CSSFontFaceSet.clear() should not clear CSS-connected members
Created attachment 436709 [details] Patch
Created attachment 436710 [details] Patch
Comment on attachment 436710 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=436710&action=review > Source/WebCore/ChangeLog:11 > + Test: fast/text/FontFaceSet-clear-css-connected.html Can this be a WPT?
Comment on attachment 436710 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=436710&action=review >> Source/WebCore/ChangeLog:11 >> + Test: fast/text/FontFaceSet-clear-css-connected.html > > Can this be a WPT? https://github.com/web-platform-tests/wpt/pull/30254
Comment on attachment 436710 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=436710&action=review > Source/WebCore/css/FontFaceSet.cpp:137 > + m_backing->remove(m_backing.get()[facesPartitionIndex]); Is this high-performance enough? Might be worth thinking that through.
Comment on attachment 436710 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=436710&action=review >> Source/WebCore/css/FontFaceSet.cpp:137 >> + m_backing->remove(m_backing.get()[facesPartitionIndex]); > > Is this high-performance enough? Might be worth thinking that through. I've never actually seen any website ever call this function, so I think sharing code and guaranteeing that all the necessary work in remove() runs is probably more valuable than squeezing more performance out of this function.
Comment on attachment 436710 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=436710&action=review >>> Source/WebCore/css/FontFaceSet.cpp:137 >>> + m_backing->remove(m_backing.get()[facesPartitionIndex]); >> >> Is this high-performance enough? Might be worth thinking that through. > > I've never actually seen any website ever call this function, so I think sharing code and guaranteeing that all the necessary work in remove() runs is probably more valuable than squeezing more performance out of this function. I guess I can at least iterate backwards, though.
Committed r281842 (241175@main): <https://commits.webkit.org/241175@main>
<rdar://problem/82617907>