Please advise me on how my SRU server should handle multi-word queries.
Should I:

   * do nothing and just throw an error,
   * munge these queries into valid CQL on the client side, or
   * munge these queries into valid CQL on the server side

After reading and re-reading the CQL documentation, I appreciate the
expressiveness of the language. At the same time, I don't expect users
to take advantage of this expressiveness, even if they are provided
with a rich query-by-example interface such as a form with lots o'
boxes, buttons, and pull-down menus. Experience has taught me that most
people, even librarians, use simple one or two word queries and expect
the world. If I'm lucky people will send phrase searches by enclosing
them in quotes. If I'm really lucky people will denote an index to
search with queries like title=foo. If I'm really, really lucky users
will do Boolean logic and pattern matching. Whether we like it or not
people's expectations are being driven by Google.

If the above is more or less true, then I expect to get the following
sorts of queries as input from users, and the majority of the queries
will be like numbers 1, 2, 3, and 4:

   1. foo
   2. bar
   3. foo bar
   4. "foo bar"
   5. title=foo bar
   6. title="foo bar"
   7. title=foo and creator=bar
   8. title=foo and (creator=bar or creator=baz)
   9. title=foo and creator=ba*

Part of my question is, "Is Query #3, above, a valid CQL query?" I
think the answer is "Yes, and such a query is treated is a proximity
search." According to the end of section #2 of the Gentle Introduction
to CQL:

   In general, multi-word terms are interpreted as requesting
   records in which a single field contains all the specified
   words, in the specified order, with no other words in
   between. This is a proximity search. But see the section
   below on relations for exceptions.

(Even if Query #3 is not valid CQL, I think it should be, but that is
beside the point.)

Another question, "Are Queries #3 and #4 intended to be equivalent?"
Again, I think the answer is yes, but I am not able not put my finger
on any documentation explicitly stating this.

My SRU client interface WILL receive queries such as the following --
people WILL enter queries such as these:

   * foo bar
   * repetitive task
   * virtual libraries
   * International Technology Education Association

My interface needs to gracefully accept such queries, process them, and
return meaningful results; I do not intend to throw back to the user
errors such as "Bad syntax. Read the documentation and try again."

My Perl-based CQL parser (beautifully written by Ed S.) is heavily
based on the cql-java parser. In both cases, queries such as the ones
above output this error:

   unknown first class relation

Furthermore, a number of the test SRU servers also output errors of
various flavors for multi-word searches:

   Illegal or unsupported boolean found.
   unknown first-class relation
   Query syntax error

Try:


http://alert.ockham.org/sru-server.cgi?
operation=searchRetrieve&version=1.1&query=repetitive+task

http://alcme.oclc.org/srw/search/SOAR?
operation=searchRetrieve&version=1.1&query=repetitive+task

http://z3950.loc.gov:7090/voyager?
operation=searchRetrieve&version=1.1&query=repetitive+task

http://greta.pica.nl:1080/sru/?
operation=searchRetrieve&version=1.1&query=repetitive+task

On the other hand, this server returns no error but actually returns
results:


http://krait.kb.nl/cgi-zoek/sru.pl?
operation=searchRetrieve&version=1.1&query=repetitive+task

Again, is the following query valid CQL:

   repetitive task

If it is a valid query, then which one of the following alternatives
should be considered equivalent to the query above:

   1. "repetitive task"
   2. repetitive and task

If the query above is valid CQL, then I need to have my CQL parser
updated to not return errors. It and apparently the cql-java parser
currently do return errors.

If the query above is invalid, then I will need to modify my client to
accept such queries (because people WILL enter such queries), transform
them into alternatives (#1 or #2), and pass along an SRU URL to my
server for execution.

Whew! What do you think? Is the mulit-word query, repetitive task, a
valid CQL query?

--
Eric "Why Am I Working At This Time Of Year?!" Morgan
University Libraries of Notre Dame