RE: An observation about "live" NodeLists

>Unfortunately code like this might well perform *very*
>badly for some implementations.

Unfortunately, _any_ deletion is going to have severe performance impacts
on some implementations of NodeList. Only the cleverest implementations are
going to even attempt to keep track of precisely which subtrees have
changed and minimize recalculation... and the work involved in tracking
that is going to impact performance and/or memory requirements elsewhere in
the implementation.

I've drawn up an essay summarizing the recent discussion and sent it to
Lauren, which is apparently the proper procedure for submitting
contributions to the FAQ -- she runs them past the WG for sanity-checking
before adding them to the document. Actually, this one's large enough that
it may go into the tutorial (there's a tutorial?) instead.

______________________________________
Joe Kesselman  / IBM Research
Unless stated otherwise, all opinions are solely those of the author.

Received on Monday, 19 October 1998 09:02:34 UTC