Re: phonetic

["Kurt D. Zeilenga" <Kurt@OpenLDAP.org>]

At 03:37 PM 7/24/00 -0400, Mark Adamson wrote:
>That's good info for 3 of the functions. I'm still curious what the
>"indexer" and "filter" functions do in this structure

These are optional routines used by to support backend-specific
search filter indexing subsystems.  Basically, the indexer generates
a set of keys to be associated with the stored values.  For each
unique key, the backend should store a list of entry IDs which
contain the value generating the key.  The filter generates keys
associated with the asserted value.  For each assertion, the
backend obtains a set of keys, obtains associated lists of entries,
and generates list of a common entries (that is, the lists are
combined through intersection).  For each filter component, lists
are then combined per filter semantics.

For phonetic matching, the stored values would likely be converted
to a canonical, phonetic representation and a single key derived
from this. The filter would likewise canonicalize the asserted
value and generate a single key.