Algebraic types and subclassing

Jan Skibinski jans@numeric-quest.com
Thu, 12 Mar 1998 19:33:45 -0600 (CST)



On Thu, 12 Mar 1998, Nick Kallen wrote:

>     A theorem prover that could reason through the traits, pre and post
> conditions, and prove that all of these things are satisfied (at compile
> time) is probably a long way off. But having the run time environment
> continually check these propositions (at run time) and abort when they are
> false is not impractical, nor *that* big of a deal (I think), and could
> obviously facilitate debugging. 

	Assertions (preconditions, postconditions, invariants) could be
	selectively disabled per module -- as some people do in
	Eiffel for time-critical classes, after entering the final
	delivery stage of software production. 

	I found the concept of assertions very valuable not only for their
	primary purpose but also as a means to produce clear and
	concise documentation.


	Jan