Editorial: restructure ownership of workers to parent-owners #2520

annevk · 2017-04-11T11:20:10Z

Instead of each worker being owned by a document, have it instead be
owned by one or more of its parents (more only in the case of
SharedWorkerGlobalScope).

This makes it possible to define JavaScript agent clusters (#2260) and
also helps with allowing service workers to have nested workers (#411).

This is marked editorial as it should have no normative impact.

This also removes a step that should have been removed as part of
4e2b006.

Instead of each worker being owned by a document, have it instead be owned by one or more of its parents (more only in the case of SharedWorkerGlobalScope). This makes it possible to define JavaScript agent clusters (#2260) and also helps with allowing service workers to have nested workers (#411). This is marked editorial as it should have no normative impact. This also removes a step that should have been removed as part of 4e2b006.

annevk · 2017-04-11T11:25:39Z

@wanderview @bakulf I'd appreciate review. I didn't go for changing the owner from document to Window at this time, but this makes such a change easier to evaluate going forward.

The main reason I'm making this change is to unblock SharedArrayBuffer.

domenic

LGTM with some nits, thanks so much.

domenic · 2017-04-13T00:12:44Z

source

@@ -79134,8 +79135,8 @@ callback <dfn>FrameRequestCallback</dfn> = void (<span>DOMHighResTimeStamp</span
  </ol>

  <p class="note">Whenever a <code>Document</code> object is <span data-x="discard a
-  Document">discarded</span>, it is also removed from the list of <span>the worker's
-  <code>Document</code>s</span> of each worker whose list contains that <code>Document</code>.</p>
+  Document">discarded</span>, it is also removed from the list of <span>the worker's owners</span>


Nit: I think better linking text would be "owners" here, although I recognize you're just going with the existing pattern. But as seen in the agents PR, I think it gets really ugly the more it's used.

Yeah. I could make these all more clearly slot-like things of the WorkerGlobalScope instead of the double naming. But then we'd also have to change "the worker's ports" and maybe some other things too. I went for a smaller change, but happy to go more ambitious.

Oh, I see. I guess being inconsistent would be not great. Up to you; we can leave it for later. In which case ignore a lot of my reviews here and in the other PR.

Note that I did set up exporting for the term I expect we'll eventually settle on.

domenic · 2017-04-13T00:14:41Z

source

+
+  <p>Each <code>WorkerGlobalScope</code> object also has a <span>set</span> of <dfn
+  id="the-worker's-workers" data-export="" data-for="WorkerGlobalScope" data-lt="workers">the
+  worker's workers</dfn>. Initially this set is empty; it is populated when the worker creates or


Reading this sentence made the grammar situation clear for me:

Workers have owners/workers

WorkerGlobalScopes have "the worker's owners"/"the worker's workers"

I apologize if I gave feedback on the agents PR that didn't fit with the above framework.

domenic · 2017-04-13T00:15:18Z

source

+  owners</dfn>. Initially this set is empty; it is populated when the worker is created or
+  obtained.</p>
+
+  <p class="note">It is a list to accomodate <code>SharedWorkerGlobalScope</code> objects.</p>


s/list/set, instead of a single owner,/ maybe

domenic · 2017-04-13T00:15:49Z

source

-  Otherwise, <var>o</var> specifies a <span data-x="concept-settings-object-global">global
-  object</span> that is a <code>Window</code> object, and the relevant <code>Document</code> is just
-  the <span>responsible document</span> specified by <var>o</var>.</p>
+  it must be removed from the list of <span>the worker's owners</span> of each worker whose list


Nit: so here this can be "from the owners of each worker..."

domenic · 2017-04-13T00:17:07Z

source

-  <p>A worker is said to be a <dfn>permissible worker</dfn> if its list of <span>the worker's
-  <code>Document</code>s</span> is not empty, or if its list has been empty for no more than a short
+  <p>A worker is said to be a <dfn>permissible worker</dfn> if its set of <span>the worker's
+  owners</span> is not empty, or if its list has been empty for no more than a short


Nit: "set of owners"

domenic · 2017-04-13T00:17:55Z

source


  <hr>

-  <p>A worker is said to be a <dfn>permissible worker</dfn> if its list of <span>the worker's
-  <code>Document</code>s</span> is not empty, or if its list has been empty for no more than a short
+  <p>A worker is said to be a <dfn>permissible worker</dfn> if its set of <span>the worker's


