Automatic message-replay for Elm!

Hello everyone! :sunny:

I’ve spent a significant amount of time designing and implementing elm-devtools, which is very close to release. The value-proposition of that project includes what I’m presenting here, but as the core ideas behind those developer-tools are very simply, I wanted to publish that core first!

Elm AMR

Automatic message-replay is an approach to “hot-reloading” that is designed around the way the Elm architecture works.

You can find more information on how the module works in the Github repo, the package site and in the comment where Richard Feldman first described “automatic message-replay”.

Thanks for taking a look. Let me know whatever you think!

13 Likes

Great work, I’ll try it out! But I have some doubts. 1) Does it take a long time to encode/decode an app with lots of msgs? 2) Every module that has it’s own Msg type need now to export an encoder/decoder? 3) And what about Elm Codec package, would it be easier to write encoders and decoders?

Hey Gabriel - Thank for asking! :sunny:

I have done a performance-test here for devtools, which employs the same strategy as this module! Performance in this module should be even better due to a couple of things:

  1. All messages are pre-encoded (as soon as they update), which makes writes to cache much faster! devtools doesn’t do this for a number of reasons that isn’t relevant to this module.
  2. This module uses something I’ve called a DenseList to ensure every message is only stored once, and then maps those to an array of integers. This makes the cache-write much smaller. I’ve also ensured encoding (on update) and decoding (on init) only does one pass across the values when recalculating the application-state.

It’s really fast IMO. The performance-stuff will only ever be relevant for applications that use RequestAnimationFrame - if this subscription is not running in your application I feel safe guaranteeing no performance problems.

This problem is addressed here. In a sentence, I’ll be building a CLI that can generate the code for you - I plan to do this after the release of devtools.

I haven’t looked into this! Thanks for sharing - I’m making note of this, and will look more into it once I get around to building code-generation for JSON-en/decoders.


Let me know if I something was unclear, or if you have more questions! :sunny:

This topic was automatically closed 10 days after the last reply. New replies are no longer allowed.