Bug 227191 - HitTestLocation should not use a fixed (1px, 1px) area for point based hittest
Summary: HitTestLocation should not use a fixed (1px, 1px) area for point based hittest
Status: RESOLVED INVALID
Alias: None
Product: WebKit
Classification: Unclassified
Component: Layout and Rendering (show other bugs)
Version: WebKit Nightly Build
Hardware: Unspecified Unspecified
: P2 Normal
Assignee: zalan
URL:
Keywords: InRadar
Depends on: 227185
Blocks:
  Show dependency tree / graph
 
Reported: 2021-06-19 21:21 PDT by zalan
Modified: 2022-06-29 14:23 PDT (History)
31 users (show)

See Also:

Attachments
Patch (36.53 KB, patch)
2021-06-19 21:34 PDT, zalan 		no flags Details | Formatted Diff | Diff
Patch (41.79 KB, patch)
2021-06-21 14:14 PDT, zalan 		no flags Details | Formatted Diff | Diff
Patch (43.21 KB, patch)
2021-06-22 12:57 PDT, zalan 		no flags Details | Formatted Diff | Diff
Show Obsolete (2) View All Add an attachment (proposed patch, testcase, etc.)

Note You need to log in before you can comment on or make changes to this bug.
Description zalan 2021-06-19 21:21:00 PDT 
at least it should be based on the device scale factor
Comment 1 zalan 2021-06-19 21:34:31 PDT 
Created attachment 431812 [details]
Patch
Comment 2 Sam Weinig 2021-06-20 13:16:05 PDT 
Comment on attachment 431812 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=431812&action=review

> Source/WebCore/ChangeLog:16
> +        Eventually the input to HitTestLocation::intersect should be a device pixel snapped rect similar to painting.

What makes doing this now difficult?
Comment 3 Sam Weinig 2021-06-20 13:16:46 PDT 
Comment on attachment 431812 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=431812&action=review

> Source/WebCore/ChangeLog:12
> +        This is an intermediate solution to support hit-test content on subpixel position.
> +        It fixes cases when we incorrectly find renderers on subpixel positions due to the integral inflate of the hit coverage area.
> +        e.g
> +          hit testing location is at [10, 10]
> +          renderer is at [10.6, 10] -> device pixel snapped to [11p, 10] on a 2x display

Seems like we should have some tests for this.
Comment 4 zalan 2021-06-21 14:14:26 PDT 
Created attachment 431907 [details]
Patch
Comment 5 zalan 2021-06-22 12:57:50 PDT 
Created attachment 431986 [details]
Patch
Comment 6 Radar WebKit Bug Importer 2021-06-26 21:21:19 PDT 
<rdar://problem/79824597>
Comment 7 Brent Fulgham 2022-06-29 14:23:22 PDT 
Alan ended up landing a more targeted fix. This change is no longer needed.