From: Hugh Sasse Staff Elec Eng <hgs@...>
Date: 2001-07-12T19:36:28+09:00
Subject: [ruby-talk:17708] Re: Old chestnut: invariants, pre/post conditions and so on

On Thu, 12 Jul 2001, Andrew Hunt wrote:
 
> I started on an implementation of Design by Contract for Ruby
> last year; it's listed in RAA (and is on the pragmaticprogrammer.com

I know of this one...

> web site).  It was just a proof of concept bit of playing around, but

which is why I was hoping to stimulate more interest or give you
other pointers to possible resources on this.

> it seemed to work mostly okay.

So it might be more stable than I thought.  I'll take a proper look
when I get a bit more time.
> 
> It's not "particularly difficult" to implement DBC for a dynamic
> language, but it is a bit different.  In the end, I think it's
> actually easier.

Then I did remember wrongly, or the twists in *my* brain denatured this!
> 
> Eiffel, for example, must know *all* of the static types and their
> relationships when compiling in order to get the DBC stuff right.
> In Ruby (and Smalltalk, etc.), we don't care about the types -- only
> the semantics of behavior.  If an object honors the pre/post/invariant,
> that's great: we don't care about it's "type", only its semantics.
> 
> In my twisted little brain, I think DBC is actually far more valuable
> in a dynamic language than in a static language. YMMV.

I think it would be valuable too, but cannot comment on the *relative* merits.

> 
> /\ndy
> 
	Hugh