While here refactoring this to a bulleted list ("all of the following are true") would greatly improve clarity

domenic · 2017-04-13T00:18:09Z

source

-  <p>A worker is said to be an <dfn>active needed worker</dfn> if any of the <code>Document</code>
-  objects in <span>the worker's <code>Document</code>s</span> are <span>fully active</span>.</p>
+  <p>A worker is said to be an <dfn>active needed worker</dfn> if any of <span>the worker's
+  owners</span> are either <code>Document</code> objects that are <span>fully active</span> or


Nit: "its owners"

domenic · 2017-04-13T00:19:22Z

source

-       settings</var>.</p></li>
+       <li><p><span data-x="set append">Append</span> the <span>relevant owner to add</span> given
+       <var>outside settings</var> to <var>worker global scope</var>'s set of <span>the worker's
+       owners</span>.</p></li>


Isn't moving from a list of relevant documents to add to a single owner a normative change? (One that aligns with reality, I imagine.)

I don't think so. Before it could be a list because an ancestor owner might be a SharedWorkerGlobal object with multiple document owners and those documents would just trickle down. Now we go through the hierarchy.

…ction later)

annevk · 2017-04-13T07:37:42Z

I went ahead and renamed them them "owner set" and "worker set" anyway. I found out that if I moved them into the WorkerGlobalScope section that would not look so weird and "the worker's ports" just remains as an item for future cleanup in the lifetime section (and probably also needs some further clarification since it's not a set that's explicitly being added too at the moment, though that might be rather easy to change if we define MessagePort allocation).

domenic · 2017-04-13T22:23:39Z

I pushed a small commit that links to and uses more Infra terms while we're in the area. LGTM!

This aligns with the changes made in whatwg/html#2520 whatwg/html@59a4750

annevk added the topic: workers label Apr 11, 2017

This was referenced Apr 11, 2017

Define agents and agent clusters #2521

Merged

Allow workers & shared workers to be created within a service worker #411

Closed

Revamp of SharedArrayBuffer PR #2518

Merged

domenic approved these changes Apr 13, 2017

View reviewed changes

rename the worker's X except for ports (can be moved from lifetime se…

bbbbfaf

…ction later)

This was referenced Apr 13, 2017

Move "the worker's ports" into the WorkerGlobalScope section #2527

Open

Integrate JavaScript SharedArrayBuffers and agents/agent clusters #2361

Closed

domenic mentioned this pull request Apr 13, 2017

Move owner set manipulation during document discarding #2533

Closed

Use moar Infra

a25f8ca

annevk merged commit 59a4750 into master Apr 14, 2017

annevk deleted the annevk/worker-ownership branch April 14, 2017 05:48

annevk mentioned this pull request Apr 14, 2017

Lifetime of service workers in relation to dedicated and shared workers w3c/ServiceWorker#1119

Open

sideshowbarker added a commit to w3c/ServiceWorker that referenced this pull request Apr 19, 2017

“the worker's Documents” ➡ “owner set”

f0b6a2a

This aligns with the changes made in whatwg/html#2520 whatwg/html@59a4750

sideshowbarker mentioned this pull request Apr 19, 2017

“the worker's Documents” ➡ “owner set” w3c/ServiceWorker#1124

Merged

sideshowbarker added a commit to w3c/ServiceWorker that referenced this pull request Jun 11, 2017

“the worker's Documents” ➡ “owner set”

ba7aa8f

This aligns with the changes made in whatwg/html#2520 whatwg/html@59a4750

jungkees pushed a commit to w3c/ServiceWorker that referenced this pull request Jun 12, 2017

“the worker's Documents” ➡ “owner set” (#1124)

443753f

This aligns with the changes made in whatwg/html#2520 whatwg/html@59a4750

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Editorial: restructure ownership of workers to parent-owners #2520

Editorial: restructure ownership of workers to parent-owners #2520

annevk commented Apr 11, 2017

annevk commented Apr 11, 2017

domenic left a comment

domenic Apr 13, 2017

annevk Apr 13, 2017

domenic Apr 13, 2017

annevk Apr 13, 2017

domenic Apr 13, 2017

domenic Apr 13, 2017

domenic Apr 13, 2017

domenic Apr 13, 2017

domenic Apr 13, 2017

domenic Apr 13, 2017

domenic Apr 13, 2017

annevk Apr 13, 2017

annevk commented Apr 13, 2017

domenic commented Apr 13, 2017

Editorial: restructure ownership of workers to parent-owners #2520

Editorial: restructure ownership of workers to parent-owners #2520

Conversation

annevk commented Apr 11, 2017

annevk commented Apr 11, 2017

domenic left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

annevk commented Apr 13, 2017

domenic commented Apr 13, 2017