Hello Webkit-Team, the two-finger-scrolling-performance is very bad and jerky at following site: http://www.apple.com/ipad/app-store/ Hardware: MacBook 3,1 OS: 10.6.3 Tested with: Safari 4.0.5 and Nightly Build r56912 Thanks in advance. Mehmet
<rdar://problem/7820819>
Created attachment 52466 [details] Patch
Comment on attachment 52466 [details] Patch > + // Check to see if the image changed; we have to do this because the call to > + // CGImageCreateCopyWithColorSpace() below can create a new image every time. > + if (m_uncorrectedContentsImage && CFEqual(m_uncorrectedContentsImage.get(), newImage)) > + return; Wow, does CFEqual actually work on CGImages!? And it compares every bit? > + m_uncorrectedContentsImage = m_pendingContentsImage = newImage; Could you do these on separate lines? It's too easy to overlook the second assignment when it's all on one line. r=me assuming you tested
(In reply to comment #3) > (From update of attachment 52466 [details]) > > + // Check to see if the image changed; we have to do this because the call to > > + // CGImageCreateCopyWithColorSpace() below can create a new image every time. > > + if (m_uncorrectedContentsImage && CFEqual(m_uncorrectedContentsImage.get(), newImage)) > > + return; > > Wow, does CFEqual actually work on CGImages!? And it compares every bit? I don't care about bit-for-bit comparisons; this just checks to see if this image is the same as the last one we set. I guess I could test for pointer equality? > > + m_uncorrectedContentsImage = m_pendingContentsImage = newImage; > > Could you do these on separate lines? It's too easy to overlook the second > assignment when it's all on one line. Will do.
http://trac.webkit.org/changeset/57039