6410 2006-01-07 06:00:54 -0800 Performance regression in text relayout due to bad BidiContext equality testing 2006-01-10 07:30:17 -0800 1 1 1 Unclassified WebKit Layout and Rendering 420+ Mac OS X 10.4 RESOLVED FIXED P2 Normal --- 1 mitz webkit-unassigned oldest_to_newest 27919 0 mitz 2006-01-07 06:00:54 -0800 The fix for bug 5172 added that for two line-ends to match (and thus for relayout to stop), they should also have the same bidi context. The bug is that the == operator for BidiContexts always returns false for contexts that don't have a parent, so in most cases, relayout will unnecessarily proceed through the last line. Fix coming up. 27921 1 5525 mitz 2006-01-07 06:12:33 -0800 Created attachment 5525 Fix the BidiContext == 27927 2 5525 darin 2006-01-07 08:45:15 -0800 Comment on attachment 5525 Fix the BidiContext == Good catch. r=me 27928 3 5525 darin 2006-01-07 08:46:45 -0800 Comment on attachment 5525 Fix the BidiContext == Another way to write this is: return c1.parent == c2.parent || (c1.parent && *c1.parent == *c2.parent); Not sure whether I like my version better or yours. 27930 4 5525 darin 2006-01-07 08:57:18 -0800 Comment on attachment 5525 Fix the BidiContext == Oops, my bad. The alternative version is: return c1.parent == c2.parent || (c1.parent && c2.parent && *c1.parent == *c2.parent); My previous suggestion was wrong and could lead to nil-deref. 28139 5 mitz 2006-01-10 07:30:17 -0800 Eric landed my asymmetric version :-) 5525 2006-01-07 06:12:33 -0800 2006-01-07 08:45:15 -0800 Fix the BidiContext == BidiContext operator==.patch text/plain 1698 mitz SW5kZXg6IFdlYkNvcmUvQ2hhbmdlTG9nCj09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT0KUkNTIGZpbGU6IC9jdnMvcm9vdC9X ZWJDb3JlL0NoYW5nZUxvZyx2CnJldHJpZXZpbmcgcmV2aXNpb24gMS45NwpkaWZmIC1wIC11IC1y MS45NyBXZWJDb3JlL0NoYW5nZUxvZwotLS0gV2ViQ29yZS9DaGFuZ2VMb2cJNyBKYW4gMjAwNiAx MDozMzoyNSAtMDAwMAkxLjk3CisrKyBXZWJDb3JlL0NoYW5nZUxvZwk3IEphbiAyMDA2IDE0OjE4 OjU5IC0wMDAwCkBAIC0xLDMgKzEsMTcgQEAKKzIwMDYtMDEtIyMgIE1pdHogUGV0dGVsICA8b3Bl bmRhcndpbi5vcmdAbWl0enBldHRlbC5jb20+CisKKyAgICAgICAgTm8gdGVzdHMgcG9zc2libGUs IG5vIGZ1bmN0aW9uYWxpdHkgY2hhbmdlcy4KKworICAgICAgICBSZXZpZXdlZCBieSBOT0JPRFkg KE9PUFMhKS4KKyAgICAgICAgCisgICAgICAgIC0gZml4IGh0dHA6Ly9idWd6aWxsYS5vcGVuZGFy d2luLm9yZy9zaG93X2J1Zy5jZ2k/aWQ9NjQxMAorICAgICAgICAgIFBlcmZvcm1hbmNlIHJlZ3Jl c3Npb24gaW4gdGV4dCByZWxheW91dCBkdWUgdG8gYmFkIEJpZGlDb250ZXh0CisgICAgICAgICAg ZXF1YWxpdHkgdGVzdGluZworCisgICAgICAgICoga2h0bWwvcmVuZGVyaW5nL2JpZGkuY3BwOgor ICAgICAgICAoa2h0bWw6Om9wZXJhdG9yPT0pOiBSZXR1cm4gdHJ1ZSBpZiBCaWRpQ29udGV4dHMg bWF0Y2ggYW5kIGJvdGggZG9uJ3QKKyAgICAgICAgaGF2ZSBhIHBhcmVudC4KKwogMjAwNi0wMS0w NyAgQW5kZXJzIENhcmxzc29uICA8YW5kZXJzY2FAbWFjLmNvbT4KIAogICAgICAgICBSZXZpZXdl ZCBieSBNYWNpZWouCkluZGV4OiBXZWJDb3JlL2todG1sL3JlbmRlcmluZy9iaWRpLmNwcAo9PT09 PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09PT09 PT09PT09ClJDUyBmaWxlOiAvY3ZzL3Jvb3QvV2ViQ29yZS9raHRtbC9yZW5kZXJpbmcvYmlkaS5j cHAsdgpyZXRyaWV2aW5nIHJldmlzaW9uIDEuMTY0CmRpZmYgLXAgLXUgLXIxLjE2NCBXZWJDb3Jl L2todG1sL3JlbmRlcmluZy9iaWRpLmNwcAotLS0gV2ViQ29yZS9raHRtbC9yZW5kZXJpbmcvYmlk aS5jcHAJMTkgRGVjIDIwMDUgMTk6NTI6NDggLTAwMDAJMS4xNjQKKysrIFdlYkNvcmUva2h0bWwv cmVuZGVyaW5nL2JpZGkuY3BwCTcgSmFuIDIwMDYgMTA6MDM6MTQgLTAwMDAKQEAgLTIzNCw3ICsy MzQsOSBAQCBib29sIG9wZXJhdG9yPT0oY29uc3QgQmlkaUNvbnRleHQmIGMxLCBjCiAgICAgICAg IHJldHVybiB0cnVlOwogICAgIGlmIChjMS5sZXZlbCAhPSBjMi5sZXZlbCB8fCBjMS5vdmVycmlk ZSAhPSBjMi5vdmVycmlkZSB8fCBjMS5kaXIgIT0gYzIuZGlyIHx8IGMxLmJhc2ljRGlyICE9IGMy LmJhc2ljRGlyKQogICAgICAgICByZXR1cm4gZmFsc2U7Ci0gICAgcmV0dXJuIGMxLnBhcmVudCAm JiBjMi5wYXJlbnQgJiYgKmMxLnBhcmVudCA9PSAqYzIucGFyZW50OworICAgIGlmICghYzEucGFy ZW50KQorICAgICAgICByZXR1cm4gIWMyLnBhcmVudDsKKyAgICByZXR1cm4gYzIucGFyZW50ICYm ICpjMS5wYXJlbnQgPT0gKmMyLnBhcmVudDsKIH0KIAogaW5saW5lIGJvb2wgb3BlcmF0b3IhPShj b25zdCBCaWRpQ29udGV4dCYgYzEsIGNvbnN0IEJpZGlDb250ZXh0JiBjMikK