JSC regresses from the Mozilla string representation of Error objects. Use document.write(Error("test")) to see the difference. It's tiny but it unnecessarily breaks the tests in 15.11.1.1.js and 15.11.4.4-1.js in ecma_3/Exceptions of the Mozilla test suite. Just backed out from the JSC change I had applied on the weekend: http://websvn.kde.org/trunk/KDE/kdelibs/kjs/error_object.cpp?rev=427879&r1=426754&r2=427879
confirming this one.
Created attachment 2832 [details] One-line fix in error_object.cpp; updated expected.html
Comment on attachment 2832 [details] One-line fix in error_object.cpp; updated expected.html Cancelling. Needs to fix layout tests.
Created attachment 2833 [details] Updated patch including fixed layout tests
Comment on attachment 2833 [details] Updated patch including fixed layout tests r=me
Carson, I'm trying to commit your fix but it looks like the patch program is getting confused. I suspect that lines in the updated test results that contain "-->" are to blame. Could you attach your updated test results? Thanks.
Created attachment 2869 [details] Updated expected.html Here it is.
Silly me, I could have just generated my own expected.html file. At any rate, I've committed this fix.