DASL draft issue: identification of query grammars

Hi,

see the message below for a DASL issue we discussed privately before and for
which we'd like to see feedback from others.

The issue is that the current (expired) DASL draft uses URIs to identify
query grammars, however XML-based query grammars need to be identified by a
Qualified XML name (a pair of namespaceUri reference and local name).

Julian

> From: Jim Davis [mailto:jrd3@alum.mit.edu]
> Sent: Monday, November 19, 2001 5:45 AM
> To: Julian Reschke; Lisa Dusseault
> Cc: mcc@watson.ibm.com; Jim Whitehead
> Subject: RE: DASL spec issue: encoding of a query grammar
>
>
> Julian last wrote about this issue in mid-September.  If it has been
> solved, you can ignore this.  If it's still alive, we should try
> to solve it.
>
> Maybe it will help if I try to explain what the DASL authors were
> trying to
> accomplish.  if there is a better way to meet the same goal, that's fine.
>
> The problem we need to solve it having a globally unique way to name query
> grammars. our proposed solution is to use a URI as a name for the grammar.
> So, e.g. we want to say "this query uses the query grammar defined by the
> Library of Congress for complex boolean bibliographic queries" and "this
> server will accept queries in any of the following N grammars".  Query
> terms have meaning only with respect to some grammar,
>
> now, is the way we proposed to do this in DASL sufficient for this?  Note
> that DASL does not need to combine the grammar URI and the query term
> string to make a longer URI, so I do not think any parsing is required.
>
> If there is a better way to meet the same goals, I for one, do not object
> to using it.  If so, please say what this is, or (so far as I am
> concerned)
> just change the spec to use it.

In my eyes, the most elegant way would be just to copy the way the supported
REPORTs are reported (see deltaV spec). Basically, there'd be a computed
live property named DAV:supported-search-grammar-set like:

<supported-search-grammar-set xmlns="DAV:">
	<supported-search-grammar><basicsearch /></supported-search-grammar>
	<supported-search-grammar><syntax1
xmlns="http://foo.bar.com/"/></supported-search-grammar>
	<supported-search-grammar><syntax2
xmlns="http://akuma.com/"/></supported-search-grammar>
</supported-search-grammar-set>

Advantages:

- in line with other DAV standards
- solves discovery of grammar names properly
- doesn't require additional headers in OPTIONS

Disadvantages:

- will only work for servers that support PROPFIND

Now, is the latter really still a problem? The DASL draft contains a few
hints that SEARCH requests with non-XML content should be possible, yet it's
silent about how this should work.

Julian

Received on Friday, 14 December 2001 16:14:26 UTC