[jira] [Commented] (SOLR-5045) Pluggable Analytics
[ https://issues.apache.org/jira/browse/SOLR-5045?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13732091#comment-13732091 ] Andrew Muldowney commented on SOLR-5045: Do you mean 2894 distributed pivot facets or SOLR-3583 the pivot facet statistics patch? Pluggable Analytics --- Key: SOLR-5045 URL: https://issues.apache.org/jira/browse/SOLR-5045 Project: Solr Issue Type: New Feature Components: search Affects Versions: 5.0 Reporter: Joel Bernstein Priority: Minor Fix For: 5.0 Attachments: SOLR-5045.patch, SOLR-5045.patch This ticket provides a pluggable aggregation framework through the introduction of a new *Aggregator* interface and a new search component called the *AggregatorComponent*. The *Aggregator* interface extends the PostFilter interface providing methods that allow DelegatingCollectors to perform aggregation at collect time. Aggregators were designed to play nicely with the CollapsingQParserPlugin introduced in SOLR-5027. The *AggregatorComponent* manages the output and distributed merging of aggregate results. This ticket is an alternate design to SOLR-4465 which had the same basic idea but a very different implementation. This implementation resolves the caching issues in SOLR-4465 and combined with SOLR-5027 plays nicely with field collapsing. It is also much less intrusive on the core code as it's entirely implemented with plugins. Initial Syntax for the sample SumQParserPlugin Aggregator: ../select?q=\*:\*wt=xmlindent=truefq=\{!sum field=popularity id=mysum\}aggregate=true *fq=\{!sum field=popularity id=mysum\}* - Calls the SumQParserPlugin telling it to sum the field popularity. *aggregate=true* - turns on the AggregatorComponent The output contains a block that looks like this: {code:xml} lst name=aggregates lst name=mysum long name=sum85/long /lst /lst {code} -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
[jira] [Commented] (SOLR-5045) Pluggable Analytics
[ https://issues.apache.org/jira/browse/SOLR-5045?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13732097#comment-13732097 ] Joel Bernstein commented on SOLR-5045: -- Andrew, This ticket just provides a pluggable aggregation framework. It probably will include a basic SUM aggregator as a reference for how the framework works. Any other aggregation implementations will be added in other tickets. So pivot facets only relate in that they could one day be implemented as a pluggable aggregation in a future ticket. Pluggable Analytics --- Key: SOLR-5045 URL: https://issues.apache.org/jira/browse/SOLR-5045 Project: Solr Issue Type: New Feature Components: search Affects Versions: 5.0 Reporter: Joel Bernstein Priority: Minor Fix For: 5.0 Attachments: SOLR-5045.patch, SOLR-5045.patch This ticket provides a pluggable aggregation framework through the introduction of a new *Aggregator* interface and a new search component called the *AggregatorComponent*. The *Aggregator* interface extends the PostFilter interface providing methods that allow DelegatingCollectors to perform aggregation at collect time. Aggregators were designed to play nicely with the CollapsingQParserPlugin introduced in SOLR-5027. The *AggregatorComponent* manages the output and distributed merging of aggregate results. This ticket is an alternate design to SOLR-4465 which had the same basic idea but a very different implementation. This implementation resolves the caching issues in SOLR-4465 and combined with SOLR-5027 plays nicely with field collapsing. It is also much less intrusive on the core code as it's entirely implemented with plugins. Initial Syntax for the sample SumQParserPlugin Aggregator: ../select?q=\*:\*wt=xmlindent=truefq=\{!sum field=popularity id=mysum\}aggregate=true *fq=\{!sum field=popularity id=mysum\}* - Calls the SumQParserPlugin telling it to sum the field popularity. *aggregate=true* - turns on the AggregatorComponent The output contains a block that looks like this: {code:xml} lst name=aggregates lst name=mysum long name=sum85/long /lst /lst {code} -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
[jira] [Commented] (SOLR-5045) Pluggable Analytics
[ https://issues.apache.org/jira/browse/SOLR-5045?