Add dialog light dismiss behavior #10737

mfreed7 · 2024-10-31T23:17:09Z

This adds a few things:

A closedby attribute that can be used to control the light dismiss behavior of dialogs:
1. <dialog closedby=none> - no user-triggered closing of dialogs at all.
2. <dialog closedby=closerequest> - user pressing ESC (or other close trigger) closes the dialog
3. <dialog closedby=any> - user clicking outside the dialog, or pressing ESC, closes the dialog. Akin to popover=auto behavior.
4. <dialog> - i.e. no closedby attribute - default/old behavior. Behaves like closedby=closerequest for modal dialogs, and closedby=none for modeless dialogs.
This PR also (per the request) adds a dialog.requestClose() which fires the cancel event, and then (if that isn't cancelled) fires the close event and closes the dialog.
I also consolidated the "light dismiss" activities into one algorithm, which will need to be called by Add popover light dismiss integration w3c/pointerevents#460 if that ever gets some attention.

Some interesting related links:

WIP dialog closedby attribute spec #10157 - much of this PR was lifted from @lukewarlow's PR
Add light dismiss functionality to <dialog> #9373 - the WHATWG discussion
Bikeshed a name for "light dismiss for dialog" openui/open-ui#834 - the OpenUI discussion
Naming for the *attribute* to enable "light dismiss for dialog" openui/open-ui#960 - naming discussion to get to closedby
Naming for the *value* for the attribute that means "just close with ESC" openui/open-ui#961 - naming discussion to get to closerequest

At least two implementers are interested (and none opposed):
- Chromium
- WebKit: Presumably supportive, based on @annevk reviews.
- Gecko: @smaug---- has said he would review, so TBD.
Tests are written and can be reviewed and commented upon at:
- https://wpt.fyi/results/html/semantics/interactive-elements/the-dialog-element?label=master&label=experimental&aligned&q=dialog-closedby.tentative.html%20or%20dialog-popover-closedby-complex.tentative.html%20or%20dialog-popover-closedby-simple.tentative.html%20or%20dialog-requestclose.tentative.html
Implementation bugs are filed:
- Chromium: https://crbug.com/376516550
- Gecko: https://bugzilla.mozilla.org/show_bug.cgi?id=1936940
- WebKit: https://bugs.webkit.org/show_bug.cgi?id=284592
MDN issue is filed: Add documentation of new dialog light dismiss behaviors mdn/content#37196
The top of this comment includes a clear commit message to use.

(See WHATWG Working Mode: Changes for more details.)

/dnd.html ( diff )
/dom.html ( diff )
/index.html ( diff )
/indices.html ( diff )
/infrastructure.html ( diff )
/interaction.html ( diff )
/interactive-elements.html ( diff )
/popover.html ( diff )
/semantics-other.html ( diff )

domenic · 2024-11-01T08:14:59Z

This is exciting! Unfortunately I am going OOO for a week starting ~now, so won't get a chance to review too soon. Hopefully others will be able to help with some reviews, but I would like to get a chance to carefully review the close watcher integration, etc. before this lands.

Would you mind commenting on what approach you took to the mutability of the closedby attribute? We discussed that a bit in #9373, and I think we came up with a tentative proposal, but I'm not sure it was battle-tested.

keithamus · 2024-11-01T12:03:59Z

Would you mind commenting on what approach you took to the mutability of the closedby attribute? We discussed that a bit in #9373, and I think we came up with a tentative proposal, but I'm not sure it was battle-tested.

Looks like CloseWatchers get a mutable enabled value, which is toggled whenever closedby is set to the applicable values - when false it skips the events and returns early. IMO this seems to nicely side-skirt the stacking issues around mutation of the attribute without adding lots of complexity in either the consuming elements and/or having some messy algorithms to insert mid-way into a stack.

keithamus

This looks great to me!

source

mfreed7 · 2024-11-01T16:07:01Z

This is exciting! Unfortunately I am going OOO for a week starting ~now, so won't get a chance to review too soon. Hopefully others will be able to help with some reviews, but I would like to get a chance to carefully review the close watcher integration, etc. before this lands.

That's totally fine - no rush. I did this one in the opposite order that I usually do: spec first, prototype after. So I'm moving on to prototype it now, which might reveal other flaws. But I thought since @lukewarlow had gotten the spec into such decent shape, that I'd clean it up and open the PR in case folks saw other flaws there that I could address in the prototype.

