[
https://issues.apache.org/jira/browse/CALCITE-7346?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=18047722#comment-18047722
]
Zhen Chen commented on CALCITE-7346:
------------------------------------
Thank you [~asolimando] . I think we can first discuss whether the description
in my JIRA ticket is acceptable. Moreover, [~mbudiu] mentioned the Double
precision issue in the PR. Can we determine a reasonable solution?
> Prevent overflow in metadata row-count when LIMIT/OFFSET literal exceeds Long
> range
> -----------------------------------------------------------------------------------
>
> Key: CALCITE-7346
> URL: https://issues.apache.org/jira/browse/CALCITE-7346
> Project: Calcite
> Issue Type: Improvement
> Reporter: Zhen Chen
> Priority: Minor
> Labels: pull-request-available
>
> Currently, the implementations of RelMdRowCount, RelMdMaxRowCount, and
> RelMdMinRowCount all use RexLiteral.longValue() when dealing with
> LIMIT/OFFSET. If the literal value of the LIMIT/OFFSET exceeds the range of
> Long, an overflow error will occur. Ideally, this would be changed to
> BigDecimal, but considering that structures like mq.getRowCount() return
> Double, we can continue using Double. This value is sufficient for most
> databases. As far as I know, ClickHouse currently supports a maximum of
> UInt64. I don't know if there are any databases that support larger values.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
