Bug 33211
| Summary: | [chromium] layout-related assertion in SelectionController is firing — fix in Chromium WebKit or maybe in WebCore | ||
|---|---|---|---|
| Product: | WebKit | Reporter: | anton muhin <antonm> |
| Component: | WebKit Misc. | Assignee: | Nobody <webkit-unassigned> |
| Status: | RESOLVED WONTFIX | ||
| Severity: | Normal | CC: | darin, dglazkov, dimich, schenney |
| Priority: | P2 | ||
| Version: | 528+ (Nightly build) | ||
| Hardware: | All | ||
| OS: | All | ||
anton muhin
After http://trac.webkit.org/changeset/52778 Chromium's RenderViewTest.ImeComposition unit test started to fail on Windows and Mac (it's disabled on Linux).
See http://code.google.com/p/chromium/issues/detail?id=31617 for more details.
| Attachments | ||
|---|---|---|
| Add attachment proposed patch, testcase, etc. |
Darin Adler
Anton explained that the assertion happens with this backtrace:
WebCore::SelectionController::recomputeCaretRect Line 936
WebCore::SelectionController::absoluteCaretBounds Line 906
WebKit::WebViewImpl::queryCompositionStatus Line 1085
Based on that it seems clear that the WebViewImpl::queryCompositionStatus function should trigger layout. The best function to call is probably Document::updateLayout.
Alternatively we could decide that the SelectionController::absoluteCaretBounds function is designed to be called by high level code and is itself responsible for triggering layout. In that case we have to look at all the absoluteCaretBounds call sites and make sure it's OK to trigger layout at each of them.