One major reason is to update a field or rather shadow a field. i have a field named "testField" in index1 and i want to update that field. When I update, I want only the new value to be reflected, not the value in old field. now parallelreader starts from the latest index, i.e index2 and searches for 'testField'. It gets a hit in the index2 itself and doesn't go forward. Thus I am shadowing the old value of 'testField' with the new value. Does that make sense?
On Wed, Oct 27, 2010 at 9:38 PM, Pulkit Singhal <pulkitsing...@gmail.com>wrote: > But why do you feel the need to have a parallel reader that combines result > sets across two indices based on docId? > > On Thu, Oct 28, 2010 at 12:17 AM, Nilesh Vijaywargiay < > nilesh.vi...@gmail.com> wrote: > > > Pulkit, > > Parallel reader takes the union of all fields for a given id. Thus if I > > want > > to add a field or modify a field of a document which has id 2 in index1, > I > > need to createa a document with id 2 in index2 with the fields I want to > > add/modify. Thus parallel reader would treat them as fields of a single > > document. > > Now if I give doc.getFields() for that document then it would list fields > > from index1 and index2. > > > > On Wed, Oct 27, 2010 at 9:04 PM, Pulkit Singhal < > pulkitsing...@gmail.com > > >wrote: > > > > > Look interesting, what is the merit in having a second index in order > to > > > keep the document id the same? Perhaps I have misunderstood. Just want > to > > > understand your motivation here. > > > > > > On Wed, Oct 20, 2010 at 2:57 PM, Nilesh Vijaywargiay < > > > nilesh.vi...@gmail.com > > > > wrote: > > > > > > > I've written a blog regarding a work around for updating index in > > Lucene > > > > using parallel reader. It's explained with results and pictures. > > > > > > > > It would be great if you have a look at it. The link: > > > > http://the10minutes.blogspot.com/2010/10/lucene-index-update.html > > > > > > > > <http://the10minutes.blogspot.com/2010/10/lucene-index-update.html > > > >Thanks > > > > Nilesh > > > > > > > > > >
