stream: read after destroy?

[ronag]

Reading the Readable.read implementation it is a bit unclear to me what happens and what should happen when this method is called after destroy().

Should it be an error (e.g. READ_AFTER_DESTROY) or a noop?

[addaleax]

I think both behaviours are okay, so I’d go with the variant that is less breaking, i.e. making it a noop.

[ronag]

@addaleax: I've done a quick fix for the most problematic parts #29491.

However, there seems to be an assumption that readable streams will still work with buffered data after destroy(), i.e. 'end' (and 'finish' for that matter) is expected to be emitted after destroy() in several tests.

I started on a "proper" fix but it quickly hit several tests #29485.

Thoughts? @mcollina