simple search api (was Re: mimetype standardisation by testsets)

Bastian, Waldo waldo.bastian at intel.com
Sun Dec 17 03:53:58 EET 2006


John,

See the reference below to DBUS sessions. Doesn't DBUS have the ability
to inform any client about connects and disconnects of other clients to
the bus?

Cheers,
Waldo
 
Intel Corporation - Channel Platform Solutions Group - Hillsboro, Oregon

>-----Original Message-----
>From: xdg-bounces at lists.freedesktop.org [mailto:xdg-
>bounces at lists.freedesktop.org] On Behalf Of Jamie McCracken
>Sent: Friday, December 15, 2006 6:12 AM
>To: Mikkel Kamstrup Erlandsen
>Cc: xdg at lists.freedesktop.org
>Subject: Re: simple search api (was Re: mimetype standardisation by
>testsets)
>
>Mikkel Kamstrup Erlandsen wrote:
>
>>
>> Question 1 : Will it benefit the search engine to have a Session
object
>> for each connection? Then Query objects are spawned by a call like
>> Magnus suggest; Query = NewQuery(Session, query_string)? Is it
correct
>> that applications doesn't need to care about sessions - just gimme
the
>> goddam query! ? :-)
>
>In Dbus , there are no specific sessions exposed as such so Im not sure
>a "session" makes sense unless you are using P2P mode.
>
>We do need to close a live query to free up resources on the server but
>theres no way to automate that as AFAIK the server does not receive
>client disconnect signals - the shared bus is the connection and thats
>never released.
>
>>
>> Question 2 : Should the results be returned with the HitsAdded
signal?
>> The Query object then has a Query.GetResults method to retrieve the
>> results. This is closer to libbeagle and spotlight and the
application
>> only spends time retrieving hits when it really wants to. It does
>> introduce some extra method calls though...
>>
>
>Dunno. It depends on the extra traffic it generates - individual dbus
>calls may have an overhead of a ms or two so they could add up when
>returning large result sets (IE  1000 hits could cost you 1 or 2
seconds
>in socket latency alone). I think sending one hit per signal is
probably
>out of the question.
>
>If we batched up the results in packets of 10 or 20 (or some config
>setting) then it might be okay to do that.
>
>I would have to experiment with tracker to find the optimal packet size
>so maybe the packet size should be search engine specific.
>
>
>--
>Mr Jamie McCracken
>http://jamiemcc.livejournal.com/
>
>_______________________________________________
>xdg mailing list
>xdg at lists.freedesktop.org
>http://lists.freedesktop.org/mailman/listinfo/xdg



More information about the xdg mailing list