Resources not always reclaimed in scorers after each search
---
Key: LUCENE-686
URL: http://issues.apache.org/jira/browse/LUCENE-686
Project: Lucene - Java
Issue Type: Bug
Com
[ http://issues.apache.org/jira/browse/LUCENE-686?page=all ]
Ning Li updated LUCENE-686:
---
Attachment: ScorerResourceGC.patch
A patch is attached:
- The patch is based on the lastest version from trunk.
- The patch includes a test called TestScorerResourceG
Hi, All,
I am an enthusiastic user of Lucene and it is very helpful to my
projects at hand. As probabilistic models such as BM25 are very popular
among research communities now, do you have any plan to incorporate some
of them in future Lucene release? I believe that will make Lucene even
more pop
Hi Jianhan,
I am not aware, however, of anyone working on a BM25 implementation.
We are a volunteer project, though, so we are always open to
contributions!
-Grant
On Oct 17, 2006, at 5:50 AM, J.Zhu wrote:
Hi, All,
I am an enthusiastic user of Lucene and it is very helpful to my
proje
Hi, Grant,
If I would like to contribute, what should I do? I am not a good Java
developer myself though. Can I work with someone also interested?
Jianhan
-Original Message-
From: Grant Ingersoll [mailto:[EMAIL PROTECTED]
Sent: 17 October 2006 11:56
To: java-dev@lucene.apache.org
Subje
As a member of a number of Lucene subprojects dev lists, I've been
comparing the way the Jira workflow is used on the different projects.
In particular, I've been noting the difference between the workflows
that Lucene Java and Hadoop use. Hadoop in particular has a state for
"patch submitted" whi
J.Zhu wrote:
If I would like to contribute, what should I do? I am not a good Java
developer myself though. Can I work with someone also interested?
In some of my group's usage of lucene over large document collections,
we have split the documents across several machines. This has lead to a
Hmmm ... do we want to change the default reply-to? I'd be +1 for that.
The stuff would appear on dev anyway, right, because the Jira events get
mailed there?
-Original Message-
From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Yonik
Seeley
Actually, it looks like only repli
Steven,
Thanks for this prod. I've been meaning to debrief the Lucene dev
group on a recent visit I made to IBM's Silicon Valley Labs where I
presented Lucene and met with their search gurus. A recurring theme
from them was the management of Java Lucene project, and what they
and we can
Looking into the payload patch is next on my list after I finish up
the benchmarking, but I know the payload patch is going to require
several sets of eyes and some discussion. I would like to see the
more flexible index format addressed and I am not sure about the
removal of Fieldable as
On 10/17/06, Steven Parkes <[EMAIL PROTECTED]> wrote:
Hmmm ... do we want to change the default reply-to? I'd be +1 for that.
The stuff would appear on dev anyway, right, because the Jira events get
mailed there?
It's being looked into already, but the dev list adds it's own
reply-to, so it mig
[
http://issues.apache.org/jira/browse/LUCENE-686?page=comments#action_12442943 ]
Paul Elschot commented on LUCENE-686:
-
Three things:
Is there an actual memory leak problem related to this?
In ReqExclScorer the two scorers can also be clos
Hi, Vic,
Unfortunately BM25 uses IDF as well so splitting documents across
machines will also affect it. How about storing these as global
statistical data for sharing the search on these machines?
The equation of BM25 is clearly stated in Robertson's paper "Simple,
proven approaches to text retr
: Actually, it looks like only replies going to [EMAIL PROTECTED] (the default)
: will be added as comments. Our setup has the reply-to set to
: [EMAIL PROTECTED] by default, so those probably won't be added.
Is this controlled on a Project basis? ... I tried making an issue in the
Solr project
[
http://issues.apache.org/jira/browse/LUCENE-686?page=comments#action_12442987 ]
Ning Li commented on LUCENE-686:
> Is there an actual memory leak problem related to this?
