Re: Issue Xpath-30: Reusing result sets

Possible alternative wording: "If a non-null result-set object is passed
in, the implementation may update its contents and return that object
rather than generating and returning a new instance."

"May" is our traditional flag that it's entirely up to the implementation
(as opposed to "should" or "must"), and this phrasing avoids any question
of whether this decision is good or bad, consistant or inconsistant, or
who/what makes the decision.

If we feel it'd be helpful to explain why someone might think this behavior
was good or bad, that could go in a second sentence... but this is starting
to get into our traditional descriptive-versus-discursive grey area; it's
not clear that whether design philosophy belongs in the spec or in the FAQ.

______________________________________
Joe Kesselman  / IBM Research

Received on Thursday, 28 March 2002 22:26:04 UTC