Would you mind commenting on what approach you took to the mutability of the closedby attribute? We discussed that a bit in #9373, and I think we came up with a tentative proposal, but I'm not sure it was battle-tested.

As @keithamus said, I just kept what @lukewarlow did in the original PR, and used a new enabled state on CloseWatcher. I haven't implemented that yet, but it seems reasonably straightforward. But let me know if you see any flaws.

source

annevk · 2024-11-05T09:59:35Z

source

+  <p>The <code data-x="attr-dialog-closedby">closedby</code> attribute's <i data-x="invalid value
+  default">invalid value default</i> and <i data-x="missing value default">missing value
+  default</i> are both the <dfn data-x="attr-closedby-auto-state">Auto</dfn> state. The <span
+  data-x="attr-closedby-auto-state">auto</span> state matches <span
+  data-x="attr-closedby-closerequest-state">closerequest</span> when the element is modal;
+  otherwise <span data-x="attr-closedby-none-state">none</span>.</p>


There's very inconsistent spelling of states here. They should match the capitalization and words of how they are introduced. And should typically be followed by the word state (though that's not linked).

Also, perhaps the statement of fact is better as a note?

There's very inconsistent spelling of states here. They should match the capitalization and words of how they are introduced. And should typically be followed by the word state (though that's not linked).

Yep, I agree. I fixed them all up, I think. Please let me know if you see any others I missed. All are now lowercase.

Also, perhaps the statement of fact is better as a note?

Sorry - can you help me understand this? Which "statement of fact"?

This CL puts the feature flag in place, adds (flag guarded) closedBy and requestClose() methods to <dialog>, connects the pointer events handling to a new dialog light dismiss method, and adds a basic set of tests. None of the actual functionality is here yet, this is just a shell. Subsequent CLs will flesh out the behavior. See spec PR for details: whatwg/html#10737 Here's the chromestatus: https://chromestatus.com/feature/5097714453577728 Bug: 376516550 Change-Id: I3727ca21476a2a3340fd18597970395d64ef7176

This CL puts the feature flag in place, adds (flag guarded) closedBy and requestClose() methods to <dialog>, connects the pointer events handling to a new dialog light dismiss method, and adds a basic set of tests. None of the actual functionality is here yet, this is just a shell. Subsequent CLs will flesh out the behavior. See spec PR for details: whatwg/html#10737 Here's the chromestatus: https://chromestatus.com/feature/5097714453577728 Bug: 376516550 Change-Id: I3727ca21476a2a3340fd18597970395d64ef7176 Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/5985491 Commit-Queue: Mason Freed <[email protected]> Reviewed-by: David Baron <[email protected]> Cr-Commit-Position: refs/heads/main@{#1378633}

This implements reflection of `closedBy` including the "limited to known values" behavior. See spec PR for details: whatwg/html#10737 Bug: 376516550 Change-Id: Iddefd573fe19fd39f4b3aebe13390235fea969b9

…ht dismiss [1/N], a=testonly Automatic update from web-platform-tests Implement basic structure for dialog light dismiss [1/N] This CL puts the feature flag in place, adds (flag guarded) closedBy and requestClose() methods to <dialog>, connects the pointer events handling to a new dialog light dismiss method, and adds a basic set of tests. None of the actual functionality is here yet, this is just a shell. Subsequent CLs will flesh out the behavior. See spec PR for details: whatwg/html#10737 Here's the chromestatus: https://chromestatus.com/feature/5097714453577728 Bug: 376516550 Change-Id: I3727ca21476a2a3340fd18597970395d64ef7176 Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/5985491 Commit-Queue: Mason Freed <[email protected]> Reviewed-by: David Baron <[email protected]> Cr-Commit-Position: refs/heads/main@{#1378633} -- wpt-commits: 1d5037cc3ff4524fd5d9f1a461afd20aebbb0839 wpt-pr: 48976

This implements reflection of `closedBy` including the "limited to known values" behavior. See spec PR for details: whatwg/html#10737 Bug: 376516550 Change-Id: Iddefd573fe19fd39f4b3aebe13390235fea969b9

This implements reflection of `closedBy` including the "limited to known values" behavior. See spec PR for details: whatwg/html#10737 Bug: 376516550 Change-Id: Iddefd573fe19fd39f4b3aebe13390235fea969b9 Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/5988652 Commit-Queue: Mason Freed <[email protected]> Reviewed-by: David Baron <[email protected]> Commit-Queue: David Baron <[email protected]> Auto-Submit: Mason Freed <[email protected]> Cr-Commit-Position: refs/heads/main@{#1379228}

