Re: [whatwg/encoding] Rename Encoding's "streams" to "token queues" (#215)

> I think rewriting everything in terms of lists would probably add clarity. I guess the one change we could make is that EOF should be explicitly be pushed into the list so we can "wait" for the list to change when it's empty.

@annevk So if I'm understanding this right, reading would mean waiting until there is at least one item in the list; and if that item is EOF, return an EOF without popping that one off the list.

Also, we should probably add a "new consumable list" operation which populates it with the EOF, and a note for users of the spec and implementers on how to create IO-backed consumable lists.

> 1. I think per our IRC conversation it's okay for this to be fully blocking and synchronous. If you need things to be non-blocking and asynchronous, use "in parallel".

Come to think about it, the encoding APIs can't be made async for web compat, which precludes running them in parallel, and aren't allowed to block because they must be callable from window agents. The way the spec text is right now, the blocking is hidden behind the scenes, but the APIs don't run into this issue because none of the lists used in those algorithms are IO-backed. So we should not make "read" always potentially blocking, but instead have it depend on whether the list is IO-backed -- or alternative, on whether the list was created with an EOF.

-- 
You are receiving this because you are subscribed to this thread.
Reply to this email directly or view it on GitHub:
https://github.com/whatwg/encoding/pull/215#issuecomment-658375479