- From: Rob Bull <bull@crxnet.com>
- Date: Thu, 22 Jun 2000 09:44:48 +0000 (GMT)
- To: Kevin Gamiel <kgamiel@islandedge.com>
- cc: Robert Waldstein <wald@LIBRARY.HO.LUCENT.COM>, www-zig@w3.org
Hi, > > > > Just to weigh in (gee, thought I had switched to being a lurker) I strongly > > disagree with the 2 points above (agreed with the rest). > > explain PDUs (actually PDUs/BER in general) have very little overhead. > > I have always found that impressive. > > I read that to mean overhead in the sense of complexity, not bytes > (Rob?). Actually, I like ber, too, but new technologies like xml do the > same thing, though in bloated form, and are more readily acceptable to > other developers. That counts for something. - this is what I was referring to - its not so much the ber aspect, its just that we reckon on a few percent of the explain record syntax information being the most useful, and the approach is that the server deliver this as one packet of data that can be used by the client. This then removes the concept of the otherwise multistage approach of a client having to do an init - explain-search(es)- present(s) - to get the same information. There is an interesting trade-off however of a very big server (many databases etc) which could result in a larger explain-lite piece of XML, against the full explain where the client can "probe" for just what it wants. In essence though, this is going to be tested in the ONE-2 project, and it may or may not work - the project will certainly report such findings to the ZIG. Rob
Received on Thursday, 22 June 2000 04:50:03 UTC