This enables `closedby=none` to *stop* the dialog from closing when the ESC key (or other close signal) is received. See spec PR for details: whatwg/html#10737 Bug: 376516550 Change-Id: I0bf7fc7a50acaffb89e183fe6e19f411de4da576 Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/5990171 Reviewed-by: David Baron <[email protected]> Commit-Queue: Mason Freed <[email protected]> Cr-Commit-Position: refs/heads/main@{#1379279}

…ht dismiss [1/N], a=testonly Automatic update from web-platform-tests Implement basic structure for dialog light dismiss [1/N] This CL puts the feature flag in place, adds (flag guarded) closedBy and requestClose() methods to <dialog>, connects the pointer events handling to a new dialog light dismiss method, and adds a basic set of tests. None of the actual functionality is here yet, this is just a shell. Subsequent CLs will flesh out the behavior. See spec PR for details: whatwg/html#10737 Here's the chromestatus: https://chromestatus.com/feature/5097714453577728 Bug: 376516550 Change-Id: I3727ca21476a2a3340fd18597970395d64ef7176 Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/5985491 Commit-Queue: Mason Freed <[email protected]> Reviewed-by: David Baron <[email protected]> Cr-Commit-Position: refs/heads/main@{#1378633} -- wpt-commits: 1d5037cc3ff4524fd5d9f1a461afd20aebbb0839 wpt-pr: 48976

…alog closedBy [2/N], a=testonly Automatic update from web-platform-tests Add attribute reflection behavior for dialog closedBy [2/N] This implements reflection of `closedBy` including the "limited to known values" behavior. See spec PR for details: whatwg/html#10737 Bug: 376516550 Change-Id: Iddefd573fe19fd39f4b3aebe13390235fea969b9 Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/5988652 Commit-Queue: Mason Freed <[email protected]> Reviewed-by: David Baron <[email protected]> Commit-Queue: David Baron <[email protected]> Auto-Submit: Mason Freed <[email protected]> Cr-Commit-Position: refs/heads/main@{#1379228} -- wpt-commits: 28946b88d5b2e037bb2506ae4984e8725ba5560b wpt-pr: 48989

This implements dialog.requestClose() and adds a test. See spec PR for details: whatwg/html#10737 Bug: 376516550 Change-Id: Iaac3d89c28844d2b54ff5b1a7b68dc356d1fd172

source

mfreed7

Thanks for the review!

source

domenic

Overall looking good, except I don't understand the light dismiss targeting behavior and I think I need you to explain it to me.

source

domenic · 2024-12-10T06:23:40Z

Oh, I spotted another issue, around the use of "request to close". I will discuss on openui/open-ui#1128.

mfreed7 · 2024-12-10T18:31:07Z

Oh, I spotted another issue, around the use of "request to close". I will discuss on openui/open-ui#1128.

Note that my most recent commit does not yet handle this. I'll work on that now.

mfreed7 · 2024-12-10T18:44:00Z

Oh, I spotted another issue, around the use of "request to close". I will discuss on openui/open-ui#1128.

Note that my most recent commit does not yet handle this. I'll work on that now.

Alright, latest commit adds something - easier than I thought. I'll implemented the equivalent thing in Chromium and it looks pretty reasonable to me. LMK what you think.

domenic · 2024-12-11T06:19:52Z

source

@@ -61878,6 +61881,13 @@ interface <dfn interface>HTMLDialogElement</dfn> : <span>HTMLElement</span> {
   <li><p>Add an <code data-x="attr-dialog-open">open</code> attribute to <span>this</span>, whose
   value is the empty string.</p></li>

+   <li><p><span>Assert</span>: <span>this</span>'s <span>node document</span>'s <span>open
+   dialogs list</span> does not <span data-x="list contains">contain</span>
+   <span>this</span>.</p></li>


I think this assert fails if you do

dialogEl.show(); dialogEl.removeAttribute("open"); dialogEl.show();

#10124 will fix this, but I'm hesitant to block this PR on that.

Yeah, you're right. The Chromium code has to work around the state of that feature flag.

Since that PR "fixes" the behavior for removing the open attribute, would you be ok with me just removing this Assert, rather than trying to bring in much of that PR?

Yeah, as long as you can verify the assert is not load-bearing, then just removing it for now seems reasonable.

