Make fragment loop not depend on spin the event loop

[annevk]

This only seems worth doing if we can address all of #3387 I think, as it's quite a bit more involved wording-wise. (It does seem a little weird that the current setup gives both the networking task source. I would have expected the DOM manipulation task source or some such for finding the fragment.)

---

/browsing-the-web.html ( diff )

[domenic]

Personally I would be in favor of merging this ASAP as I think it's a significant clarity improvement just by being more explicit, e.g. using clearer looping constructs.

[domenic]

Personally I would rather move us toward a style of stating the task source in the same line as the "queue a task" (i.e. "queue a task on the networking task source to..."). But I can understand if you want to do that as a larger sweeping change instead of one by one.

[annevk]

I'd prefer making "queue a task" more explicit about its arguments first.

[domenic]

@annevk shall we merge this?

[annevk]

Reading this again it seems like the new version doesn't protect against multiple tasks being queued that might potentially do the same thing. What is desired seems to be that at the end of the task a new task is queued to the "in parallel" thread to run the steps again. We could perhaps emulate this with a second taskRun variable or some such upon which the "in parallel" thread waits for a change.

[domenic]

Reading this again it seems like the new version doesn't protect against multiple tasks being queued that might potentially do the same thing.

You mean, because navigate happened twice? I think that's what the "If the Document object has no parser..." business prevents. Either way, I don't really see what changed that would make things different before this PR versus after.

Let's take a step back. The intention of this section seems to be basically to, at some later time, queue a task to scroll to the fragment, if doing so still makes sense. If that doesn't work, try again later.

Here's my attempt at reframing the spec-code to match that more Englishey description.

3. The navigation algorithm has now matured.

4. Try to scroll to the fragment for the Document.

To try to scroll to the fragment for Document document,

1. In parallel,

   1. Wait a user-agent-defined amount of time.

   2. Queue a task on the neworking task source to run these steps:

      1. If document has no parser, or its parser has stopped parsing, or the user agent has reason to believe the user is no longer interested in scrolling to the fragment, then abort these steps.

      2. Otherwise, scroll to the fragment given in document's URL. If this does not find an indicated part of the document, then try to scroll to the fragment for document.

How does that sound?

[annevk]

I meant because of the "while true". I think your rephrasing would work.

[domenic]

@annevk did another take based on my above comment; let me know what you think.