[rspec-devel] Are rspec code actually "driven" out from the spec incrementally?

Dan North dan at tastapod.com
Fri Jun 22 08:23:58 EDT 2007


David Chelimsky wrote:
> Sorry if that's not as exciting as you had imagined. I wonder how
> things would have been different if we had started with rspec
> self-hosting from day 1.
>   
I can confidently suggest that two things would have happened. Firstly 
you would have had to get to a minimum bootstrap level of functionality 
/somehow/, so you'd have used test/unit to get there initially. 
Secondly, you would have made it self-hosting /far too early/ and then 
whenever it broke - which would have been frequently - you would have 
been unable to diagnose the problem because /it wasn't running!/

At least that's what happened to me with jbehave :)

Also I met Ward Cunningham at a conference a couple of years ago and he 
told me the same thing happened to him at Microsoft with a well-known C# 
testing framework. So it seems there's no easy way round it.

With rbehave I took a different approach, which was to use rspec to 
develop it and use an example application (game of life) to eat my own 
dogfood. Much happier with that arrangement - rspec was working just 
fine all the way through!

> Cheers,
> David
>   
Cheers,
Dan
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://rubyforge.org/pipermail/rspec-devel/attachments/20070622/2fc3f3a6/attachment.html 


More information about the rspec-devel mailing list