RE: proposed additions for discovery, sorting, and typed values

A few comments on your proposal, Jim ...

Schema Discovery:

The text describes a property for discovering which operators are supported.
There's a particular case in which I'm interested here. Suppose a server
offers searching on every resource -- making it easy for the client to
perform searches, by essentially making every URI available as a search
arbiter. In this case clients don't have to go find an arbiter -- they can
whatever URI is at hand. Would the queryschema property have to be defined
on each on of those resources? That seems like it be potentially a lot of
data stored on each resource just to support the search method. A PROPFIND
response that returned all properties might be very large when doen over a
container (even if depth=1).

My feeling if such a property would work best only if there where a small
number of arbiters. One way to get around this and yet have discoverability
for all resources is to use the DASL: part of the OPTIONS response; it could
point the client to a resource that has the relevant queryschema property.
This would allow the information to exist for any resource without the
burden of actually defining the property on each searchable resource. 

Typing:

We may want to leverage typing work that is already being done in XML. For
example, XML-Data defines a number of types and a mechanism for doing typing
in XML:

	http://www.w3.org/TR/1998/NOTE-XML-data/

Not all of this document deals with typing but there is a nice section on
it. These types are very oriented to the kinds of searches we are talking
about -- they basically match simple data types we are all familiar with. 

Thanks,
-Saveen

-----Original Message-----
From: Jim Davis [mailto:jdavis@parc.xerox.com]
Sent: Wednesday, March 11, 1998 10:02 PM
To: www-webdav-dasl@w3.org
Subject: proposed additions for discovery, sorting, and typed values


DASL needs methods for sorting results and for schema discovery.  It also
needs three valued logic and explicit typing.

I have proposals for how to do all these.

sorting is obvious, we add a sortby tag to the simplesearch.  The only
tricky issues are 1) ensuring that one can also sort by 'relevance' as most
full text search engines do, and 2) allowing one to sort up or down, and 3)
finding a way to talk about differences in sorting order among various char
sets.  My current proposal only addresses the first issue

schema discovery means the ability to determine, for a given server, not
only what query syntax it supports, but what properties are searchable,
sortable, etc.   I suggest the way to do this is to say that for each
searchable resource (an arbiter, as the current draft calls it), for each
query syntax it supports, it define a property whose value is the (query
syntax specific) schema.  For the simple search, this would list the
searchable properties, available operators, etc.

I have a detailed proposal for a set of tags to do this for simple search.

We need data types so we can tell the difference between the string "7" and
the number 7.  (Perhaps Alan B will send some email explaining more on this
point.)

We need three valued logic so we can search for e.g. resources where some
properties might be undefined, and so we can handle things like divide by
zero with clarity.  This means we also need constants for the true, false,
and unknown values.

Rather than send my specific proposal as a huge email file I have placed it
on the Web, in both Word and plain text.  The plain text was generated from
the Word and is not as nicely formatted but hopefuly conveys the ideas.

Perhaps Saveen will be kind enough to pick up my Word file (that is, if he
agrees with my proposals) and merge them into the official draft proposal.

I should admit that the current proposal shows the syntax but does not
explain or justify it very well.  If you've worked with lots of search
systems than it may make sense, but otherwise I'll have to add more
language to it.

I just wanted to get it out now so there could be some discussion on the
basic framework.  If there is agreement then we can pursue the details of
syntax further.

Please see

http://www.parc.xerox.com/istl/members/jdavis/dasl-plus.doc

http://www.parc.xerox.com/istl/members/jdavis/dasl-plus.txt

Best regards

Jim


------------------------------------
http://www.parc.xerox.com/jdavis/
650-812-4301

Received on Monday, 16 March 1998 14:15:26 UTC