On Thu, May 16, 2002 at 02:38:57PM -0400, Ray Denenberg wrote:
> Ok, it's time we addressed Joe Zeeman's suggestion (see separately
> forwarded message).  Joe is suggestion that truncation be explicit in
> the query syntax.  How do we feel about that?
>
> So there are three possibilities:
> 1. Implicit truncation, as in the original rule where type 104 always
> applies.
> 2. Truncation defined as part of the index, as in the bath searches.
> 3. Explicit truncation, expressed in the query string.
>
> Note that if we go with either 1 or 3, we will have to abandon the idea
> of defining Bath searches.


Being in a different time zone, I am probably behind already on this one
(lots more email in this thread to read), so can I change topic slightly
and ask: is a goal of CQL to be very much like CCL (ISO8777) if only
because there is no reason to be different? By "very much like" I mean
if CCL defines the functionality, then use the CCL syntax (unless there
is a good reason not to).

If its to be like CCL, then truncation (using '?') is supported as a
syntax of the query language. So the CQL Bath profile would be *derived*
from the Z39.50 Bath profile. So you would define index names for
words in title, first words in title, etc. Each index for the bath
profile would list all attributes in Explain. But then CQL query operators
such as truncation and relation would *replace* any existing attribute
value for its type. (If an index did not define a value for that
type, then the new type/value pair would be *added* to the attribute
list.)

This means all indexes defined and 'explain'ed would have full attribute
lists. Using query operators (such as truncation - abc? - and relation)
would modify the attribute list of the index.

Alan