refactor(notifier)!: no initial state for makePublishKit

[turadg]

Extracted from #5704

Description

All publishing kits start empty: #5621

So makePublishKit shouldn't take an initial value and there's no need to have a separate makeEmptyPublishKit.

Security Considerations

--

Documentation Considerations

--

Testing Considerations

Updated

[erights]

LGTM, thanks!

[gibson042]

This test is assuming too much about the implementation details of makePublishKit rather than its external interface. We should just drop the subscribeAfter argument and publishCount assertion.

Suggested change

	const subAll = await Promise.all([
	subscriber.subscribeAfter(),
	subscriber.subscribeAfter(expectedInitialCount - 1n),
	subscriber.subscribeAfter(0n),
	]);
	const [subFirst] = subAll;
	t.like(subFirst, {
	head: { done: false },
	publishCount: expectedInitialCount,
	publishCount: 1n,
	});
	const subAll = await Promise.all([
	subscriber.subscribeAfter(),
	subscriber.subscribeAfter(),
	]);
	const [subFirst] = subAll;
	t.like(subFirst, {
	head: { done: false },
	});

[turadg]

This test is assuming too much about the implementation details

You mean the publicCount: 1n part? (The 0n argument is legit.)

I agree the interface doesn't guarantee that the next value will be one greater. Because there is a publish() it does guarantee that it will be greater though, right?

When I read this test in master I thought it was about the counts after publish but the first test in the file checks those. This seems to really be testing publishing particular values (of the initialValues) loop. Since there is no "initial value" anymore, and I trust there are already tests for .publish(), I'll just remove this test.

@gibson042 PTAAL

[turadg]

Looking at the CI output I saw a number of floating promise warnings. Tacking on c2271c9 to resolve those and will merge by automerge:rebase

[gibson042]

Removing this eliminates coverage that arbitrary values can be published, but that coverage probably is of little value.

[gibson042]

Might as well be consistent here.

Suggested change

	void Promise.resolve(sub1.subscribeAfter()).then(result =>
	sub1FirstAll.push(result),
	);
	Promise.resolve(sub1.subscribeAfter())
	.then(result => sub1FirstAll.push(result))
	.catch(t.fail);

[gibson042]

Suggested change

	void Promise.resolve(sub1.subscribeAfter(0n)).then(result =>
	sub1FirstAll.push(result),
	);
	Promise.resolve(sub1.subscribeAfter(0n))
	.then(result => sub1FirstAll.push(result))
	.catch(t.fail);