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13719512#comment-13719512 ] Otis Gospodnetic commented on SOLR-5045: [~joel.bernstein] how does this play with SOLR-2894? Overlap? Is the plan to be able to use this approach here to implement SOLR-2894 later on? Pluggable Analytics --- Key: SOLR-5045 URL: https://issues.apache.org/jira/browse/SOLR-5045 Project: Solr Issue Type: New Feature Components: search Affects Versions: 5.0 Reporter: Joel Bernstein Priority: Minor Fix For: 5.0 Attachments: SOLR-5045.patch, SOLR-5045.patch This ticket provides a pluggable aggregation framework through the introduction of a new *Aggregator* interface and a new search component called the *AggregatorComponent*. The *Aggregator* interface extends the PostFilter interface providing methods that allow DelegatingCollectors to perform aggregation at collect time. Aggregators were designed to play nicely with the CollapsingQParserPlugin introduced in SOLR-5027. The *AggregatorComponent* manages the output and distributed merging of aggregate results. This ticket is an alternate design to SOLR-4465 which had the same basic idea but a very different implementation. This implementation resolves the caching issues in SOLR-4465 and combined with SOLR-5027 plays nicely with field collapsing. It is also much less intrusive on the core code as it's entirely implemented with plugins. Initial Syntax for the sample SumQParserPlugin Aggregator: ../select?q=\*:\*wt=xmlindent=truefq=\{!sum field=popularity id=mysum\}aggregate=true *fq=\{!sum field=popularity id=mysum\}* - Calls the SumQParserPlugin telling it to sum the field popularity. *aggregate=true* - turns on the AggregatorComponent The output contains a block that looks like this: {code:xml} lst name=aggregates lst name=mysum long name=sum85/long /lst /lst {code} -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
[jira] [Commented] (SOLR-5045) Pluggable Analytics
[ https://issues.apache.org/jira/browse/SOLR-5045?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13719614#comment-13719614 ] Joel Bernstein commented on SOLR-5045: -- Yeah, the plan eventually would be to port the techniques used in SOLR-2894 to a pluggable Aggregator. Ideally pluggable analytics would lead to the implementation of different aggregation libraries. Since they can be implemented as pure plugins, developers wouldn't have to worry about getting their library committed. Interesting commercial opportunity for developing and maintaining a high performance analytic library for Solr, above and beyond what the community provides. Pluggable Analytics --- Key: SOLR-5045 URL: https://issues.apache.org/jira/browse/SOLR-5045 Project: Solr Issue Type: New Feature Components: search Affects Versions: 5.0 Reporter: Joel Bernstein Priority: Minor Fix For: 5.0 Attachments: SOLR-5045.patch, SOLR-5045.patch This ticket provides a pluggable aggregation framework through the introduction of a new *Aggregator* interface and a new search component called the *AggregatorComponent*. The *Aggregator* interface extends the PostFilter interface providing methods that allow DelegatingCollectors to perform aggregation at collect time. Aggregators were designed to play nicely with the CollapsingQParserPlugin introduced in SOLR-5027. The *AggregatorComponent* manages the output and distributed merging of aggregate results. This ticket is an alternate design to SOLR-4465 which had the same basic idea but a very different implementation. This implementation resolves the caching issues in SOLR-4465 and combined with SOLR-5027 plays nicely with field collapsing. It is also much less intrusive on the core code as it's entirely implemented with plugins. Initial Syntax for the sample SumQParserPlugin Aggregator: ../select?q=\*:\*wt=xmlindent=truefq=\{!sum field=popularity id=mysum\}aggregate=true *fq=\{!sum field=popularity id=mysum\}* - Calls the SumQParserPlugin telling it to sum the field popularity. *aggregate=true* - turns on the AggregatorComponent The output contains a block that looks like this: {code:xml} lst name=aggregates lst name=mysum long name=sum85/long /lst /lst {code} -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
[jira] [Commented] (SOLR-5045) Pluggable Analytics
[ https://issues.apache.org/jira/browse/SOLR-5045?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13712548#comment-13712548 ] Jack Krupansky commented on SOLR-5045: -- One interesting test case: There has been some interest in adding median to the stats component. The difficulty is that you need to build up the frequency distribution so that you can find the value that is = half of the values, which is a lot more effort than simply adding values to an accumulator. Pluggable Analytics --- Key: SOLR-5045 URL: https://issues.apache.org/jira/browse/SOLR-5045 Project: Solr Issue Type: New Feature Components: search Affects Versions: 5.0 Reporter: Joel Bernstein Priority: Minor Fix For: 5.0 Attachments: SOLR-5045.patch, SOLR-5045.patch This ticket provides a pluggable aggregation framework through the introduction of a new *Aggregator* interface and a new search component called the *AggregatorComponent*. The *Aggregator* interface extends the PostFilter interface providing methods that allow DelegatingCollectors to perform aggregation at collect time. Aggregators were designed to play nicely with the CollapsingQParserPlugin introduced in SOLR-5027. The *AggregatorComponent* manages the output and distributed merging of aggregate results. This ticket is an alternate design to SOLR-4465 which had the same basic idea but a very different implementation. This implementation resolves the caching issues in SOLR-4465 and combined with SOLR-5027 plays nicely with field collapsing. It is also much less intrusive on the core code as it's entirely implemented with plugins. Initial Syntax for the sample SumQParserPlugin Aggregator: ../select?q=\*:\*wt=xmlindent=truefq=\{!sum field=popularity id=mysum\}aggregate=true *fq=\{!sum field=popularity id=mysum\}* - Calls the SumQParserPlugin telling it to sum the field popularity. *aggregate=true* - turns on the AggregatorComponent The output contains a block that looks like this: lt;lst name=quot;aggregatesquot;gt; {lt;lst name=quot;mysumquot;gt;} lt;long name=quot;sumquot;gt;85lt;/longgt; lt;/lstgt; lt;/lst -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
