On V8, JavaTypeString is now distinct from JavaTypeObject, but both need to represented by a JNI class description when forming a JNI method signature.
It looks like the culprit is http://trac.webkit.org/changeset/82194, which introduced JavaTypeString, but didn't update JavaMethodJobject::signature() to account for this.
Created attachment 96728 [details] Patch
Comment on attachment 96728 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=96728&action=review > Source/WebCore/ChangeLog:8 > + Covered by existing tests. I don't understand. Then, should this patch be changing expectations to pass an existing failing test? > Source/WebCore/bridge/jni/JavaMethodJobject.cpp:135 > + requiresClassName |= (type == JavaTypeString); Is this code perf sensitive? Unless the compiler is smarter than I think it is, it seems like we lost a shortcut for JavaTypeObject.
Created attachment 101469 [details] Patch
Committed r91362: <http://trac.webkit.org/changeset/91362>
This patch broke Mac builds: http://build.webkit.org/builders/SnowLeopard%20Intel%20Debug%20%28Build%29/builds/1387 http://build.webkit.org/builders/Leopard%20Intel%20Debug%20%28Build%29/builds/37433 Speculative build fix landed: http://trac.webkit.org/changeset/91366
I think that actually the break was due to http://trac.webkit.org/changeset/91355
(In reply to comment #7) > I think that actually the break was due to http://trac.webkit.org/changeset/91355 You're right. Commented on the right bug :(