Re: Language design values (Re: message primitive)

[Marius Vollmer <mvo at zagadka dot ping dot de>]

"Daschbach, John L" <John.Daschbach@pnl.gov> writes:

> A purely orthogonal set of functions would mean that given any
> function you could not duplicate it's functionality with any
> combination of the remaining functions.

Hmm, I only know orthogonality in the linear algebra sense, which you
seem to be using here too, but there it is really defined via the
inner product, not some via some kind of expressability notion.  That
is, sin and cos are orthogonal because the integral of sin*cos is zero
(over whole periods), yet sin(x) = cos(x-pi/2).

I think it makes sense to interpret `orthogonality' in computer
systems as `independence'.  I.e., two features are orthogonal if the
definition of one feature is independent from the definition of the
other.  Or put another way, the definition of one feature does not
change at all when you change the definition of the other.

- Marius