You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@calcite.apache.org by "Julian Hyde (JIRA)" <ji...@apache.org> on 2016/10/18 23:38:58 UTC

[jira] [Updated] (CALCITE-1429) Druid adapter must send "fromNext" when requesting rows from Druid

     [ https://issues.apache.org/jira/browse/CALCITE-1429?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Julian Hyde updated CALCITE-1429:
---------------------------------
    Summary: Druid adapter must send "fromNext" when requesting rows from Druid  (was: Error of fetching data in select query for Druid)

> Druid adapter must send "fromNext" when requesting rows from Druid
> ------------------------------------------------------------------
>
>                 Key: CALCITE-1429
>                 URL: https://issues.apache.org/jira/browse/CALCITE-1429
>             Project: Calcite
>          Issue Type: Bug
>          Components: druid
>            Reporter: Jiarong Wei
>            Assignee: Julian Hyde
>
> As mentioned in [Druid documentation|http://druid.io/docs/0.9.1.1/querying/select-query.html],
> {quote}
> Note that in the second query, an offset is specified and that it is 1 greater than the largest offset found in the initial results. To return the next "page", this offset must be incremented by 1 (should be decremented by 1 for descending query), with each new query, but with option {{fromNext}} enabled, this operation is not needed. When an empty results set is received, the very last page has been returned.
> {quote}
> I think Druid adapter [presumes|https://github.com/apache/calcite/blob/master/druid/src/main/java/org/apache/calcite/adapter/druid/DruidQuery.java#L935] this feature is turned on but it's not set in [code|https://github.com/apache/calcite/blob/master/druid/src/main/java/org/apache/calcite/adapter/druid/DruidQuery.java#L627].
> Also, {{previousOffset}} may not be reasonable, see [DruidQuery|https://github.com/apache/calcite/blob/master/druid/src/main/java/org/apache/calcite/adapter/druid/DruidQuery.java#L940]. I think whether we have already fetched all data is decided by the next request. If we don't have any data returned on next request, that means fetching is done.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)