Re: usage of parallelreader

[Chuck Williams <ch...@manawiz.com>]

I also need the ability to achieve rapid updates.  ParallelReader is
attractive because my content naturally segments into a set of large and
small stored fields where the small fields need to be accessed quickly,
plus stable and mutable indexed fields where the mutable fields need to
be updated quickly.  Fortunately all the large fields are stable. 
Doug's comments in parallel reader seem to indicate it is designed to
solve this type of problem, but it is not clear how to perform the
update on only the mutable field index while keeping all indices
synchronized.

Is there a reason the update contrib cited below was not pursued; it
seems Otis had some interest in it initially?  It seems that integrating
this contrib with ParallelReader might be a good approach.  I'm going to
explore that and other approaches to this problem.  Any relevant
background or suggestions would be greatly appreciated.

Thanks,

Chuck

----- Original Message -----
*From:* "Daan de Wit" <d....@wis.nl>
*To:* java-user@lucene.apache.org
*Sent:* 12/15/2005 07:08:41 AM
*Subject:* usage of parallelreader


>Hi all,
>
> 
>
>Within our application it is possible for users to add reactions for
>files. It is a requirement that a search returns a file if the query
>matches the contents or a reaction. I think it would be best to use to
>different indexes, one for the file and another one for the reactions.
>Searching could then be done via ParallelReader. If I understand that
>class correctly I have to ensure that the documents in the two indexes
>stay in the same order. If I'm not mistaken, updating a document must
>thus be done by first deleting the to-update document from both indexes,
>and then adding it back to both indexes. If this is correct, then it
>would be much easier to implement this scenario using one index and
>maybe use the Document update contrib
><http://issues.apache.org/bugzilla/show_bug.cgi?id=34629>  by Nicolas
>Maisonneuve. 
>
>Could anyone please enlighten me?
>
> 
>
>Daan
>
>
>  
>


---------------------------------------------------------------------
To unsubscribe, e-mail: java-dev-unsubscribe@lucene.apache.org
For additional commands, e-mail: java-dev-help@lucene.apache.org

Re: usage of parallelreader

[Otis Gospodnetic <ot...@yahoo.com>]

moving this to java-user.

Hello,

Is anyone using ParallelReader the way Chuck is describing his scenario?  I remember looking at the ParallelReader a few months ago and asking myself the same question: how can I update only one of the parallel indices and still keep the documents in two indices in sync?

Thanks,
Otis

----- Original Message ----
From: Chuck Williams <ch...@manawiz.com>
To: java-dev@lucene.apache.org
Sent: Fri 16 Dec 2005 08:32:51 PM EST
Subject: Re: usage of parallelreader

I also need the ability to achieve rapid updates.  ParallelReader is
attractive because my content naturally segments into a set of large and
small stored fields where the small fields need to be accessed quickly,
plus stable and mutable indexed fields where the mutable fields need to
be updated quickly.  Fortunately all the large fields are stable. 
Doug's comments in parallel reader seem to indicate it is designed to
solve this type of problem, but it is not clear how to perform the
update on only the mutable field index while keeping all indices
synchronized.

Is there a reason the update contrib cited below was not pursued; it
seems Otis had some interest in it initially?  It seems that integrating
this contrib with ParallelReader might be a good approach.  I'm going to
explore that and other approaches to this problem.  Any relevant
background or suggestions would be greatly appreciated.

Thanks,

Chuck

----- Original Message -----
*From:* "Daan de Wit" <d....@wis.nl>
*To:* java-user@lucene.apache.org
*Sent:* 12/15/2005 07:08:41 AM
*Subject:* usage of parallelreader


>Hi all,
>
> 
>
>Within our application it is possible for users to add reactions for
>files. It is a requirement that a search returns a file if the query
>matches the contents or a reaction. I think it would be best to use to
>different indexes, one for the file and another one for the reactions.
>Searching could then be done via ParallelReader. If I understand that
>class correctly I have to ensure that the documents in the two indexes
>stay in the same order. If I'm not mistaken, updating a document must
>thus be done by first deleting the to-update document from both indexes,
>and then adding it back to both indexes. If this is correct, then it
>would be much easier to implement this scenario using one index and
>maybe use the Document update contrib
><http://issues.apache.org/bugzilla/show_bug.cgi?id=34629>  by Nicolas
>Maisonneuve. 
>
>Could anyone please enlighten me?





---------------------------------------------------------------------
To unsubscribe, e-mail: java-user-unsubscribe@lucene.apache.org
For additional commands, e-mail: java-user-help@lucene.apache.org