Limiting search to a subset of documents is normally down via query
filters.
From my point of view have several nutch beans will slow dwn things,
since you have to open several index searcher, detailer, tcp ip
connections etc.
I'm actually not sure if it is possible to have multiple bean instances.
Stefan
Am 08.01.2006 um 16:30 schrieb Raghavendra Prabhu:
I would like to explain the benefits of pooling -
Let it say we try to have a categorisation.
Say that nutch needs to search different set of indexes and the
user will be
able to select the categories
In this scenario the nutchbeans instantiated will be different
For ex folderone has some segments
foldertwo has some segments
The user should be able to search either folderone or foldertwo or
both of
them
If he selects folderone -a nutchbean will be created
if he selects foldertwo - a nutchbean will be created
In this case the same bean cannot be made persistent as the beans will
different
So by implementing pooling for nutchbean we can scale nutch to a
higher
level which can support the above combination .
In the above case,different users select different nutchbeans
(according to
what they want to search)
We can provide a solution to the above problem later on in case we
implement
pooling .
Please let me know if what i am telling is correct.If i am wrong
please feel
free to correct me.
Rgds
Raghavendra Prabhu
On 1/8/06, Stefan Groschupf <[EMAIL PROTECTED]> wrote:
If it is cached in the servlet context than the bean is shared
between all users, isn't it?
Am 08.01.2006 um 07:35 schrieb Raghavendra Prabhu:
I know that it is cached in the servlet context
But the mechanism of pooling will help nutchbean be shared across
different
users.
Whereas right now it will be persistent across a single user .
So people accessing from different machines will use the same
nutchbean and
again check it back into the pool
So in a way -it provides persistence across different users
Rgds
Prabhu
On 1/6/06, Stefan Groschupf <[EMAIL PROTECTED] > wrote:
NutchBean is cached in the servlet context. So I guess it is not
recreated for each request.
Stefan
Am 05.01.2006 um 22:38 schrieb Raghavendra Prabhu:
No i dont think so
What i am suggesting is we have nutch beans instantiated and we
store it .
Whenever an user comes and searches ,he will be given a
NutchBean .
After he searches he returns it to the pool and during the same
time when
some one searches he would get the same bean (note new bean is not
created )
Only if a bean is not available ,does a new bean get created .
This makes it faster as different users share the same NutchBean
and it does
not create a new nutchbean
Note:NutchBean is shared across different users whereas right now
it is only
for a single user and garabage collected
Here we control the NutchBean instantiation and we have to come up
with a
way to free it .
On 1/6/06, Byron Miller <[EMAIL PROTECTED]> wrote:
If i'm not mistaken doesn't the opensearch servlet get
around this issue? You could then post process the xml
through a stylesheet/css or your favorite scripting
language.
-byron
--- Raghavendra Prabhu < [EMAIL PROTECTED]> wrote:
Right now
Whenever an user comes and searches ,a NutchBean is
created
We should have a mechanism where this nutchbean is
pooled .I mean is created
and stored so that it can be given to the user
Immediately after the user has used the Nutch Bean
,he returns it back
(example at orkut ,we get a message saying doughnut
not available)
This will make search result faster and more
efficient
Only when paraller users are there will nutchbeans
get created
Any comments on the above issue
Rgds
Prabhu
---------------------------------------------------------------
company: http://www.media-style.com
forum: http://www.text-mining.org
blog: http://www.find23.net
---------------------------------------------------------------
company: http://www.media-style.com
forum: http://www.text-mining.org
blog: http://www.find23.net
