Re: "Common feature sets"

Graham writes:
> Quoting Ted:
> >Let me ask first whether you agree that the constant aspect might be
> >applied to a grouping below the set.  
> 
> Er... I'm not sure I understand the question.
> 
> I have been thinking that the term "constant" is probably unhelpful, but
> thus far have failed to find a better alternative.  I think the term
> "common" might be closer (as in "common factor" or "common base values").
> This, in turn, I think relates to Bill's "indirect reference".
> 
> The bit I really don't understand is "below the set".  This seems to
> suggest one of two things (a) to subsets of some set of interest (which
> set?), or (b) to the primitive elements from which the set is constructed
> (feature collections?  feature tags and values?).

If three represenations each require 5 features, three of which are constant
and the other two of which interdepend, how do we represent the three.  So
imagine the features are:

(feeble, greeble, beeble, habid, rabid)
(feeble, greeble, beeble, humid, rumid)
(feeble, greeble, beeble, zord, dord)

Can we make it something like:

2634062802=(feeble, greeble, beeble)
(2634062802, habid, rabid)
(2634062802, humid, rumid)
(2634062802, zord, dord)

Obviously, using 2634062802 rather than "feeble, greeble, beeble" isn't a
big win, but a cksum will be more effective when we have things like

u.http://www.microsoft.com/win-nt/features/svideo
u.http://www.netscape.com/communicator/features/open-gl-texturing
u.https://www.datefellows.com/watermarks/version4

as features.

I hope that is at least moderately clearer.
				best,
					Ted Hardie