[ https://issues.apache.org/jira/browse/SOLR-12638?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16573481#comment-16573481 ]
David Smiley commented on SOLR-12638: ------------------------------------- Yeah something like that. This feature you propose bares a lot of resemblance to "atomic updates" (not to be confused with _in-place_ updates – e.g. numeric DV updates). That code is largely handled by AtomicUpdateDocumentMerger called by DistributedUpdateProcessor. It seems to me this proposed feature would work with that same code in some way. For example, after potentially doing an atomic update (i.e. the document is partial), we then also potentially see if it's a nested doc by checking for \_ROOT\_ as you said. A tricky part may be considering the update log via RealtimeGetComponent (see DistributedUpdateProcessor.getUpdatedDocument). That API allows fetch-by-ID, and we can input the root document ID. Hopefully we should get back the whole child hierarchy... we may need to enhance RTG to ensure it does so when consulting the index (seems reasonable to me). That's where you might employ the child doc transformer. I believe the children will be attached to root docs in the updateLog, since they are "underneath"/attached in the serialization of each parent document. Note that an atomic update of a child document won't actually consider the update log since the update log is keyed by parent/root ID, not the ID of the child document. > Provide a simple API to update block (update child documents) > ------------------------------------------------------------- > > Key: SOLR-12638 > URL: https://issues.apache.org/jira/browse/SOLR-12638 > Project: Solr > Issue Type: Sub-task > Security Level: Public(Default Security Level. Issues are Public) > Reporter: mosh > Priority: Major > > I have been toying with the thought of using this transformer in conjunction > with NestedUpdateProcessor and AtomicUpdate to allow SOLR to completely > re-index the entire nested structure. This is just a thought, I am still > thinking about implementation details. Hopefully I will be able to post a > more concrete proposal soon. -- This message was sent by Atlassian JIRA (v7.6.3#76005) --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org