[rspec-users] have_one and have_present

Jerry West jerry.west at ntlworld.com
Tue May 8 10:28:39 EDT 2007


I have to agree, Nick.  Specs should be about behaviour and not 
implementation, though the two will inevitably overlap.  Even specifying 
associations (another recent discussion) might be too much - what the 
client is really interested in is the behaviour associations offer (or 
rather the problem-specific solutions that behaviour allows you to 
offer).  So long as one regards such a spec as being shorthand for the 
end behaviour there should be no problem, but if we are too specific in 
our specifications we might end up dictating implementation choices 
which are neither necessary nor, perhaps desirable.

I like your valid/invalid syntax too.

Best wishes,
  Jerry

PS: "Independent" confirmation of an implementation detail ('did I type 
it right?') can be very helpful and I know I have used tests & specs for 
exactly that purpose, but in my case I suspect it's really a lack of 
confidence in my own abilities and understanding!  Is it really the role 
of a spec to double check that the coder has typed the right thing?


More information about the rspec-users mailing list