Re: Academic name for associative array when used to pair column names with data

On Wednesday, March 19, 2014 6:08:48 AM UTC-7, william.d..._at_gmail.com wrote:
> So, there's no proper way to distinguish by nomenclature a tuple of just the data values from a tuple of the pairing of the attribute names with the data values?

"Relation", "tuple" and other terms have special relational meanings. I really don't have any better suggestion than what might be suggested to you by "relational tuple" vs "[math] ordered [n-] tuple". (Stroll wiki from "tuple".) Perhaps "rtuple" vs "otuple". (While you're at it, how about renaming the meaningless "data", no better than "flag" or "count".)

However:
1. Your code presumably appears in the context of a _specification_ of an _abstraction_, ie a reader/implementer presumably understands what abstraction is specified/implemented, and that that will involve relevant terminology. The abstraction is, relational relations. 2. Your code is _implementing_ an abstraction so its data structure's variables' types are _implementation_ types not _implemented_ types. So realize that you should seek a name appropriate to a variable mapping like or representing a relational (or ordered) tuple in a certain way, rather than to a variable that is a relational tuple. The latter would be for users of the abstraction.

It is far more important that one clarifies the abstraction (externally) and the correspondence between it and the implementation (internally) (both representation states and algorithm states) than care what one's temporary names are. If an implementer doesn't bother to define & name implementation-private abstractions/types themself then I would expect them to append the names they ought to have used for such types onto the variables that ought to have been declared to be of such types.

philip Received on Thu Mar 20 2014 - 01:58:39 CET