[Koha-devel] Koha Z39.50 client process discussion

[Frederic Demians]

Koha Z39.50 client process isn't easy to understand from a end-user
perspective.

Current process:

◊ In cataloguing, a search to multiple z3950 target is sent.
◊ Result is displayed. 20 first results are displayed for each target.
◊ Each record is stored in import_* tables for further usage via
'reservoir'.
◊ By clicking on Import, one record is copied in data entry page.
◊ Cataloger can modify biblio record, localized it, etc.
◊ In next cataloguing operation, a search will be done both in
catalog and reservoir (import_biblios table), and will retrieve
eventually biblio records from previous z3950 searches.

Some design choices are disputable:

◊ Why just 20 records per host are displayed?
◊ Why all retrieved records are stored in reservoir? People generally
don't understand that. Why 20 and not 5 or 100 records? When a
search is done by author, it is to catalog a specific book from
this author. But with the current design, the library will fill
its reservoir with biblio records describing books which will
generally never enter into its catalog.
◊ Search in reservoir is done with BreedingSearch function only on
'title' and 'isbn' fields. So it's impossible to retrieve a biblio
record by author.
◊ There is no user interface allowing to see what's going on with
biblio records stored in reservoir and coming from z3950 searches.
There is such an interface for staged records: those imported
explicitly into reservoir from files (in Tools).
◊ So if a library uses to do search in 4 z3950 hosts, each search may
add 4 x 20 = 60 records in import_biblios and import_records
tables. The number of records added this way will grow indefinitely
without any mean to clean reservoir.

Some ideas of improvement:

◊ Add a syspref 'ReservoirZ3950' which when at false disable biblio
records adding from z3950 searches.
◊ Add an UI in Tools to see the number of biblio records in reservoir
added from z3950 searches.
◊ Add a 'clear' option in Tool to delete all records in reservoir
coming from z3950, and eventually also coming from file imports.

Any comment/idea?

Also here:

http://bugs.koha.org/cgi-bin/bugzilla3/show_bug.cgi?id=3069
--
Frédéric