[jira] [Commented] (LUCENE-3017) FST should differentiate between final vs non-final stop nodes
[ https://issues.apache.org/jira/browse/LUCENE-3017?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13018819#comment-13018819 ] Michael McCandless commented on LUCENE-3017: I hear you :) I think Lucene's needs put "pressure" on the traditional FST bounds... so we need to stretch things a bit. > FST should differentiate between final vs non-final stop nodes > -- > > Key: LUCENE-3017 > URL: https://issues.apache.org/jira/browse/LUCENE-3017 > Project: Lucene - Java > Issue Type: Improvement >Reporter: Michael McCandless >Assignee: Michael McCandless >Priority: Minor > Fix For: 4.0 > > Attachments: LUCENE-3017.patch > > > I'm breaking out this one improvement from LUCENE-2948... > Currently, if a node has no outgoing edges (a "stop node") the FST > forcefully marks this as a final node, but it need not do this. Ie, > whether that node is final or not should be orthogonal to whether it > has arcs leaving or not. -- This message is automatically generated by JIRA. 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] (LUCENE-3017) FST should differentiate between final vs non-final stop nodes
[ https://issues.apache.org/jira/browse/LUCENE-3017?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13018511#comment-13018511 ] Dawid Weiss commented on LUCENE-3017: - I know. I was just pointing out the fact that it does get fairly complex, but I don't have any constructive ideas how to make it simpler, so I'll simply shut up now :) > FST should differentiate between final vs non-final stop nodes > -- > > Key: LUCENE-3017 > URL: https://issues.apache.org/jira/browse/LUCENE-3017 > Project: Lucene - Java > Issue Type: Improvement >Reporter: Michael McCandless >Assignee: Michael McCandless >Priority: Minor > Fix For: 4.0 > > Attachments: LUCENE-3017.patch > > > I'm breaking out this one improvement from LUCENE-2948... > Currently, if a node has no outgoing edges (a "stop node") the FST > forcefully marks this as a final node, but it need not do this. Ie, > whether that node is final or not should be orthogonal to whether it > has arcs leaving or not. -- This message is automatically generated by JIRA. 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] (LUCENE-3017) FST should differentiate between final vs non-final stop nodes
[ https://issues.apache.org/jira/browse/LUCENE-3017?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13018401#comment-13018401 ] Michael McCandless commented on LUCENE-3017: Well... for LUCENE-2948, I need this to handle term prefixes that are in the terms index and are also valid terms. For example, I could have term "foo", a prefix of many other terms ("foobar", "foobaz", etc), and so the path f-o-o is in the terms index (pointing to a block that has all these other terms), ie ending on a zero-arc node. If that ending zero-arc node is final, I know "foo" is a valid term and I must seek to the block to load it, but if it's not final, I know it cannot exist in the index, and I can fail-fast (return NOT_FOUND from seek("foo")). > FST should differentiate between final vs non-final stop nodes > -- > > Key: LUCENE-3017 > URL: https://issues.apache.org/jira/browse/LUCENE-3017 > Project: Lucene - Java > Issue Type: Improvement >Reporter: Michael McCandless >Assignee: Michael McCandless >Priority: Minor > Fix For: 4.0 > > Attachments: LUCENE-3017.patch > > > I'm breaking out this one improvement from LUCENE-2948... > Currently, if a node has no outgoing edges (a "stop node") the FST > forcefully marks this as a final node, but it need not do this. Ie, > whether that node is final or not should be orthogonal to whether it > has arcs leaving or not. -- This message is automatically generated by JIRA. 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] (LUCENE-3017) FST should differentiate between final vs non-final stop nodes
[ https://issues.apache.org/jira/browse/LUCENE-3017?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13018282#comment-13018282 ] Dawid Weiss commented on LUCENE-3017: - Ehm... an automaton with zero-arc nodes that are not final is no no longer an automaton, but a graph of some sort... I mean -- what is the interpretation of an empty non-final node? "There is no sequence in the input that corresponds to this path and it is a prefix of some path in the input that you can't get from this automaton", right? This slowly becomes very confusing... the patch looks all right, but I'm thinking if the API overall is still clear. > FST should differentiate between final vs non-final stop nodes > -- > > Key: LUCENE-3017 > URL: https://issues.apache.org/jira/browse/LUCENE-3017 > Project: Lucene - Java > Issue Type: Improvement >Reporter: Michael McCandless >Assignee: Michael McCandless >Priority: Minor > Fix For: 4.0 > > Attachments: LUCENE-3017.patch > > > I'm breaking out this one improvement from LUCENE-2948... > Currently, if a node has no outgoing edges (a "stop node") the FST > forcefully marks this as a final node, but it need not do this. Ie, > whether that node is final or not should be orthogonal to whether it > has arcs leaving or not. -- This message is automatically generated by JIRA. 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