On Tue, Mar 31, 2009 at 11:34 AM, Josh Chisholm <joshuachish...@gmail.com>wrote:
> We were using the following style of setting up a world: > > World do > def a_helper > ... > end > end > > I don't know where I came up with that. I now realise we should have > been returning a world class here, but the above style seems to be > supported. Is it? > The result of this block is nil, so that wouldn't work so well... I'll add a guard against this so you get a better error message. > > Anyway, we got away with it so far. And it still works after upgrading > cucumber to 0.2+, but when an exception is raised from a step that is > called by another step, we don't get a full backtrace through the step > definitions, but instead get this cryptic message: > > undefined method `exception=' for nil:NilClass (NoMethodError) > > Here's the full gist: > http://gist.github.com/87941 > > FWIW I made the weird message go away: > > http://github.com/joshski/cucumber/commit/27a6f718451a417e9d42f9713911e6712aa48e8b > That doesn't really solve the problem. It shouldn't be allowed to use the nil singleton as a World since lots of modules would get mixed into it and mess things up. > > I suspect this solution is wrong because we weren't doing the right > thing. Obviously we can just return a world class like the wiki > suggests. If returning nil from the World block is unsupported, maybe > it should be caught early? > Done: http://github.com/aslakhellesoy/cucumber/commit/6c7f039ac2874775558734108f40ed1c61e1466b Aslak > > Cheers, > > Josh > _______________________________________________ > rspec-users mailing list > rspec-users@rubyforge.org > http://rubyforge.org/mailman/listinfo/rspec-users >
_______________________________________________ rspec-users mailing list rspec-users@rubyforge.org http://rubyforge.org/mailman/listinfo/rspec-users
