lucene-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Jason Rutherglen <jason.rutherg...@gmail.com>
Subject Re: Deprecating InstantiatedIndexWriter
Date Sat, 03 Oct 2009 19:34:57 GMT
Karl,

Sure no problems, LUCENE-1313 is using a RAMDir for now.

-J

On Sat, Oct 3, 2009 at 11:00 AM, Karl Wettin <karl.wettin@gmail.com> wrote:
> I suppose that should have been suggested before 2.9 rather than after...
>
> There are at least three reasons to why I want to do this:
>
> The code is based on the behaviour or the Directory IndexWriter as of 2.3
> and I have not been touching it since then. If there will be changes in the
> future one will have to keep IIW in sync, something that's easy to forget.
> There is no locking which will cause concurrent modification exceptions when
> accessing the index via searcher/reader while committing.
> It use the old token stream API so it has to be upgraded in case it should
> stay.
>
> The java- and package level docs have since it was committed been suggesting
> that one should consider using II as if it was immutable due to the
> locklessness. My suggestion is that we make it immutable for real.
>
> Since II is ment for small corpora there is very little time lost by using
> the constructor that builts the index from an IndexReader. I.e. rather than
> using InstantiatedIndexWriter one would have to use a Directory and an
> IndexWriter and then pass an IndexReader to a new InstantiatedIndex.
>
>
> Any objections?
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: java-dev-unsubscribe@lucene.apache.org
> For additional commands, e-mail: java-dev-help@lucene.apache.org
>
>

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


Mime
View raw message