newlandsvalley / elm-webmidi-ports Goto Github PK

I see the model only includes "inputDevices" so I suspect this may not be the case.

I can probably write this, but I'm opening this to track it and hopefully get some input on how to implement this.

Input and output devices are already tracked and exposed to Elm, but right now you can't do anything with them.

For input:
I think the most straightforward thing would be to expose the ID of the midi device in the EncodedEvent and Event messages. That way clients can easily ignore it if they don't care, but they also have all the information if they want to use it. There might be a race if we receive a MIDI message followed by the device immediately disconnecting it's possible when a client sees the decoded event the attached device ID doesn't match up to any device. I don't think this is much of a problem though.

For output:
I think the simplest thing here would be to have two different send functions. One which sends to everything (like the existing one) and another which accepts a list of device ids.

Thoughts?

When trying to play the piano and midiInstrument examples I get errors that SoundFont.Ports cannot be found. I'm assuming this refers to your other project https://github.com/newlandsvalley/elm-soundfont-ports . However, I'm not sure how I'm supposed to install this so these examples work.

Can you add something to the existing documentation to cover this?

I haven't looked into how to test Elm with ports, but I don't suspect this will be too hard since the ports already make a good separation between this library and the web MIDI JS.

Right now most of the Elm files are formatted with elm-format. Would you accept a PR to reform the Elm files?

I like autoformatters because they give me one less thing to think about when I'm writing code. However, they don't work well if everyone isn't on board as they'll otherwise clutter up unrelated changes and make the history harder to follow.

I'm going to get to work on this.

For the first version I'm thinking something similar to how inputs are currently handled where we don't differentiate between devices so any output will be sent to every device.

This would just ensure the main package and the examples continue to build. If tests are eventually added this would have even more benefit.

Travis CI seems to be the popular option since it's well integrated and free for open source stuff. I'll eventually get around to looking at it.