Michael Latta <[EMAIL PROTECTED]> writes: > David, > > It seems to me that the root of the problem is that the specification > is incorrect. Since Rails returns association proxies the > specification fails because it does not specify what the behavior > should be. I would suggest that instead of patching the change > matcher, that you should add a change_contents matcher that matches > the contents of a collection vs. the contents of a collection. That > way the framework is not guessing what was meant, but relying on the > specification to be correct. Since that is really what you want to > specify (the contents have changed). I think this is cleaner.
I think this is a leaky abstraction, and letting it leak out like that is worse than having to do a little framework mojo on our part. Pat p.s. does anyone else think that "magic" sounds deragotory while "mojo" sounds cool? I propose we say the rspec internals "got their mojo workin" rather than "magic" :) _______________________________________________ rspec-users mailing list [email protected] http://rubyforge.org/mailman/listinfo/rspec-users
