We should update uniteIfNonZero() to use isZero() instead of a !width() && !height(). See: https://bugs.webkit.org/show_bug.cgi?id=76177#c3
Created attachment 123133 [details] Update uniteIfNonZero to use isZero
Created attachment 123135 [details] Update uniteIfNonZero to use isZero A note on why I added no tests: I did a little research on what this change will affect and I came up empty handed. I instrumented the code and ran all our tests and visited various popular/complex pages, neither of which registered a difference between the old behavior and the new. The primary use of uniteIfNonZero in regular code is in RenderInline.cpp which, as far as I can tell in my testing, only passes integer values through to this method and should not be affected by this change. I still think this is change will be clearly useful in preventing future bugs.
Comment on attachment 123135 [details] Update uniteIfNonZero to use isZero View in context: https://bugs.webkit.org/attachment.cgi?id=123135&action=review > Source/WebCore/ChangeLog:15 > + * rendering/svg/SVGRenderSupport.cpp: > + (WebCore::SVGRenderSupport::computeContainerBoundingBoxes): It would be helpful to explain this change in ChangeLog. Readers needn't spend effort trying to figure out how it's related to bug title when it's not!
Created attachment 123171 [details] Update uniteIfNonZero to use isZero Adding a comment in the ChangeLog so that readers are not confused by the unrelated change in SVGRenderSupport.cpp.
Comment on attachment 123171 [details] Update uniteIfNonZero to use isZero OK
Comment on attachment 123171 [details] Update uniteIfNonZero to use isZero Clearing flags on attachment: 123171 Committed r105643: <http://trac.webkit.org/changeset/105643>
All reviewed patches have been landed. Closing bug.