Thanks, I recorded this as <https://bugs.openjdk.java.net/browse/JDK-8055870> It seems to be related to the static local variable type inference logic that I'm responsible for. I'll look into this ASAP. We just pushed a backport of JDK 9 optimistic typing feature to 8u-dev, and local static type inference is part of that.
I'll also try to figure out internally who's resposible for the bugreport page and see to it that a "Nashorn" subcategory gets added. Attila. On Aug 25, 2014, at 6:26 AM, Tal Liron <[email protected]> wrote: > Sorry for using the mailing list, still unclear for me how to open Nashorn > bugs (the standard form in bugreport.java.com does not have a "nashorn" > subcategory). The bug is on trunk in > http://hg.openjdk.java.net/jdk8u/jdk8u-dev/nashorn. > > The following program will cause an exception when "isEmpty2" is called: > > Error: java.lang.ClassCastException: java.lang.String cannot be cast to > jdk.nashorn.internal.runtime.Undefined > > The only difference between isEmpty1 and isEmpty2 is the actual use of "k". > If "k" is not referenced, the exception happens. > > > function isEmpty1(v) { > for (var k in v) { > print(v[k]+'\n') > return false > } > return true > } > > function isEmpty2(v) { > for (var k in v) { > return false > } > return true > } > > x = {test: 'test'} > > isEmpty1(x) > isEmpty2(x) >
