Bug 58334
Summary: | LayoutTestController.computedStyleIncludingVisitedInfo-based tests don't work in WK2 | ||
---|---|---|---|
Product: | WebKit | Reporter: | Brady Eidson <beidson> |
Component: | Tools / Tests | Assignee: | Nobody <webkit-unassigned> |
Status: | NEW | ||
Severity: | Normal | CC: | samuel.white, sam |
Priority: | P2 | ||
Version: | 528+ (Nightly build) | ||
Hardware: | All | ||
OS: | All |
Brady Eidson
LayoutTestController.computedStyleIncludingVisitedInfo-based tests don't work in WK2
All such tests use both computedStyleIncludingVisitedInfo and keepWebHistory, both of which *are* implemented in WKTR.
However, something must not be working right. Currently the 4 known failures due to this bug are:
fast/history/nested-visited-test.html
fast/history/self-is-visited.html
fast/history/sibling-visited-test.html
state-url-sets-links-visited.html
All of these visually pass in a WK2 browser.
The failures all involve the returned style for the inspected nodes not correctly showing their visitedness.
Two obvious theories:
1 - computedStyleIncludingVisitedInfo doesn't work right
2 - Visited links go through a round trip to the UIProcess and don't get back to the WebProcess in time, therefore we're being bitten by IPC asynchronicity
Attachments | ||
---|---|---|
Add attachment proposed patch, testcase, etc. |
Brady Eidson
Added a new failing test to the skipped list and reclassified the 3 others in r83595
Samuel White
platform/mac/accessibility/search-predicate.html was added to the skipped list in r103704. It fails because visited links are not properly reported as visited in WK2 and do not report the appropriate style change.
Brady Eidson
(In reply to comment #2)
> platform/mac/accessibility/search-predicate.html was added to the skipped list in r103704. It fails because visited links are not properly reported as visited in WK2 and do not report the appropriate style change.
To the end user visited links work fine in WebKit 2. See my original hypothesis from 2011-4-12:
>1 - computedStyleIncludingVisitedInfo doesn't work right
>2 - Visited links go through a round trip to the UIProcess and don't get back to the WebProcess in time, therefore we're being bitten by IPC asynchronicity