Re: [boost] Re: Standard Matrix interfaces

From: Philipp K. Janert (janert_at_[hidden])
Date: 2001-08-28 17:21:59

(...)
> This is particularly true for most OO interfaces,
> since covariance ensures no purely abstract interface
> can exist in almost all real world cases.
> A 'purely' abstract interface is one in which
> no methods return 'significant' concrete types,
> nor accept them as arguments. The existence of
> such types in method signatures is the proof that
> the interface is just another representation.
(...)

Excuse my ignorance, but: What does the word 
"covariance" mean in OO or Computer Science? 
I only know it from Differential Geometry.

Best,

                Ph.