Re: storing survey answers of different data types

Marshall wrote:
... So in this case,
> assuming I adopt you desired goal, I'd see (once, you know, I
> actually have something to look at) ...

Okay Marshall, I'm starting to think I'll never make my point with A-algebra per se, but ignoring that, the simplest imperative language definition I can come up with is based on one or the other of these (the connectives are A-algebra ala what I think is the very respectable wording in http://www.dcs.warwick.ac.uk/~hugh/TTM/APPXA.pdf without the "<" and ">"):

i) A MINUS D is semantically equivalent to:

((A{h} AND (NOT D)) AND (A AND (NOT D)))

where h is any subset of A's heading.

So, when the join of two projections h1 and h2 of A equals A, A MINUS D is also equal to:

((A{h1} AND (NOT D)) AND (A{h2} AND (NOT D)))

or

ii) In an implementation language, the result of "Delete D from A", which might be defined by some BNF syntax or other, is semantically equivalent to the extension given by:

(A MINUS D) {heading of A}

is semantically equivalent to

((A{h} MINUS D) AND (A MINUS D)) {heading of A}

where MINUS is defined the same way TTM already defines it..

Not to argue against the lattice algebra (which I couldn't for several reasons, one being that I'm still not sure whether the current thinking on that uses an 'or' op that follows de Morgan), just wondering what you think, eg., basing a system on lattice algebra would need to make some formal connection such as one of the above two if an implementation were to be 'unambiguous' in the eyes of some. Received on Fri May 01 2009 - 05:29:02 CEST