[Chromium] Ensure layout has happened before calling into hasNonEmptyBoundingBox()
Created attachment 147929 [details] Patch
Comment on attachment 147929 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=147929&action=review > Source/WebKit/chromium/src/WebNode.cpp:218 > + m_private->document()->updateLayout(); most code in WebCore of this nature calls updateLayoutIgnorePendingStylesheets() - it deals with a m_hasNodesWithPlaceholderStyle case that seems important
Created attachment 147939 [details] Patch
(In reply to comment #2) > (From update of attachment 147929 [details]) > View in context: https://bugs.webkit.org/attachment.cgi?id=147929&action=review > > > Source/WebKit/chromium/src/WebNode.cpp:218 > > + m_private->document()->updateLayout(); > > most code in WebCore of this nature calls updateLayoutIgnorePendingStylesheets() - it deals with a m_hasNodesWithPlaceholderStyle case that seems important Ok, switched to using that method.
Comment on attachment 147939 [details] Patch Clearing flags on attachment: 147939 Committed r120637: <http://trac.webkit.org/changeset/120637>
All reviewed patches have been landed. Closing bug.
What are the side-effects of this change? Also I'm surprised by the lack of testing? is testing impossible/impractical?
(In reply to comment #7) > What are the side-effects of this change? Side-effects include correctness improvements for callers of the two affected methods, and presumably some minimal performance degradations to compensate. (The impact on performance appears not to be causing any issues for existing clients.) > Also I'm surprised by the lack of testing? is testing impossible/impractical? AFAIK, the Chromium WebKit API is generally not tested within the WebKit project. However, the correctness issue was caught by a Chromium test, so there is some test coverage.