Side-effect free containers for E

[Jonathan S. Shapiro]

> My problem comes from what I think Dean said, which is that only one bit
> can be on in the data byte.  I.e. You can't have a key which provides the
> function of two facets in one key.

Regardless of who said it, I believe that this is wrong. You can have the
function of multiple facets in a single key by designing a union facet and
using that. This is in fact what you are doing in a KeyKOS domain
implementation in any case.

shap