lucene-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Michael McCandless <>
Subject Re: Lucene 3.0.3 Release Date
Date Fri, 29 Oct 2010 12:32:00 GMT
I think we could back port all of these.

But many of them are "minor"...

Some comments below:

On Fri, Oct 29, 2010 at 7:09 AM, Robert Muir <> wrote:
> On Fri, Oct 29, 2010 at 4:28 AM, Uwe Schindler <> wrote:
> I just need the information, if there is anything else that should go
> into 2.9.4/3.0.3.
> Here is a list of bugs fixed in 3.x that weren't backported... i didnt
> spend a lot of time on these. perhaps its easy to reopen all of them
> and set version to 3.0/2.9, and we can discuss on the the list/jira
> that way
> * LUCENE-2216: OpenBitSet.hashCode returned different hash codes for
>  sets that only differed by trailing zeros. (Dawid Weiss, yonik)
> * LUCENE-2235: Implement missing PerFieldAnalyzerWrapper.getOffsetGap().
>  (Javier Godoy via Uwe Schindler)
> * LUCENE-2249: ParallelMultiSearcher should shut down thread pool on
>  close.  (Martin Traverso via Uwe Schindler)

This seems more serious since it's a resource leak...

> * LUCENE-2273: FieldCacheImpl.getCacheEntries() used WeakHashMap
>  incorrectly and lead to ConcurrentModificationException.
>  (Uwe Schindler, Robert Muir)

Not sure if this is relevant to 3.0.x/2.9.x?

> * LUCENE-2328: Index files fsync tracking moved from
>  IndexWriter/IndexReader to Directory, and it no longer leaks memory.
>  (Earwin Burrfoot via Mike McCandless)

The mem leak here is miniscule and it was a big patch so I think we
shouldn't back port.

> * LUCENE-2365: IndexWriter.newestSegment (used normally for testing)
>  is fixed to return null if there are no segments.  (Karthick
>  Sankarachary via Mike McCandless)
> * LUCENE-2074(just the subtask): Reduce buffer size of lexer back to
> default on reset.
>  (Ruben Laguna, Shai Erera via Uwe Schindler)
> * LUCENE-2496: Don't throw NPE if IndexWriter is opened with CREATE on
>  a prior (corrupt) index missing its segments_N file.  (Mike
>  McCandless)
> * LUCENE-2513: when opening writable IndexReader on a not-current
>  commit, do not overwrite "future" commits.  (Mike McCandless)

We should backport this one -- admittedly a corner case, but, it can
remove commits from your index...

> * LUCENE-2533: fix FileSwitchDirectory.listAll to not return dups when
>  primary & secondary dirs share the same underlying directory.
>  (Michael McCandless)
> * LUCENE-2549: Fix TimeLimitingCollector#TimeExceededException to record
>  the absolute docid.  (Uwe Schindler)
> * LUCENE-2580: MultiPhraseQuery throws AIOOBE if number of positions
>  exceeds number of terms at one position (Jayendra Patil via Mike McCandless)
> * LUCENE-2616: FastVectorHighlighter: out of alignment when the first value is
>  empty in multiValued field (Koji Sekiguchi)
> * LUCENE-2272: Fix explain in PayloadNearQuery and also fix scoring
> issue (Peter Keegan via Grant Ingersoll)
> * LUCENE-2658: Exceptions while processing term vectors enabled for multiple
>  fields could lead to invalid ArrayIndexOutOfBoundsExceptions.
>  (Robert Muir, Mike McCandless)

Looks serious...

> contrib/CHANGES:
>  * LUCENE-2035: TokenSources.getTokenStream() does not assign
> positionIncrement.
>   (Christopher Morris via Mark Miller)
>  * LUCENE-2055 (doc fix only, recommend to use snowball): Deprecated
> RussianTokenizer, RussianStemmer, RussianStemFilter,
>   FrenchStemmer, FrenchStemFilter, DutchStemmer, and DutchStemFilter. For
>   these Analyzers, SnowballFilter is used instead (for Version > 3.0), as
>   the previous code did not always implement the Snowball algorithm correctly.
>   Additionally, for Version > 3.0, the Snowball stopword lists are used by
>   default.  (Robert Muir, Uwe Schindler, Simon Willnauer)
>  * LUCENE-2278: FastVectorHighlighter: Highlighted term is out of alignment
>   in multi-valued NOT_ANALYZED field. (Koji Sekiguchi)
>  * LUCENE-2284: MatchAllDocsQueryNode toString() created an invalid XML tag.
>   (Frank Wesemann via Robert Muir)
>  * LUCENE-2277: QueryNodeImpl threw ConcurrentModificationException on
>   add(List<QueryNode>). (Frank Wesemann via Robert Muir)
>  * LUCENE-2184: Fixed bug with handling best fit value when the proper
> best fit value is
>                not an indexed field.  Note, this change affects the APIs. (Grant
>  * LUCENE-2359: Fix bug in CartesianPolyFilterBuilder related to
> handling of behavior around
>                the 180th meridian (Grant Ingersoll)
> * LUCENE-2524: FastVectorHighlighter: use mod for getting colored tag.
>  (Koji Sekiguchi)
> * LUCENE-2653: ThaiWordFilter depends on the JRE having a Thai
> dictionary, which is not
>  always the case. If the dictionary is unavailable, the filter will now throw
>  UnsupportedOperationException in the constructor.  (Robert Muir)

To unsubscribe, e-mail:
For additional commands, e-mail:

View raw message