Re: [mediacapture-transform] MediaStreamTrackProcessor threading model is underspecified (#27)

> Can you elaborate on why that is the bare minimum if this is going to ship anywhere?

It is not possible to write a compatible implementation if the respective order of operations is not specified normatively. The links above don't mandate any thread setup, they mandate a specific order in which thing need to appear to the user, using a thread setup to explain it.

Web Codecs uses various threads internally (and even hardware decoders and even other chips, or even networked attached stuff, it is irrelevant), but is specified in a certain way. Web Audio also offloads a bunch of stuff to other threads and even ping-pongs between threads in various scenario, but is specified using only a single thread as you see.

You might think that you can just describe things roughly, but that won't work, this has real-world consequences. Things need to be described in a way that more than one implementation can be written without looking at each other source code, and pass the same tests.

Things matter if they are observable, in which case they need to be specified.

-- 
GitHub Notification of comment by padenot
Please view or discuss this issue at https://github.com/w3c/mediacapture-transform/issues/27#issuecomment-852202849 using your GitHub account


-- 
Sent via github-notify-ml as configured in https://github.com/w3c/github-notify-ml-config

Received on Tuesday, 1 June 2021 15:08:23 UTC