Right now no. For example, in FS based directories, the index inputs te
: "patch submitted" which, as it's used on Hadoop, seems to facilitate
: communication. State changes (open->patch submitted,
: patch-submitted->open) seem to help communications between contributors
: and reviewers. Looking at the Lucene Java Jira, sometimes "[patch]" is
: put at the beginning of
Steven Parkes wrote:
Is there sufficient interest to consider this for Lucene Java? (I'd
write "any interest", but since I'm interested, there's at least some.)
+1
If there is agreement, I can gladly change Lucene Java to use Hadoop's
Jira workflow.
The other thing I was thinking of was th
Yonik Seeley wrote:
It's being looked into already, but the dev list adds it's own
reply-to, so it might not be as simple as changing the reply-to in
outbound JIRA messages.
Right, that's the problem.
Doug
-
To unsubscribe, e
[
http://issues.apache.org/jira/browse/LUCENE-686?page=comments#action_12443014 ]
Paul Elschot commented on LUCENE-686:
-
In SpanTermQuery in the spans package, there is also a TermPositions that might
need to be closed.
This means that closin
Chris Hostetter wrote:
Is this controlled on a Project basis? ... I tried making an issue in the
Solr project (where i figured i'd be spaming less people) to experiement
with this and couldn't get any messages sent to [EMAIL PROTECTED] to show up in
the
test issue i created...
Jira's outgoing
Chris Hostetter wrote:
Perhaps we should have two new statuses: "rough patch available" and
"polished patch available" ?
The other thing i would like to see tracked better is the distinction
between issues that unit tests attached and issues that do not ... this is
somewhat orthoginal to patch a
Is it explicitly stated anywhere what the workflow is/should be?
-Grant
On Oct 17, 2006, at 1:51 PM, Doug Cutting wrote:
Steven Parkes wrote:
Is there sufficient interest to consider this for Lucene Java? (I'd
write "any interest", but since I'm interested, there's at least
some.)
+1
I
Chuck Williams wrote:
I think the last discussion ended with the main counter-argument being
lack of support by gjc. Current top of GJC News:
*June 6, 2006* RMS approved the plan to use the Eclipse compiler as
the new gcj front end. Work is being done on the |gcj-eclipse| branch;
it can alread
: But as long as there is no real a memory leak, what is the point of
: adding this close functionality?
I think the concern is not so much that Lucene core as is has any leaks --
but that subclasses of core implimentations have no mechanism for safely
managing resources. Custom Directory implim
Grant Ingersoll wrote:
Is it explicitly stated anywhere what the workflow is/should be?
I can't see the workflow when I'm not logged in. I think you might have
to be a Jira administrator to view workflows. Some information about
them is up at:
http://www.atlassian.com/software/jira/docs/v
Think it's up to us. Hadoop has customized their workflow.
It's not the easiest process in the world. I know it takes a certain
amount of effort to transition issues from one workflow to another, so
should we want to do this, we should try to not plan on lots of tweaks?
-Original Message-
On Tuesday 17 October 2006 20:32, Chris Hostetter wrote:
>
> : But as long as there is no real a memory leak, what is the point of
> : adding this close functionality?
>
> I think the concern is not so much that Lucene core as is has any leaks --
> but that subclasses of core implimentations have
Doug Cutting wrote:
I've attached an image of it to this message.
Well, that didn't work. Instead I added it to Hadoop's wiki:
http://wiki.apache.org/lucene-hadoop-data/attachments/JiraWorkflow/attachments/workflow.png
Doug
--
: When custom Scorers and/or Directories need a close method, it can
: also be provided by subclassing Scorer, IndexSearcher and Directory
: in the custom code.
: Not providing this close method in the Lucene core passes the message
: that a working implementation is possible without it.
it seems
: > Is this controlled on a Project basis? ... I tried making an issue in the
: > Solr project (where i figured i'd be spaming less people) to experiement
: > with this and couldn't get any messages sent to [EMAIL PROTECTED] to show
up in the
: > test issue i created...
:
: Jira's outgoing messag
Looking at some currently closed issues, it looks like it should be
possible to add comments and links to closed issues. It provides the
comment button anyway. I haven't tried to push it. We could test on the
test SOLR issue.
