- From: David Brownell <david-b@pacbell.net>
- Date: Wed, 01 Mar 2000 11:06:28 -0800
- To: Steve DeRose <Steven_DeRose@brown.edu>
- Cc: www-dom@w3.org
Steve DeRose wrote: > > At 11:13 AM -0000 3/1/00, Miles Sabin wrote: > > > > There're several other queries like this which could usefully > > be added and which might allow for similar optimizations: > > precedes in document order; is an ancestor of; least common > > ancestor; depth from root etc. > > Yup; there's a must-read paper in Digital Libraries 98 (+/- 1? sorry, don't > have the reference handy, authors were from Korea, though) on ways to > optimize the heck out of all these commonplace operations. Also some good > work in VLDB '99 (though the most relevant ones from there don't actually > mention XML in the titles). Got any suggestions about how to weigh the requirements for those (big/huge documents) apps against the other ones? Seems to me that if there's a real need for such methods it'd be appropriate to collect them into a new interface, which could optionally be implemented by DOM vendors who are courting applications working with that type of data. I'm not yet convinced that those operations are common enough to warrant more than another optional functionality module in DOM. Even though you called them "commonplace"; perhaps in your problem domain they are. - Dave
Received on Wednesday, 1 March 2000 14:06:24 UTC