Stuart Ballard writes: > Andrew Haley <aph <at> redhat.com> writes: > > > > No, I don't mean that. I mean that if we change this behaviour today, > > we can't guarantee that someone won't chnage it back tomorrow. > > > > So, use this as a temporary hack if you must, but prepare to be > > disappointed. > > So are you saying you'd oppose the idea of adding a Mauve test to verify this > behavior, on the grounds that it's only for RI-compatibility rather than > spec-compatibility?
Yes. > There are a LOT of Mauve tests, as I understand it, which test > things that are completely unspecified other than by the RI's > behavior. Swing, as a whole, isn't even close to specified enough > to be independently implementable without that kind of testing. Sun > themselves have a WONTFIX bug in their BTS saying so, so I'm not > saying anything controversial here. Sure, but underspecified stuff is way different from this, which isn't underspecified at all. In the case of underspecified stuff we have to go to Sun's library to find out what is supposed to happen. This case is totally different: the test case is an illegal Java program. In no circumstances should Mauve contain illegal programs! > IMHO the way to go here would be to recognize that this behavior is > unspecified, but still put in a Mauve test for it so it doesn't > regress, and try to work with Sun (in the new era of cooperation we > now live in) to get it specified in future. It is fully specified now. Andrew.
