[wdspec] add fixture to wait for several WebDriver BiDi events about wpt HOT 8 OPEN

I see, the proposed helper API sounds good to me.

from wpt.

API suggestions

With the idea of a single fixture, we can pass the expected events as a list (because the order will matter).

on_events =  wait_for_events([expectation1, expectation2, ...])

Individual items could either be a dict:

on_events =  wait_for_events([
  { "event": "browsingContext.contextCreated", "predicate": lambda e: is_event_valid(e) },
  { "event": "network.responseCompleted", "count": 5 },
  { "event": "browsingContext.load" },
])

Or a list (slightly more compact):

on_events =  wait_for_events([
  ["browsingContext.contextCreated", 1, lambda e: is_event_valid(e)],
  ["network.responseCompleted", 5],
  ["browsingContext.load", 1], # we could omit the "count" if it's 1 and there is no lambda?
])

Either option sounds fine to me, slight preference for the first one.

Return value

About the return value, the snippets above assume each "expectation" returns its own array of events.

on_events =  wait_for_events([
  ["browsingContext.contextCreated", 1, lambda e: is_event_valid(e)],
  ["network.responseCompleted", 5],
  ["browsingContext.load", 1],
])

# do something

[context_created_events, response_completed_events, load_events] = await on_events

Other examples

Some other examples if we want to use this for simpler scenarios.

Wait for N events:

on_events =  wait_for_events([
  ["network.responseCompleted", 5],
])

# do something

[response_completed_events] = await on_events

Wait for 1 event

on_events =  wait_for_events([
  ["network.responseCompleted", 1],
])

# do something

[[response_completed_event]] = await on_events

(although we will probably keep wait_for_event around anyway)

Checking no event emitted

Currently we are also usually using an array of events when we want to check that no event is emitted. I would suggest to replace those with a simpler:

    on_event = wait_for_event(CONTEXT_CREATED_EVENT)
    # do something
    wait = AsyncPoll(bidi_session, timeout=0.5)
    with pytest.raises(TimeoutException):
        await wait_for_future_safe(on_event)

I would not try to make that part of the new fixture, seems like it would be hard to follow.

from wpt.

With the idea of a single fixture, we can pass the expected events as a list (because the order will matter).

By the order will matter, you mean that listeners will be consumed in order or that 5 responseCompleted events should come before the single load event?

from wpt.

With the idea of a single fixture, we can pass the expected events as a list (because the order will matter).

By the order will matter, you mean that listeners will be consumed in order or that 5 responseCompleted events should come before the single load event?

The latter. I thought it would be useful to use this to enforce a strict sequence of events, and also I thought that's what the Node API did.

from wpt.

I think Node API only enforces that listeners are processed in order in which they are created. I think being more strict is for the better but I wonder how can one express if a certain set of events needs to arrive at any (relative) order with this proposal?

from wpt.

I think Node API only enforces that listeners are processed in order in which they are created.

Ah, looks like I misunderstood the other API.

I think being more strict is for the better but I wonder how can one express if a certain set of events needs to arrive at any (relative) order with this proposal?

One way would be to call the fixture several times, eg

on_responsecompleted_events =  wait_for_events([["network.responseCompleted", 5]])
on_load_events =  wait_for_events([["browsingContext.load", 2]])

Or we could make this "strict" behavior optional, and require to pass a flag when calling the fixture to enable it?

from wpt.

What is the behavior if I listen for the same event?:

on_responsecompleted_events =  wait_for_events([["network.responseCompleted", 1]])
on_responsecompleted_events2 =  wait_for_events([["network.responseCompleted", 2]])

and say only two event are actually arriving?

from wpt.

What is the behavior if I listen for the same event?:

on_responsecompleted_events =  wait_for_events([["network.responseCompleted", 1]])
on_responsecompleted_events2 =  wait_for_events([["network.responseCompleted", 2]])

and say only two event are actually arriving?

With what I had in mind: for 2 distinct calls both would be resolved if 2 events arrive.

from wpt.