Add note about rounding coordinates for click, auxclick, contextmenu #404

patrickhlauke · 2021-08-04T12:38:28Z

See the comment about web compat issues with the "upgraded" click/auxclick/contextmenu as pointer events #100 (comment)

Closes #100

patrickhlauke · 2021-08-04T12:39:43Z

Currently set this as a note, as I'm not sure if we want to set this in stone as a requirement (though I guess then that invalidates the SHOULD I used ... not sure). Let's discuss at today's meeting.

patrickhlauke · 2021-08-04T12:40:57Z

Related: should we also mention anything related to this explicitly in the mouse compatibility section? So user agents know that they should also round/convert to long when generating compat mouse events?

patrickhlauke · 2021-08-04T12:45:38Z

Last tangential point: currently, the definition for {{PointerEvent}} links to https://w3c.github.io/pointerevents/#dom-pointerevent - should it not go to what is essentially the interface https://w3c.github.io/pointerevents/#pointerevent-interface ? (if so, just need to repoint the <dfn> appropriately). To me this would make more sense, but not sure what the conventions are for specs...

patrickhlauke · 2021-08-04T13:12:57Z

And last additional point, I promise: wondering if the current note in 4.1

The PointerEvent interface inherits from MouseEvent, defined in UI Events and extended by CSSOM View Module.

should actually be expanded to make clear what that last bit means ... fractional coordinates. took me a while to work out that nowhere in the PE spec we actually explicitly say anything about fractional coordinates/that they're double

…oordinates normative

… it more obvious/easily scannable

patrickhlauke · 2021-08-04T20:48:08Z

Made the changes discussed in today's meeting:

expanded the note at the end of 4.1 (to foreshadow the topic a bit more about what CSSOM proposes, the fact that it's still only a proposal not a standard, and that there are UA requirements for those UAs that do already follow this)
made the proposed note for 4.2.12 normative instead

While I was in 4.2.12, also cleaned up the structure of the prose there to make the distinction between events that came from pointing devices and those that didn't more obvious when scanning over the section.

patrickhlauke · 2021-08-05T12:17:05Z

Just adding here, now that I thought it through some more: so the compat problems/issues would come about in user agents where PointerEvents have been implemented already using fractional coordinates, but the same implementation hasn't been extended to regular MouseEvents, right? If a browser decided to already implement fractional coordinates for MouseEvents, this would actually lead to problems again (essentially, the opposite type of problem this tries to address), right?

In short, the problem only appeared because Chrome (and others?) didn't apply the CSSOM approach for MouseEvents, but just for PointerEvents, right? So my proposed sentence

For user agents that already implement this proposed extension, there are additional normative requirements when it comes to the click, auxclick, and contextmenu events

should probably more specifically be about

For user agents that already implement this proposed extension for PointerEvents, but not for MouseEvents, there are additional normative requirements when it comes to the click, auxclick, and contextmenu events

likewise, in

However, this change — when applied to click, auxclick, and contextmenu — has proven to lead to web compatibility issues with legacy code.

it should probably read more along the lines of

However, this change — when applied to PointerEvents, but not to regular MouseEvents — has proven to lead to web compatibility issues with legacy code when applied to click, auxclick, and contextmenu.

i.e. if the user agent hasn't implemented fractional coordinates anywhere (not in MouseEvents, not in PointerEvents), it's a non-issue. and it's also a non-issue if the user agent has applied this change to both MouseEvents and PointerEvents. it's only a problem when it's only implemented in one and not the other, right?

patrickhlauke · 2021-08-05T12:35:22Z

Just pushed a change to this PR to clarify this further. I think this now more accurately explains where the problem lies, and when this requirement needs to be implemented.

mustaqahmed · 2021-08-05T17:42:25Z

Just pushed a change to this PR to clarify this further. I think this now more accurately explains where the problem lies, and when this requirement needs to be implemented.

I am afraid it was more accurate before this last commit! Let me add a few points to focus on, in case they call for more discussion:

Double coordinates (as per CSSOM View spec) are known to be better but we can't move MouseEvent coordinates away from long for compat reasons. We logically expect that all UAs would pick double coordinates for PointerEvents (as per CSSOM) even if those UAs remain stuck with long coordinates in MouseEvents (as per UI Event spec).
The expectation above applied only to "original" PointerEvents (pointerdown, pointermove etc). The "newly promoted" PointerEvents (click, auxclick and contextmenu) suffer from the similar compat issues like MouseEvents.
The two points above are orthogonal to compat MouseEvents.
The compat event section only defines the firing of followup MouseEvents, without touching any other details of the events---because those details belong to the UIEvent spec. Adding more details here means more redundancy and may possibly call for more monkey-patching.

