Re: Inheritance/defaulting of attributes

Paul Prescod (papresco@technologist.com)
Tue, 07 Oct 1997 23:53:54 -0400

Tim Bray wrote:
> Documents are different. To start with, it's far from clear to me
> that the term "inheritance," standing by itself, has a useful
> semantic in the context of documents.

I agree. Inheritance in OOP only had meaning after it was defined.
Still, if we are talking about "element type inheritance" I think the
obvious meaning would be that element sub-types would "inherit"
"properties" (attributes and content model) from super-types.

> This has a practical consequence. We should, in future, struggle
> for greater precision; for example, however it is clear that
> however LOCATOR attributes depend on those in the parent EXTENDED
> linking element, it ain't inheritance in the O-O style. -Tim

Actually, in OOP that is called "acquisition". It is evolving a language
and understanding of its own. See:

Environmental Acquisition - A New Inheritance Mechanism

http://www.cs.technion.ac.il/~david/Papers/Tech_Reports/lpcr9507.PS.gz
in:
http://www.cs.technion.ac.il/~david/publications.html

So we are not in completely uncharted waters although there will
certainly be some differences.

Paul Prescod