[
https://issues.apache.org/jira/browse/SOLR-12697?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17343492#comment-17343492
]
Christine Poerschke commented on SOLR-12697:
--------------------------------------------
Thanks [~TomGilke] for opening the PR! I've left initial comments on it.
Having re-read the notes on this issue and researched and learnt more re: the
various implementation possibilities I think the {{DocValues.getNumeric}}
approach (like you have it in the PR) is a good one _and_ there might also be
some additional functionality that could be supported (outside the scope of
this JIRA ticket) via the other implementation possibilities if there were use
cases for it.
----
[~slivotov] and [~erickerickson]'s comments on 18/May/19 are related to use of
the SolrDocumentFetcher implementation possibility and how it has a document
cache but that
{quote}... All fieldValue features are by the flow isolated from each other, so
there is no possibility to gather all required fields and then fetch them from
the stored document fields. ...
{quote}
So I was curious about how one might connect them and conceptually something
like
{code:java}
[ {
"name": "a", "class":
"org.apache.solr.ltr.feature.PrefetchingFieldValueFeature",
"params": { "field" : "aaa", "prefetch_fields" : [ "bbb", "ccc" ] }
}, {
"name": "b", "class":
"org.apache.solr.ltr.feature.PrefetchingFieldValueFeature",
"params": { "field" : "bbb", "prefetch_fields" : [ "aaa", "ccc" ] }
}, {
"name": "c", "class":
"org.apache.solr.ltr.feature.PrefetchingFieldValueFeature",
"params": { "field" : "ccc", "prefetch_fields" : [ "aaa", "bbb" ] }
} ]
}
{code}
might be a way i.e. a new feature that uses the SolrDocumentFetcher
implementation always and via the {{prefetch_fields}} there is a possibility to
gather fields.
----
[~kaessmann] on 07/May/21 mentioned the
[CollapseQParser.getTopFieldCacheReader|https://github.com/apache/solr/blob/main/solr/core/src/java/org/apache/solr/search/CollapsingQParserPlugin.java#L476]
code and there {{UninvertingReader}} reader is used and since that "that
exposes indexed values as if they also had docvalues" then conceptually a new
{{org.apache.solr.ltr.feature.UninvertingFieldValueFeature}} feature for
{{stored="false" docValues="false" indexed="true"}} fields could be supported
though performance wise there could be issues, depending on the use case.
> pure DocValues support for FieldValueFeature
> --------------------------------------------
>
> Key: SOLR-12697
> URL: https://issues.apache.org/jira/browse/SOLR-12697
> Project: Solr
> Issue Type: Sub-task
> Components: contrib - LTR
> Reporter: Stanislav Livotov
> Priority: Major
> Attachments: SOLR-12697.patch, SOLR-12697.patch, SOLR-12697.patch,
> SOLR-12697.patch, SOLR-12697.patch
>
> Time Spent: 20m
> Remaining Estimate: 0h
>
> [~slivotov] wrote in SOLR-12688:
> bq. ... FieldValueFeature doesn't support pure DocValues fields (Stored
> false). Please also note that for fields which are both stored and DocValues
> it is working not optimal because it is extracting just one field from the
> stored document. DocValues are obviously faster for such usecases. ...
> (Please see SOLR-12688 description for overall context and analysis results.)
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
