lucene-java-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Nilesh Vijaywargiay <nilesh.vi...@gmail.com>
Subject Re: Update lucene index
Date Tue, 12 Oct 2010 18:49:45 GMT
Thanks Anshum.
Any one who used parallelReader, can yo provide me with a direction for this
particular issue? May be a good example using parallelReader to solve this
kind of problem would do.



On Mon, Oct 11, 2010 at 11:34 PM, Anshum <anshumg@gmail.com> wrote:

> ParallelReader though theoretically sounds useful, I doubt if how much the
> overhead of maintaining and synchronizing the document ids would be. I
> haven't used it so far, perhaps someone who's used the ParallelReader for
> such a purpose on production environment/scale may help you.
>
> --
> Anshum Gupta
> http://ai-cafe.blogspot.com
>
>
> On Tue, Oct 12, 2010 at 11:49 AM, Nilesh Vijaywargiay <
> nilesh.vijay@gmail.com> wrote:
>
> > Anshum,
> > Yea, I would like to do what you mentioned. The source won't be available
> > at
> > later time and also other reason I understand is that if I delete and add
> a
> > document, it would get a different ID and would return inconsistent
> results
> > later if I search index built on this documents. Does that make sense?
> >
> > What do you think about ParallelReader. Is it remotely helpful here?
> >
> > Nilesh
> >
> >
> > On Mon, Oct 11, 2010 at 11:13 PM, Anshum <anshumg@gmail.com> wrote:
> >
> > > So what you essentially want is still being able to update a document.
> To
> > > update a particular field in the lucene doc, you'd need to mark that
> > field
> > > as updated/deleted and create a new document with new field information
> > > (that somehow gets merged while searching or gets taken care of in a
> > > similar
> > > manner). This as of now is not possible, and even if you were to do it
> > > using
> > > some logical workaround, it'd be much better to take the approach of
> just
> > > adding another doc and deleting the current one completely (marking it
> > > deleted unless you need to optimize).
> > > So by the way, is there a reason for you wanting to do so? is it that
> you
> > > only index data coming from a stream and you don't have access to the
> > > original source at a later time?
> > >
> > > --
> > > Anshum Gupta
> > > http://ai-cafe.blogspot.com
> > >
> > >
> > > On Tue, Oct 12, 2010 at 11:35 AM, Nilesh Vijaywargiay <
> > > nilesh.vijay@gmail.com> wrote:
> > >
> > > > Hi Group,
> > > >
> > > > I understand that the process of updating a document in lucene index
> is
> > > to
> > > > delete the document and add it again. But I do not want to delete the
> > > > document.
> > > >
> > > > I was thinking of a approach where I can create a new index with only
> > the
> > > > fields I want to update. And then may be I can some how merge these
> two
> > > > indexes. I
> > > >
> > > > I read about multiple searcher but that applies where you have two
> > index
> > > > containing different documents. I think parallel reader might help me
> > > here
> > > > but was unable to understand clearly as I am very new to lucene.
> > > >
> > > > What do you think my approach should be?
> > > >
> > > > Nilesh
> > > >
> > >
> >
>

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message