| Summary: | REGRESSION (r294160): Occasional infinite loops under updateLayersForInteractionRegions | ||||||
|---|---|---|---|---|---|---|---|
| Product: | WebKit | Reporter: | Tim Horton <thorton> | ||||
| Component: | New Bugs | Assignee: | Tim Horton <thorton> | ||||
| Status: | RESOLVED FIXED | ||||||
| Severity: | Normal | CC: | ap, cdumez, simon.fraser, webkit-bug-importer, wenson_hsieh | ||||
| Priority: | P2 | Keywords: | InRadar | ||||
| Version: | WebKit Nightly Build | ||||||
| Hardware: | Unspecified | ||||||
| OS: | Unspecified | ||||||
| Attachments: |
|
||||||
|
Description
Tim Horton
2022-05-18 16:27:25 PDT
Created attachment 459560 [details]
Patch
I will figure out the changelog shenanigans. Comment on attachment 459560 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=459560&action=review Seems reasonable to me! > Source/WebCore/platform/graphics/IntRectHash.h:49 > + static WebCore::IntRect emptyValue() { return { -2, -2, -2, -2 }; } Out of curiosity, is there a reason why we opt for this value? (CoreGraphics has a notion of `CGRectNull` that's something along the lines of { 0, 0, ∞, ∞ }, for instance) Comment on attachment 459560 [details]
Patch
(Chris, Tim and I discussed this on Slack and agreed to go with a separate hashtraits object that supports empty rects)
(In reply to Wenson Hsieh from comment #5) > Comment on attachment 459560 [details] > Patch > > (Chris, Tim and I discussed this on Slack and agreed to go with a separate > hashtraits object that supports empty rects) Rather, I am just going to change my code to not emit empty rects. Pull request: https://github.com/WebKit/WebKit/pull/761 Committed r294467 (250729@main): <https://commits.webkit.org/250729@main> Reviewed commits have been landed. Closing PR #761 and removing active labels. |