Re: Implementing NodeList

Seems to be a common point of confusion; I think I was to blame for the
last discussion of this question.

The recommended solution for the "deep" NodeList implied by
getElementsByTagName seems to be to make nodes aware of when/how their
descendent subtree has changed, make this flavor of NodeList aware of when
something below the node it started from has changed, and cause the
NodeList to flush any previous work and recalculate itself at that time.
There may be some other optimizations possible.
______________________________________
Joe Kesselman  / IBM Research
Unless stated otherwise, all opinions are solely those of the author.

Received on Tuesday, 18 August 1998 14:51:28 UTC