Looking at the Jira docs, it looks like you can configure closed issues
On Tuesday 17 October 2006 21:10, Chris Hostetter wrote:
>
> : When custom Scorers and/or Directories need a close method, it can
> : also be provided by subclassing Scorer, IndexSearcher and Directory
> : in the custom code.
> : Not providing this close method in the Lucene core passes the messag
Vic Bancroft wrote on 10/17/2006 02:44 AM:
> In some of my group's usage of lucene over large document collections,
> we have split the documents across several machines. This has lead to
> a concern of whether the inverse document frequency was appropriate,
> since the score seems to be dependant
It looks like there is per-project configuration:
http://www.atlassian.com/software/jira/docs/latest/issue_creation_email.
html. It also looks like it requires a mailbox per project, which [EMAIL
PROTECTED]
doesn't sound like.
-Original Message-
From: Chris Hostetter [mailto:[EMAIL PROTE
Steven Parkes wrote:
My main concern is that things get lost in lists that grow without
bound.
I've never been to concerned about the size of the open bug list. It
can be searched, if one wishes to find whether there's already an issue
before filing a new one. The lists I try to keep small
Steven Parkes wrote:
It looks like there is per-project configuration:
http://www.atlassian.com/software/jira/docs/latest/issue_creation_email.
html. It also looks like it requires a mailbox per project, which [EMAIL
PROTECTED]
doesn't sound like.
That half of things is working. If you reply
A comment from a committer or contributor should be sufficient to
explain why something has not been committed, fixed or whatever,
and what action might next be needed. The scarcest resource is
committers. So we want to be able to focus their activities.
"Patch Available" is a call to
[
http://issues.apache.org/jira/browse/LUCENE-365?page=comments#action_12443056 ]
Yonik Seeley commented on LUCENE-365:
-
Paul, would it be possible to get a version with a license granted to the ASF?
Also, a single patch file would be preferr
[
http://issues.apache.org/jira/browse/LUCENE-365?page=comments#action_12443056 ]
Yonik Seeley commented on LUCENE-365:
-
Paul, would it be possible to get a version with a license granted to the ASF?
Also, a single patch file would be preferr
[
http://issues.apache.org/jira/browse/LUCENE-544?page=comments#action_12443058 ]
Matt Ericson commented on LUCENE-544:
-
I have been working on this exact same problem
Have you created an tests for it?
I am attaching my version. My version
[ http://issues.apache.org/jira/browse/LUCENE-544?page=all ]
Matt Ericson updated LUCENE-544:
Attachment: QueryParserPatch
This is my version of the Query Parser that will allow the users to boost some
fields
> MultiFieldQueryParser field boost multipl
A new scorer that requires reclaiming resources could be used by many
other scorers such as boolean scorers and conjunction scorers. Then
those scorers should have a closing method and so do the ones use
those scorers... A general closing method would be better, wouldn't
it?
-
[ http://issues.apache.org/jira/browse/LUCENE-544?page=all ]
Otis Gospodnetic updated LUCENE-544:
Attachment: (was: QueryParserPatch)
> MultiFieldQueryParser field boost multiplier
>
>
>
[
http://issues.apache.org/jira/browse/LUCENE-365?page=comments#action_12443062 ]
Paul Elschot commented on LUCENE-365:
-
The ASF licence is in the sources, this is from before jira.
I'll add a patch with the ASF licence soon.
> [PATCH] Per
[
http://issues.apache.org/jira/browse/LUCENE-365?page=comments#action_12443062 ]
Paul Elschot commented on LUCENE-365:
-
The ASF licence is in the sources, this is from before jira.
I'll add a patch with the ASF licence soon.
> [PATCH] Per
[ http://issues.apache.org/jira/browse/LUCENE-544?page=all ]
Matt Ericson updated LUCENE-544:
Attachment: QueryParserPatch
My Version of the QueryParser that will allow you to boost your fields
This version used a Map to keep track of what field to boos
[ http://issues.apache.org/jira/browse/LUCENE-443?page=all ]
Yonik Seeley resolved LUCENE-443.
-
Fix Version/s: 2.1
Resolution: Fixed
Assignee: Yonik Seeley
Thanks! I just committed this.
