Re: Optional Backpointers from Targets to References

ccjason@us.ibm.com wrote:

> I could imagine finding backpointers using a DASL like query might be
> very slow relative to an architected property/mechanism for
> maintaining backpointer info.

Not necessarily.  It's up to the server implementer: if the customers want
backpointer-like functionality, then the server can maintain backpointers
under the covers and use them to respond to the DASL query.  Other servers
which don't expect this to be a common request can just calculate the
response on the fly, and the client won't care which way it's working.  Just
like some servers will implement full-text search via an index, and others
will just grep the files.

I think I prefer the DASL approach; it's just as optional as the property
approach, but it behaves better in the downlevel server case.

> Perhaps you could explain why you say
> DASL is always more efficient for you.

He means that it's more efficient not to have a feature than to have a
feature that never gets used.

--
/====================================================================\
|John (Francis) Stracke    |My opinions are my own.|S/MIME supported |
|Software Retrophrenologist|=========================================|
|Netscape Comm. Corp.      | What do you mean, *you're*              |
|francis@netscape.com      |   a solipsist?                          |
\====================================================================/
New area code for work number: 650

Received on Thursday, 24 September 1998 13:11:16 UTC