RE: draft-reschke-webdav-search-05 - a few questions on the draft

Hi,

you are of course right about why sending multiple requests and compiling a
single result is hard for a client. However for the same reasons, it isn't
simple for the server as well. The goal of the DAV:basicsearch grammar is to
provide a very small set of functionality that will be available across
*all* servers (therefore, support for it is REQUIRED). Thus, adding a
requirement that is unlikely to be implementable by everybody isn't going to
fly.

On the other hand, we *do* have the choice of *allowing* a server to support
multiple scopes. In this case we'd still need to define how this relates to
sorting, ranking and result truncation. So please make a proposal for
language that we could add to section 5.4 to define this.

Regards, Julian


--
<green/>bytes GmbH -- http://www.greenbytes.de -- tel:+492512807760

> -----Original Message-----
> From: www-webdav-dasl-request@w3.org
> [mailto:www-webdav-dasl-request@w3.org]On Behalf Of yamuna prakash
> Sent: Tuesday, September 30, 2003 5:24 PM
> To: www-webdav-dasl@w3.org
> Subject: RE: draft-reschke-webdav-search-05 - a few questions on the
> draft
>
>
>
> Another possible usecase is pagination! Consider I restrict the
> results to
> top 500 documents (say sorted by relevance). However I would paginate the
> results into say 10 pages (50 per page). I could issue the first
> search to
> identify the documents and issue another search to retrieve the
> data for the
> first 50 documents identified by the first search. I could then
> send this to
> the client while I continue building the remaining 9 pages
> asynchronously.
> The chances that end-users would click beyond the first page is very low.
> However I cannot preclude the possibility. The reasons for pagination are
> well established and doing it the way mentioned here I would think would
> improve the response time (although there are a lot of things that could
> prove me wrong especially the issuing multiple search requests).
>
> [Note: I am not saying DASL should support pagination but instead that it
> would seem to preclude the possibility]
>
> Coming back to the searching in different collections use-case, issuing
> multiple search requests I believe not only places a huge burden on the
> client but I am not sure I like the performance characteristics
> either. For
> example even considering that the client can build the
> consolidated search
> results, it seems to me to be accurate the client cannot make use of the
> max-results feature in a very efficient manner. Lets say we want
> to present
> the end-user the top 500 documents. If I have sort clause then it
> seems to
> me the max-results clause goes pretty much out of the window.
> (ex:consider
> sorting by date you need the most recently modified documents, in
> which case
> it not possible to apply max-results at all on the individual search
> requests.) If I have to retrieve the entire result sets and then
> build the
> consolidated resultset and apply max-results that to me would
> seem to have
> horrendous performance characteristics.
>
> [I am guessing Julian had all these issues in mind when she said its not
> trivial:)]
>
> prakash
>
> >From: "yamuna prakash" <yamunap@hotmail.com>
> >To: www-webdav-dasl@w3.org
> >Subject: RE: draft-reschke-webdav-search-05 - a few questions on
> the draft
> >Date: Tue, 30 Sep 2003 07:25:18 +0000
> >
> >
> >Actually I guess searching from the root might not be an option
> if I have
> >to exclude certain collections from the search... in which case it seems
> >like the client has to do a whole host of things if it is to present the
> >search results in a consolidate manner to the end-users.
> >
> >prakash
> >
> >>From: "yamuna prakash" <yamunap@hotmail.com>
> >>To: www-webdav-dasl@w3.org
> >>Subject: RE: draft-reschke-webdav-search-05 - a few questions
> on the draft
> >>Date: Tue, 30 Sep 2003 07:11:55 +0000
> >>
> >>
> >>Currently we have a use case where in we need to search under different
> >>collections and then build a consolidated search result and
> present it to
> >>the user. We would like to avoid searching from the root with the
> >>possibility of having to specify infinite depth, to ensure the
> search does
> >>not take too much time. As Elias mentioned in his email, this can be
> >>achieved by issuing multiple requests, but as Julian mentioned this of
> >>course opens us up to the issue of what having to manage top
> >>(max-results), sorting, etc i.e. in effect the client having to build a
> >>layer on top of DASL.
> >>
> >>
> >>I am quite new to DASL and WebDAV and in fact I am still
> finding my way so
> >>to speak.
> >>
> >>Given the above fact, my next question might sound incredibly stupid...
> >>but...
> >>
> >>if each collection has different properties ( I am not sure if this is
> >>possible) and if each collection acts as a node in a tree, I am
> curious as
> >>to how searching within multiple scopes would be different from
> searching
> >>in a tree with arbitrary depth?
> >>
> >>Is it the case that all collection nodes are consistent (i.e. have the
> >>same properties) and thus searching within multiple scopes is a
> different
> >>beast?
> >>
> >>
> >>thanks,
> >>prakash
> >>
> >>
> >>>From: "Julian Reschke" <julian.reschke@gmx.de>
> >>>To: "Wallmer, Martin" <Martin.Wallmer@softwareag.com>,"'yamuna
> prakash'"
> >>><yamunap@hotmail.com>, <julian.reschke@gmx.de>
> >>>CC: <www-webdav-dasl@w3.org>
> >>>Subject: RE: draft-reschke-webdav-search-05 - a few questions on the
> >>>draft
> >>>Date: Tue, 30 Sep 2003 08:28:46 +0200
> >>>
> >>>
> >>>Correct.
> >>>
> >>>Keep in mind that allowing multiple issues may be more complex than you
> >>>think. For instance, it's trivial to just repeat the same query for the
> >>>different scopes, but how does this affect sorting/ranking/max-results?
> >>>
> >>>Julian
> >>>
> >>>--
> >>><green/>bytes GmbH -- http://www.greenbytes.de -- tel:+492512807760
> >>>
> >>>
> >>
> >>_________________________________________________________________
> >>Get Married!  http://www.bharatmatrimony.com/cgi-bin/bmclicks1.cgi?74
> >>Search from 7 lakh Brides & Grooms.
> >>
> >
> >_________________________________________________________________
> >Reconnect with old pals. Relive past joys.
> >http://www.batchmates.com/msn.asp All it takes is a click!
> >
>
> _________________________________________________________________
> Reconnect with old pals. Relive past joys.
http://www.batchmates.com/msn.asp
All it takes is a click!

Received on Friday, 3 October 2003 07:37:57 UTC