[EDIT: I mentioned these points only for discussion. Most likely they don't deserve to be included in the PR here.]

patrickhlauke · 2021-08-05T18:55:04Z

We logically expect that all UAs would pick double coordinates for PointerEvents (as per CSSOM) even if those UAs remain stuck with long coordinates in MouseEvents (as per UI Event spec)

we don't normatively, explicitly, actually say this in our spec at the moment, right? A UA could implement PEs but with long rather than double? (as normatively the WebIDL only says we extend MouseEvent, and the CSSOM change is mentioned in an informative note only.

we can't move MouseEvent coordinates away from long for compat reasons

but IF a user agent decided to apply the CSSOM fractional coordinates to their regular MouseEvent as well, then rounding click/auxclick/contextmenu to long would then cause the same weird mismatch, no? or is the expectation that the CSSOM change will never actually make it to regular MouseEvent ?

The two points above are orthogonal to compat MouseEvents.

yes and my PR doesn't mention compat events. when I talk about MouseEvent, I mean the actual ones.

mustaqahmed · 2021-08-05T20:03:51Z

we don't normatively, explicitly, actually say this in our spec at the moment, right? A UA could implement PEs but with long rather than double? (as normatively the WebIDL only says we extend MouseEvent, and the CSSOM change is mentioned in an informative note only.

we can't move MouseEvent coordinates away from long for compat reasons

but IF a user agent decided to apply the CSSOM fractional coordinates to their regular MouseEvent as well, then rounding click/auxclick/contextmenu to long would then cause the same weird mismatch, no? or is the expectation that the CSSOM change will never actually make it to regular MouseEvent ?

I ruled out this consideration because this is known to be impractical, therefore presumed all UAs ignore CSSOM extension for MouseEvents. After seeing (again!) that this can't be normative in the PE spec, I am finally able to see the motivation of your last patch. Phew!

The two points above are orthogonal to compat MouseEvents.

yes and my PR doesn't mention compat events. when I talk about MouseEvent, I mean the actual ones.

Looks like I misinterpreted "regular MouseEvent" as "compat MouseEvent", for no good reason! I am taking back my compat event points.

patrickhlauke · 2021-08-05T20:09:53Z

it's a tangled web that we weave...this CSSOM thing stretched my brain even more than that target discussion a few months ago...

index.html

…ript" This reverts commit 9718d17.

patrickhlauke · 2021-08-24T00:04:01Z

per the discussion at last week's meeting https://www.w3.org/2021/08/18-pointerevents-minutes.html changed the "round" to reference Math.floor and point to ECMAScript.

mustaqahmed

Two super picky comments!

index.html

mustaqahmed · 2023-02-15T19:00:16Z

I am looking into a WPT for this!

smaug---- · 2023-02-21T20:36:41Z

web-platform-tests/wpt#38525

mustaqahmed · 2023-02-28T21:22:35Z

Here I was modifying pointerevent_fractional_coordinates.html to cover integer coordinates in click-like events, and I think we still need that. Did I miss anything?

smaug---- · 2023-02-28T21:24:27Z

oops, I posted the link to wrong issue.

mustaqahmed · 2023-02-28T21:46:02Z

No worries, I updated the issue you meant: #456

This will resolve the "needs-wpt" label in w3c/pointerevents#404 Change-Id: I06e27cd6e3adb790dbf1355c9f8cc856335bebb7

This will resolve the "needs-wpt" label in w3c/pointerevents#404 Change-Id: I06e27cd6e3adb790dbf1355c9f8cc856335bebb7 Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/4797838 Reviewed-by: Kevin Ellis <[email protected]> Auto-Submit: Mustaq Ahmed <[email protected]> Commit-Queue: Kevin Ellis <[email protected]> Cr-Commit-Position: refs/heads/main@{#1187412}

mustaqahmed · 2023-08-23T20:23:14Z

A WPT has been added.

…tional coordinates., a=testonly Automatic update from web-platform-tests Add "click" to the existing WPT for fractional coordinates. This will resolve the "needs-wpt" label in w3c/pointerevents#404 Change-Id: I06e27cd6e3adb790dbf1355c9f8cc856335bebb7 Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/4797838 Reviewed-by: Kevin Ellis <[email protected]> Auto-Submit: Mustaq Ahmed <[email protected]> Commit-Queue: Kevin Ellis <[email protected]> Cr-Commit-Position: refs/heads/main@{#1187412} -- wpt-commits: bea5b03a40531e0beadde1a05228beaec5a98c33 wpt-pr: 41582

This will resolve the "needs-wpt" label in w3c/pointerevents#404 Change-Id: I06e27cd6e3adb790dbf1355c9f8cc856335bebb7 Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/4797838 Reviewed-by: Kevin Ellis <[email protected]> Auto-Submit: Mustaq Ahmed <[email protected]> Commit-Queue: Kevin Ellis <[email protected]> Cr-Commit-Position: refs/heads/main@{#1187412}

