.map (always True) s
    stop = Signal.map (always False) (delay t s)
  in Signal.foldp (\ n _ -> n) False (Signal.merge start stop)

Also, I'll make a reply to your message on how I go about testing changes so everyone can see.

[jcollard]

Even more readable:

since : Time -> Signal a -> Signal Bool
since t s =
  let initial = Signal.constant False
      start = Signal.map (always True) s
      stop = Signal.map (always False) (delay t s)
  in Signal.mergeMany [initial, start, stop]

[jcollard]

Even more readable:

since : Time -> Signal a -> Signal Bool
since t s =
  let initial = Signal.constant False
      start = Signal.map (always True) s
      stop = Signal.map (always False) (delay t s)
  in Signal.mergeMany [initial, start, stop]

[jwmerrill]

You're right about the parity of mine being reversed. I'm going to go ahead and edit it to be correct.

Your version doesn't have exactly the same behavior as the existing since. For example, if I ask for since second Mouse.clicks and then click my mouse at half-second intervals, your version will alternate between True and False, but the existing version will never become True until I stop clicking at half second intervals.

Here's an Elm 0.13 program that compares the 3 implementations:

import Mouse
import Signal

joesince t s =
  let initial = Signal.constant False
      start = Signal.lift (always True) s
      stop = Signal.lift (always False) (delay t s)
  in Signal.merges [initial, start, stop]

jasonsince t s = 
  let
    start = Signal.lift (always 1) s
    stop = Signal.lift (always -1) (delay t s)
    delaydiff = foldp (+) 0 (merge start stop)
  in
    lift ((/=) 0) delaydiff

main = asText <~ Signal.lift3 (,,)
    (since second Mouse.clicks)
    (joesince second Mouse.clicks)
    (jasonsince second Mouse.clicks)

Try it out, and see what happens when you click your mouse regularly with a period of less than one second.

[jwmerrill]

You're right about the parity of mine being reversed. I'm going to go ahead and edit it to be correct.

Your version doesn't have exactly the same behavior as the existing since. For example, if I ask for since second Mouse.clicks and then click my mouse at half-second intervals, your version will alternate between True and False, but the existing version will never become True until I stop clicking at half second intervals.

Here's an Elm 0.13 program that compares the 3 implementations:

import Mouse
import Signal

joesince t s =
  let initial = Signal.constant False
      start = Signal.lift (always True) s
      stop = Signal.lift (always False) (delay t s)
  in Signal.merges [initial, start, stop]

jasonsince t s = 
  let
    start = Signal.lift (always 1) s
    stop = Signal.lift (always -1) (delay t s)
    delaydiff = foldp (+) 0 (merge start stop)
  in
    lift ((/=) 0) delaydiff

main = asText <~ Signal.lift3 (,,)
    (since second Mouse.clicks)
    (joesince second Mouse.clicks)
    (jasonsince second Mouse.clicks)

Try it out, and see what happens when you click your mouse regularly with a period of less than one second.

[jcollard]

You're correct. Thank you for pointing it out.

[jcollard]

You're correct. Thank you for pointing it out.

[evancz]

Thanks guys! :D

[evancz]

Thanks guys! :D