[jira] [Commented] (SOLR-5045) Pluggable Analytics
[ https://issues.apache.org/jira/browse/SOLR-5045?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13712554#comment-13712554 ] Jack Krupansky commented on SOLR-5045: -- Can I script some custom analytics? Or is that simply a question of how this new component hooks in with the proposed JavaScriptRequestHandler (SOLR-5005)? Pluggable Analytics --- Key: SOLR-5045 URL: https://issues.apache.org/jira/browse/SOLR-5045 Project: Solr Issue Type: New Feature Components: search Affects Versions: 5.0 Reporter: Joel Bernstein Priority: Minor Fix For: 5.0 Attachments: SOLR-5045.patch, SOLR-5045.patch This ticket provides a pluggable aggregation framework through the introduction of a new *Aggregator* interface and a new search component called the *AggregatorComponent*. The *Aggregator* interface extends the PostFilter interface providing methods that allow DelegatingCollectors to perform aggregation at collect time. Aggregators were designed to play nicely with the CollapsingQParserPlugin introduced in SOLR-5027. The *AggregatorComponent* manages the output and distributed merging of aggregate results. This ticket is an alternate design to SOLR-4465 which had the same basic idea but a very different implementation. This implementation resolves the caching issues in SOLR-4465 and combined with SOLR-5027 plays nicely with field collapsing. It is also much less intrusive on the core code as it's entirely implemented with plugins. Initial Syntax for the sample SumQParserPlugin Aggregator: ../select?q=\*:\*wt=xmlindent=truefq=\{!sum field=popularity id=mysum\}aggregate=true *fq=\{!sum field=popularity id=mysum\}* - Calls the SumQParserPlugin telling it to sum the field popularity. *aggregate=true* - turns on the AggregatorComponent The output contains a block that looks like this: {code:xml} lt;lst name=quot;aggregatesquot;gt; lt;lst name=quot;mysumquot;gt; lt;long name=quot;sumquot;gt;85lt;/longgt; lt;/lstgt; lt;/lst {code} -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
[jira] [Commented] (SOLR-5045) Pluggable Analytics
[ https://issues.apache.org/jira/browse/SOLR-5045?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13712566#comment-13712566 ] Joel Bernstein commented on SOLR-5045: -- You have the flexibility to calculate median, atleast on a single server. Not sure what the best approach to this would be. Distributed median may be harder. You'd have to build up distributions in a way that can be merged. Scripting is a very cool thing. I need to do some research though on SOLR-5005 and see if can be applied. Pluggable Analytics --- Key: SOLR-5045 URL: https://issues.apache.org/jira/browse/SOLR-5045 Project: Solr Issue Type: New Feature Components: search Affects Versions: 5.0 Reporter: Joel Bernstein Priority: Minor Fix For: 5.0 Attachments: SOLR-5045.patch, SOLR-5045.patch This ticket provides a pluggable aggregation framework through the introduction of a new *Aggregator* interface and a new search component called the *AggregatorComponent*. The *Aggregator* interface extends the PostFilter interface providing methods that allow DelegatingCollectors to perform aggregation at collect time. Aggregators were designed to play nicely with the CollapsingQParserPlugin introduced in SOLR-5027. The *AggregatorComponent* manages the output and distributed merging of aggregate results. This ticket is an alternate design to SOLR-4465 which had the same basic idea but a very different implementation. This implementation resolves the caching issues in SOLR-4465 and combined with SOLR-5027 plays nicely with field collapsing. It is also much less intrusive on the core code as it's entirely implemented with plugins. Initial Syntax for the sample SumQParserPlugin Aggregator: ../select?q=\*:\*wt=xmlindent=truefq=\{!sum field=popularity id=mysum\}aggregate=true *fq=\{!sum field=popularity id=mysum\}* - Calls the SumQParserPlugin telling it to sum the field popularity. *aggregate=true* - turns on the AggregatorComponent The output contains a block that looks like this: {code:xml} lst name=aggregates lst name=mysum long name=sum85/long /lst /lst {code} -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org