Dan Book wrote: >Sorry if I missed it, but is there any reason not to only allow all >attributes before the signature (as presumably it originally was)? That is indeed the obvious and simple solution, and it is still the only solution we've seen that's entirely satisfactory in how it treats attributes semantically. The reason why we're not jumping straight to this solution is that many people disliked that syntax the first time round and insisted on changing it in the manner that turned out to introduce this bug. There's a forlorn hope that someone might have a miraculous idea that fixes the semantics without changing the syntax back. Failing that, it remains to be seen to what extent knowledge of this bug and related issues changes the views of people who called for signatures to precede attributes the first time. -zeframThread Previous | Thread Next