> ConjunctionScorer tune-up
> ---
[ http://issues.apache.org/jira/browse/LUCENE-415?page=all ]
Yonik Seeley closed LUCENE-415.
---
Fix Version/s: 1.9
Resolution: Fixed
Closing (a jira bug prevented me from closing this earlier).
> Merge error during add to index (IndexOutOfBoundsExcep
The cases I'm thinking of are those issues, be they bug fixes or
critical functionality, that are submitted by non-contributors, people
that aren't going to do the coding themselves. I might be very
interested in some of those, probably the bugs in particular. I'm just
trying to figure out how I'll
[
http://issues.apache.org/jira/browse/LUCENE-365?page=comments#action_12443077 ]
Yonik Seeley commented on LUCENE-365:
-
> The ASF licence is in the sources
Yeah, I was just trying to be safe.
It seems like there might be a small distinction
[
http://issues.apache.org/jira/browse/LUCENE-365?page=comments#action_12443077 ]
Yonik Seeley commented on LUCENE-365:
-
> The ASF licence is in the sources
Yeah, I was just trying to be safe.
It seems like there might be a small distinction
Steven Parkes wrote:
It wasn't really about having a list of needs clarification. It was more
about bounding open. I suppose it's my product development side showing.
Generally we tried not to leave issues open indefinitely, for fear of
not getting back to a customer. Perhaps there's nothing comp
[ http://issues.apache.org/jira/browse/LUCENE-365?page=all ]
Paul Elschot updated LUCENE-365:
Attachment: DisjunctionSumScorer20061017.patch
This patch obsoletes DisjunctionSumScorer and ScorerDocQueue as attached
earlier.
All tests pass here.
TestDisj
[ http://issues.apache.org/jira/browse/LUCENE-365?page=all ]
Paul Elschot updated LUCENE-365:
Attachment: DisjunctionSumScorer20061017.patch
This patch obsoletes DisjunctionSumScorer and ScorerDocQueue as attached
earlier.
All tests pass here.
TestDisj
[ http://issues.apache.org/jira/browse/LUCENE-544?page=all ]
Otis Gospodnetic resolved LUCENE-544.
-
Resolution: Fixed
I decided to go with Matt's version - smaller change to the class + a unit
test. Thanks Matt!
Karl: if any functionality from you
Just because you've gotten back doesn't mean the issue is gone.
No, just clarifying whose court the ball is in. In the patch available
case, it's clear. Looking at the some long standing open issues,
sometimes discussion peters out and it's not clear whether submitter and
commenters are on
So is Hadoop's workflow acceptable to all?
+1
I can put the diagram up on Wiki if this is approved. Also, is there
a way to add links from JIRA?
-Grant
-
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands
On 10/17/06, Doug Cutting <[EMAIL PROTECTED]> wrote:
So is Hadoop's workflow acceptable to all?
+1 for Hadoop's JIRA workflow.
As far as how that workflow should be used, I think we need a wiki page.
-Yonik
-
To unsubscribe,
[
http://issues.apache.org/jira/browse/LUCENE-365?page=comments#action_12443135 ]
Yonik Seeley commented on LUCENE-365:
-
Thanks Paul,
This patch seemed to revert the following:
http://issues.apache.org/jira/secure/attachment/12324730/Disjunct
[
http://issues.apache.org/jira/browse/LUCENE-365?page=comments#action_12443135 ]
Yonik Seeley commented on LUCENE-365:
-
Thanks Paul,
This patch seemed to revert the following:
http://issues.apache.org/jira/secure/attachment/12324730/Disjunct
[ http://issues.apache.org/jira/browse/LUCENE-365?page=all ]
Yonik Seeley resolved LUCENE-365.
-
Fix Version/s: 2.1
Resolution: Fixed
Assignee: Yonik Seeley (was: Lucene Developers)
> [PATCH] Performance improvement to DisjunctionSumScore
: > : When custom Scorers and/or Directories need a close method, it can
: > : also be provided by subclassing Scorer, IndexSearcher and Directory
: > it seems like that would handicap adoption of new Queries/Directories ...
: > I don't know how many people would have been interested in
: > Consta
62 matches
Mail list logo