Update signal graph nodes in topological order.

[shashi]

Currently every event requires visiting every node in the signal graph. This is an experiment to use topological order in the signal graph to only visit the nodes that need to update in the right order. This approach also makes for clearer signal propagation code (without count++ in recv etc.). This is O(n*log(n)) in the worst case, compared to O(n) but I expect this will be faster in most cases with multiple Input nodes. I tried it out on most examples in on elm-lang.org, it works. I couldn't think of a reason why this would behave any differently.

This was suggested by @stevengj on JuliaGizmos/Reactive.jl#4

[johnpmayer]

This looks pretty intrusive. Would you mind introducing the idea on the
mailing list?
On Jun 12, 2014 12:50 PM, "Shashi Gowda" [email protected] wrote:

Currently every event requires visiting every node in the signal graph.
This is an experiment to use topological order in the signal graph to only
visit the nodes that need to update in the right order. This approach also
makes for clearer signal propagation code (without count++ in recv etc.).

This was suggested by @stevengj https://github.com/stevengj on

JuliaGizmos/Reactive.jl#4 JuliaGizmos/Reactive.jl#4

You can merge this Pull Request by running

git pull https://github.com/shashi/Elm topo_sort

Or view, comment on, or merge it at:

#658
Commit Summary

• Update signal graph nodes in topological order.

File Changes

• M libraries/Native/Show.js
  https://github.com/elm-lang/Elm/pull/658/files#diff-0 (2)
• M libraries/Native/Signal/Keyboard.js
  https://github.com/elm-lang/Elm/pull/658/files#diff-1 (49)
• M libraries/Native/Signal/Signal.js
  https://github.com/elm-lang/Elm/pull/658/files#diff-2 (95)
• M runtime/Init.js
  https://github.com/elm-lang/Elm/pull/658/files#diff-3 (57)
• M runtime/Utils.js
  https://github.com/elm-lang/Elm/pull/658/files#diff-4 (119)

Patch Links:

• https://github.com/elm-lang/Elm/pull/658.patch
• https://github.com/elm-lang/Elm/pull/658.diff

—
Reply to this email directly or view it on GitHub
#658.

[shashi]

Sure! 👍 Done. https://groups.google.com/forum/#!topic/elm-discuss/YcvPwhcq_sU

[evancz]

This is a cool idea, but I am going to close for now. I think we need to show benchmarks for changes that are for signal performance and I want to keep an architecture that is without glitches.

[shashi]

👍 I will get around to doing the benchmarks soon.