Well, it's load bearing in the case you bring up above - removing the open attribute. But the spec isn't super clear about that now anyway, so I guess it only bears code-implementation weight, not spec weight. LMK if that makes sense.

source

domenic · 2024-12-11T06:25:38Z

source

@@ -83166,7 +83448,7 @@ dictionary <dfn dictionary>CloseWatcherOptions</dfn> {
  <p>The <dfn method for="CloseWatcher"><code
  data-x="dom-CloseWatcher-requestClose">requestClose()</code></dfn> method steps are to <span
  data-x="close-watcher-request-close">request to close</span> <span>this</span>'s <span>internal
-  close watcher</span>.</p>
+  close watcher</span> with false.</p>


I think this should also be true, to match <dialog>. (This may require updating some tests to use the sendCloseRequest() helper instead of using requestClose().)

See my comment here openui/open-ui#1128 (comment). I'm a little worried about compat, and I'd prefer to split out this change from the dialog change. If that works for you?

... per my next comment openui/open-ui#1128 (comment) I've made this change. I'll leave this comment open to ensure you notice it.

domenic

Provisionally approving under the assumption the broken assert is removed. Great stuff!

This will still need the rest of the checkboxes filled out, and I'm going OOO for the rest of the year. But I might check in since there's probably not much left to do here.

One thing I haven't checked recently is the WPTs. I'd encourage adding lots and lots, e.g. for the stuff we changed recently around allowing cancelable requestClose(), tree order vs. flat tree order vs. dialog open order, etc. This stuff is just so subtle you can never have enough tests.

source

smaug---- · 2024-12-12T12:57:34Z

FWIW, I started to review the PR but need to still think whether there are any unexpected behaviors related to click handling and targeting.

mfreed7 · 2024-12-12T17:50:18Z

Provisionally approving under the assumption the broken assert is removed. Great stuff!

Thanks for the detailed review!

This will still need the rest of the checkboxes filled out, and I'm going OOO for the rest of the year. But I might check in since there's probably not much left to do here.

Ok, sounds good. No rush - January is ok by me.

One thing I haven't checked recently is the WPTs. I'd encourage adding lots and lots, e.g. for the stuff we changed recently around allowing cancelable requestClose(), tree order vs. flat tree order vs. dialog open order, etc. This stuff is just so subtle you can never have enough tests.

Yeah, I'll give it more thought. I did add some complex cases, but I'll see what else I can add. I agree it's complicated and needs good testing.

mfreed7 · 2024-12-12T17:50:31Z

FWIW, I started to review the PR but need to still think whether there are any unexpected behaviors related to click handling and targeting.

Great, thanks for reviewing. Let me know what you find.

Per the conversation here: openui/open-ui#1128 (comment) There's a desire to ship closeWatcher.requestClose() along with dialog.requestclose() both not requiring user activation. The spec PR has been updated accordingly: whatwg/html#10737 (in this commit: whatwg/html@76619aa) Fixed: 383593252 Change-Id: I8c05e352d8b4964407a1cee36f35372e192e2ca5 Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/6092173 Auto-Submit: Mason Freed <[email protected]> Reviewed-by: David Baron <[email protected]> Reviewed-by: Mason Freed <[email protected]> Commit-Queue: Mason Freed <[email protected]> Cr-Commit-Position: refs/heads/main@{#1396007}

keithamus reviewed Nov 1, 2024

View reviewed changes

source Outdated Show resolved Hide resolved

mfreed7 mentioned this pull request Nov 1, 2024

Support disabling CloseWatcher integration in <dialog> #10592

Open

annevk added addition/proposal New features or enhancements needs implementer interest Moving the issue forward requires implementers to express interest needs tests Moving the issue forward requires someone to write tests topic: dialog The <dialog> element labels Nov 5, 2024

annevk reviewed Nov 5, 2024

View reviewed changes

source Outdated Show resolved Hide resolved

annevk reviewed Nov 5, 2024

View reviewed changes

mfreed7 mentioned this pull request Nov 5, 2024

Add light dismiss functionality to <dialog> #9373

Open

chromium-wpt-export-bot mentioned this pull request Nov 5, 2024

Implement basic structure for dialog light dismiss [1/N] web-platform-tests/wpt#48976

Merged

chromium-wpt-export-bot mentioned this pull request Nov 5, 2024

Add attribute reflection behavior for dialog closedBy [2/N] web-platform-tests/wpt#48989

