[
https://issues.apache.org/jira/browse/SOLR-7548?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14545810#comment-14545810
]
Joel Bernstein commented on SOLR-7548:
--------------------------------------
One of the main reasons to not make the /export handler the default is that it
is a little painful to get started with because it is so picky.
So, I think most people would want to prototype some Streaming Expressions with
a small amount of data using the default handler. Then go into production with
/export handler on the larger data set.
As Yonik mentions, ideally we could do everything from the default handler, but
the "/export" handler enforces a bunch of rules that enable it export and sort
really large data sets. If we can find a way to loosen those rules and still
handle the export of large data sets that would be great.
We could also have a flag in the default handler to enforce certain rules when
run in "streaming" mode. There are still other technical hurdles to getting the
default handler to behave like the /export handler though.
> CloudSolrStream Limits Max Results to rows Param
> ------------------------------------------------
>
> Key: SOLR-7548
> URL: https://issues.apache.org/jira/browse/SOLR-7548
> Project: Solr
> Issue Type: Bug
> Affects Versions: Trunk
> Reporter: Dennis Gove
> Priority: Minor
> Labels: Streaming
> Fix For: Trunk
>
>
> The CloudSolrStream stream class accepts a set of params to be passed to the
> standard query handler. If the provided params doesn't include "rows=N" then
> maximum # of records returned by this stream is the configured default rows
> value (generally 10, but perhaps more).
> As CloudSolrStream would generally be the first part of a larger set of
> stream expressions it seems counterintuitive to limit the first set by this
> value.
> This ticket is to address this so that either we set pass a param of rows=MAX
> where MAX is the max value we can pass (max int or max long I suppose) or
> make it so that default value is ignored when in a streaming context.
> Example:
> Imagine we have a collection "people" with 90 documents in it
> The following query would return at most 10 documents (assuming 10 is the
> default)
> {code}
> search(people,q=*:*,fl="id,name_s,gender_s,nick_s",sort="name_s desc")
> {code}
> The following query would return all documents
> {code}
> search(people,q=*:*,fl="id,name_s,gender_s,nick_s",sort="name_s
> desc",rows=100)
> {code}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org
For additional commands, e-mail: dev-h...@lucene.apache.org
