On 11 Jul 2007, at 17:12, Anthony Carlos wrote: > Perhaps we need to bundle "The Elements of RSpec Style" with the > documentation on the website. I'm happy to collect and edit > suggestions.
I would like to enthusiastically (albeit fundamentally unhelpfully) voice my support for this idea. The main win of RSpec over Test::Unit is the framework of linguistic constraints it provides to help guide people down the right path, but the main weakness is that it still doesn't go far enough -- most conversations I have with people about BDD are concerned with the continuing struggle to work out what the "right way" to write specs is. Toy examples (cf Stack, Account) are unhelpful in this regard because they demonstrate the common-sense mechanics of RSpec without really delivering any accumulated wisdom about how real specs should look and behave, and that's the hard part to grasp when you're trying to learn this stuff. I realise there's a well-intentioned progressive reticence to preach a dogmatic gospel about how specs should be named, designed and constructed, but as with Rails I believe that people derive genuine comfort and encouragement from being told The One Correct Way when they're starting out, and once they become competent enough to have dissenting opinions of their own they rarely seem to have a problem with enacting them. There are lots of gems of pragmatic wisdom dotted throughout the rspec-users archives -- yeah, David's "one WHEN/THEN per example" is a great one -- and I think it'd be tremendously useful to collect and promote these as Truth even if no expert quite believes that they are. Cheers, -Tom _______________________________________________ rspec-users mailing list rspec-users@rubyforge.org http://rubyforge.org/mailman/listinfo/rspec-users