Add note about rounding coordinates for click, auxclick, contextmenu

30b3348

patrickhlauke marked this pull request as draft August 4, 2021 12:38

Expand note in 4.1, make requirement for click/auxclick/contextmenu c…

b07c68a

…oordinates normative

patrickhlauke marked this pull request as ready for review August 4, 2021 20:35

Clean up/restructure the prose for click/auxclick/contextmenu to make…

de7bc73

… it more obvious/easily scannable

patrickhlauke requested review from mustaqahmed, flackr and smaug---- August 4, 2021 20:48

smaug---- approved these changes Aug 5, 2021

View reviewed changes

Clarify further when this is a problem

5bca296

patrickhlauke requested a review from smaug---- August 5, 2021 12:35

mustaqahmed approved these changes Aug 5, 2021

View reviewed changes

flackr reviewed Aug 6, 2021

View reviewed changes

index.html Outdated Show resolved Hide resolved

patrickhlauke added 3 commits August 24, 2021 00:52

Replace vague "round" with Math.floor and reference to ECMAScript

9718d17

Revert "Replace vague "round" with Math.floor and reference to ECMASc…

d0b7b97

…ript" This reverts commit 9718d17.

Replace vague "round" with "Math.floor" and reference to ECMASCript

0cd83b0

Remove the "normative" from inside the note

bd755db

patrickhlauke requested a review from flackr August 24, 2021 00:07

patrickhlauke requested a review from mustaqahmed August 24, 2021 00:07

mustaqahmed reviewed Aug 24, 2021

View reviewed changes

index.html Outdated Show resolved Hide resolved

Readability tweak

34864ef

patrickhlauke merged commit 76c9195 into gh-pages Sep 1, 2021

patrickhlauke deleted the patrickhlauke-click-auxclick-contextmenu-coords branch September 1, 2021 15:12

patrickhlauke added the wpt label Aug 17, 2022

smaug---- added needs-wpt Investigation whether the issue needs a wpt test has been done and wpt is missing and removed wpt labels Nov 23, 2022

smaug---- removed the needs-wpt Investigation whether the issue needs a wpt test has been done and wpt is missing label Feb 21, 2023

mustaqahmed added the needs-wpt Investigation whether the issue needs a wpt test has been done and wpt is missing label Feb 28, 2023

chromium-wpt-export-bot mentioned this pull request Aug 22, 2023

Add "click" to the existing WPT for fractional coordinates. web-platform-tests/wpt#41582

Merged

mustaqahmed removed the needs-wpt Investigation whether the issue needs a wpt test has been done and wpt is missing label Aug 23, 2023

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Add note about rounding coordinates for click, auxclick, contextmenu #404

Add note about rounding coordinates for click, auxclick, contextmenu #404

patrickhlauke commented Aug 4, 2021 •

edited by pr-preview bot

Loading

patrickhlauke commented Aug 4, 2021

patrickhlauke commented Aug 4, 2021 •

edited

Loading

patrickhlauke commented Aug 4, 2021

patrickhlauke commented Aug 4, 2021

patrickhlauke commented Aug 4, 2021 •

edited

Loading

patrickhlauke commented Aug 5, 2021

patrickhlauke commented Aug 5, 2021

mustaqahmed commented Aug 5, 2021 •

edited

Loading

patrickhlauke commented Aug 5, 2021

mustaqahmed commented Aug 5, 2021

patrickhlauke commented Aug 5, 2021

patrickhlauke commented Aug 24, 2021 •

edited

Loading

mustaqahmed left a comment

mustaqahmed commented Feb 15, 2023

smaug---- commented Feb 21, 2023

mustaqahmed commented Feb 28, 2023

smaug---- commented Feb 28, 2023

mustaqahmed commented Feb 28, 2023

mustaqahmed commented Aug 23, 2023

Add note about rounding coordinates for click, auxclick, contextmenu #404

Add note about rounding coordinates for click, auxclick, contextmenu #404

Conversation

patrickhlauke commented Aug 4, 2021 • edited by pr-preview bot Loading

patrickhlauke commented Aug 4, 2021

patrickhlauke commented Aug 4, 2021 • edited Loading

patrickhlauke commented Aug 4, 2021

patrickhlauke commented Aug 4, 2021

patrickhlauke commented Aug 4, 2021 • edited Loading

patrickhlauke commented Aug 5, 2021

patrickhlauke commented Aug 5, 2021

mustaqahmed commented Aug 5, 2021 • edited Loading

patrickhlauke commented Aug 5, 2021

mustaqahmed commented Aug 5, 2021

patrickhlauke commented Aug 5, 2021

patrickhlauke commented Aug 24, 2021 • edited Loading

mustaqahmed left a comment

Choose a reason for hiding this comment

mustaqahmed commented Feb 15, 2023

smaug---- commented Feb 21, 2023

mustaqahmed commented Feb 28, 2023

smaug---- commented Feb 28, 2023

mustaqahmed commented Feb 28, 2023

mustaqahmed commented Aug 23, 2023

patrickhlauke commented Aug 4, 2021 •

edited by pr-preview bot

Loading

patrickhlauke commented Aug 4, 2021 •

edited

Loading

patrickhlauke commented Aug 4, 2021 •

edited

Loading

mustaqahmed commented Aug 5, 2021 •

edited

Loading

patrickhlauke commented Aug 24, 2021 •

edited

Loading