Thanks. I'm not particularly interested in trying to run a malformed Test object "correctly," so failing usefully is definitely the goal. If you want to comment on the contentious one---which I'd left on Scott's plate since I know he has opinions on it---you can dig up the other thread, 'review request: JUnit "back door" closing;... that aims to "fail in a useful way" whenever we have something go drastically wrong during the servlet reportResultAndGetNextMethod bookkeeping, presumably for other causes since we've now fixed this one. Problem is, I couldn't find a graceful way to signal the failure from inside a server-side RPC with broken stuff around me, so as proposed it has a very ugly System.exit(). :-/
On Wed, Oct 22, 2008 at 9:16 AM, BobV <[EMAIL PROTECTED]> wrote: > +1 to at least failing in a useful way. > > LGTM. > > -- > Bob Vawter > Google Web Toolkit Team > --~--~---------~--~----~------------~-------~--~----~ http://groups.google.com/group/Google-Web-Toolkit-Contributors -~----------~----~----~----~------~----~------~--~---