Merged

chromium-wpt-export-bot mentioned this pull request Nov 8, 2024

Implement dialog.requestClose() [4/N] web-platform-tests/wpt#49050

Merged

annevk reviewed Nov 8, 2024

View reviewed changes

lukewarlow reviewed Dec 5, 2024

View reviewed changes

source Show resolved Hide resolved

Remove extra section

f5a0919

mfreed7 commented Dec 6, 2024

View reviewed changes

source Show resolved Hide resolved

source Show resolved Hide resolved

lukewarlow reviewed Dec 6, 2024

View reviewed changes

source Show resolved Hide resolved

mfreed7 added 2 commits December 6, 2024 11:26

Remove exception

3f413d3

Add closewatcher override for requestClose

d9ed19f

annevk reviewed Dec 9, 2024

View reviewed changes

source Show resolved Hide resolved

source Outdated Show resolved Hide resolved

lukewarlow mentioned this pull request Dec 9, 2024

[invokers] Should invokers on dialog have a cancel action? openui/open-ui#938

Open

Address 2 nits

ff98f98

mfreed7 mentioned this pull request Dec 9, 2024

What should "nested" dialog light dismiss look like? openui/open-ui#1128

Closed

domenic reviewed Dec 10, 2024

View reviewed changes

source Show resolved Hide resolved

source Outdated Show resolved Hide resolved

source Outdated Show resolved Hide resolved

source Show resolved Hide resolved

source Show resolved Hide resolved

source Show resolved Hide resolved

source Show resolved Hide resolved

Address domenic comments

2fb87b1

Side-step closewatcher abuse prevention

6aa7386

domenic reviewed Dec 11, 2024

View reviewed changes

allowPreventClose -> requireHistoryActionActivation

344d87f

domenic approved these changes Dec 12, 2024

View reviewed changes

annevk reviewed Dec 12, 2024

View reviewed changes

source Show resolved Hide resolved

mfreed7 added 2 commits December 12, 2024 09:44

Change closeWatcher.requestClose() also

76619aa

Address two comments

57df925

past mentioned this pull request Dec 13, 2024

Upcoming WHATNOT meeting on 2024-12-12 #10825

Closed

mfreed7 mentioned this pull request Dec 13, 2024

Add documentation of new dialog light dismiss behaviors mdn/content#37196

Open

chromium-wpt-export-bot mentioned this pull request Dec 13, 2024

Make closeWatcher.requestClose() not require user activation web-platform-tests/wpt#49681

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Add dialog light dismiss behavior #10737

Add dialog light dismiss behavior #10737

mfreed7 commented Oct 31, 2024 •

edited by pr-preview bot

Loading

domenic commented Nov 1, 2024

keithamus commented Nov 1, 2024

keithamus left a comment

mfreed7 commented Nov 1, 2024

annevk Nov 5, 2024

annevk Nov 5, 2024

mfreed7 Nov 8, 2024

mfreed7 left a comment

domenic left a comment

domenic commented Dec 10, 2024

mfreed7 commented Dec 10, 2024

mfreed7 commented Dec 10, 2024 •

edited

Loading

domenic Dec 11, 2024

mfreed7 Dec 11, 2024

domenic Dec 12, 2024

mfreed7 Dec 12, 2024

domenic Dec 11, 2024

mfreed7 Dec 11, 2024

mfreed7 Dec 12, 2024

domenic left a comment

smaug---- commented Dec 12, 2024

mfreed7 commented Dec 12, 2024

mfreed7 commented Dec 12, 2024

Add dialog light dismiss behavior #10737

Are you sure you want to change the base?

Add dialog light dismiss behavior #10737

Conversation

mfreed7 commented Oct 31, 2024 • edited by pr-preview bot Loading

domenic commented Nov 1, 2024

keithamus commented Nov 1, 2024

keithamus left a comment

Choose a reason for hiding this comment

mfreed7 commented Nov 1, 2024

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

mfreed7 left a comment

Choose a reason for hiding this comment

domenic left a comment

Choose a reason for hiding this comment

domenic commented Dec 10, 2024

mfreed7 commented Dec 10, 2024

mfreed7 commented Dec 10, 2024 • edited Loading

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

domenic left a comment

Choose a reason for hiding this comment

smaug---- commented Dec 12, 2024

mfreed7 commented Dec 12, 2024

mfreed7 commented Dec 12, 2024

mfreed7 commented Oct 31, 2024 •

edited by pr-preview bot

Loading

mfreed7 commented Dec 10, 2024 •

edited

Loading