[Yazlist] Performance tweaks for async. search?

Mike Taylor mike at indexdata.com
Thu Nov 29 18:00:29 CET 2007


Johannes Leveling writes:
 > >  > So, I want to start processing records at
 > >  > 17:29:35-28/11 (from the Zebra log), but records become available after
 > >  > 17:29:56-28/11 (21 seconds later).
 > >
 > > Er.  Now I'm lost.  You want to start processing the record
 > > before you've retrieved them?  Asyncronous event handling won't
 > > help you with that, you need a TARDIS!
 >
 > No, I want to start processing partial results.

Well, that sounds reasonable enough.

 > (I need asynchronous event handling to have non-blocking calls, so
 > I can retrieve and convert records in parallel - at least in
 > theory.)

Yes; and not just in theory.  IRSpy does do this (although its
definition of "processing" records is pretty simple-minded).

 > > Are you piggy-backing your searches so that the initial search
 > > response includes some of the matching records?
 >
 > Yes.

Well, good.

 > Aren't there some records available in the first response?

I would think so, yes.  If you're using ZOOM, a result-set object is
created at search time, and of course it's initially empty.  Once
you've got a response from the server, the hit-count and zero or more
records should be filled in.  You can check whether it's so using
$rs->size() and $rs->record_immediate().  Don't just use $rs->record()
as that will synchronously wait for the requested record if it's not
already available, whereas $rs->record_immediate() will return undef.

 _/|_	 ___________________________________________________________________
/o ) \/  Mike Taylor    <mike at indexdata.com>    http://www.miketaylor.org.uk
)_v__/\  "Bring out your dead ..." -- Monty Python and the Holy Grail.




More information about the Yazlist mailing list