New issue
Advanced search Search tips

Issue 766665 link

Starred by 2 users

Issue metadata

Status: Assigned
Owner:
Cc:
Components:
EstimatedDays: ----
NextAction: ----
OS: ----
Pri: 3
Type: Task

Blocking:
issue 645116



Sign in to add a comment

InputDeviceCapabilities is Chrome-only

Project Member Reported by rbyers@chromium.org, Sep 19 2017

Issue description

It's been two years since we first enabled the UIEvent.sourceCapabilities API ( issue 476530 ), and Chrome is still the only browser to ship it.

Usage is at ~0.04% of page views (low but non-trivial): https://www.chromestatus.com/metrics/feature/timeline/popularity/1098

Searching GitHub I do find some projects using it legitimately:
https://github.com/search?l=JavaScript&q=sourceCapabilities&type=Code&utf8=%E2%9C%93

We should consider if there's anything we could do to support adoption by other browsers, eg:
 - File bugs on the other browsers to track considering adding support
 - Add support for more of the use cases people felt were interesting
 - Once we have the automation support, write automated web-platform-tests
 - Of the hits in HTTP Archive, see if we can demonstrate better behavior on any sites in Chrome

This API is really only useful on browsers which support TouchEvents on desktop (i.e. not Safari).  I think Edge may have recently moved to a model where touch events are enabled on desktop if a touchscreen is attached, but perhaps this isn't in the main builds yet?

Of course it's possible that I was just wrong about the value of this API and we should remove it.  Of course _I_ don't think that's the case ;-)
 

Comment 1 by mustaq@chromium.org, Sep 25 2017

Cc: dtapu...@chromium.org
I think InputDeviceCapabilities.pointerMovementScrolls should be a stronger incentive: the stylus behavior is platform dependent and should be detectable from the web.  Do you think we should first ship .pointerMovementScrolls?

Comment 2 by rbyers@chromium.org, Sep 25 2017

Agreed that (issue 645116) probably has more cross-browser impact.  Perhaps we should poke the Edge team on that - see if they're planning on supporting it?
Owner: nzolghadr@chromium.org
Status: Assigned (was: Untriaged)
I'll do the follow up and update the issue.

Comment 4 by mustaq@chromium.org, Oct 13 2017

Blocking: 645116
Just remembered this discussion (about touch-action for pen vs touch), which suggests that .pointerMovementScrolls is not very useful:
  https://github.com/w3c/pointerevents/issues/203#issuecomment-299573524
I kind of agree with it after looking at the big picture.

So we shouldn't really invest in pointerMovementScrolls unless we fix this issue.

Looking into HTTP Archive as Rick suggested above seems to be the right first step to me.

Comment 5 by mustaq@chromium.org, Nov 24 2017

HTTP Archive data suggests that ~0.02% pages use this features.

[Detailed results] out of 16343951 pages in table har.2017_10_15_chrome_requests_bodies, we have:
- "sourceCapabilities" in 3454 pages,
- "firesTouchEvents" in 3445 pages, and
- "sourceCapabilities.firesTouchEvents" in 3305 pages.

However, chromestatus shows that page views for the feature has double to above 0.10+% in the past month: https://www.chromestatus.com/metrics/feature/timeline/popularity/1098


Comment 6 by mustaq@chromium.org, Nov 24 2017

Tried to see if there is a common library includes in the above pages that could be causing the recent spike in usage.  Here is the js names with frequency of occurrences (in a sample of 50 sites containing "sourceCapabilities"):
     22 this.js
     26 b.js
     27 min.js
     35 s.js
     40 virtualpage-canal-mobile-component.js
     47 e.js
     63 t.js
     84 component.js

Web/github search didn't point to any of them for "sourceCapabiilties".
Correcting my analysis on the last (deleted) post:

- Out of the 3.4k pages in HTTP Archive with 'sourceCapabilities', ~25% (~850) use hammer.js or hammer.min.js.  No other js ref really stand out.

- The recent increase in chromestatus caused by changes in metrics.  Apparently the new usage number (0.1% ) is more accurate.


The fact that ~0.02% of the pages getting ~0.1% usage suggests that some popular sites may be dominating the usage.  Navid, can we use UKM data to find them?

Sign in to add a comment