Sysex access information in MIDIPort?

[qdot]

Due to the fact that a MIDIPort object can outlive a MIDIAccess object, we could have ports that are created, but whether or not they support sysex is opaque unless you actively open and call send on them with a sysex message. It also means that we could fall into issues like ports being created between two MIDIAccess objects with differing sysex permissions, and having a problem telling which is which.

I have no idea exactly how you'd get into that situation, but it could happen with the way I'm translating the spec as it is at the moment. Would it be worth it to push that info into ports, or is that overkill?

[cwilso]

My opinion is that it's probably overkill. Yes, it's possible to get into those two situations (1 - you don't know if the MIDIPort you have has sysex access, and 2 - you have two MIDIAccess objects, one with sysex and one without) - but only intentionally, and in short just "you shouldn't do that if you want to remember what MIDIAccess you're coming from."

[toyoshim]

From a viewpoint of implementation, it's easy to expose it as boolean MIDIPort#sysexEnabled.
So, if it is convenient for web developers, I'm fine with adding it. Also, if the idea to allow sysex permission port by port revives, the attribute will be needed.

[cwilso]

OK.

[mjwilson-google]

TPAC 2024 notes:

Firefox and Chromium are both still positive for this change.