[
https://issues.apache.org/jira/browse/CALCITE-5245?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17596919#comment-17596919
]
Julian Hyde commented on CALCITE-5245:
--------------------------------------
[~goldmedal], Can you make clear in the summary / description that you are
talking about generating SQL, i.e. the JDBC adapter. I initially thought you
were proposing support for ORDINAL and OFFSET in the parser (which would also
be a valid request).
I agree with [~libenchao] that this might be better solved at the RexNode
level, where type information is present. Calcite's ITEM operator complies with
ISO standard SQL, so one idea is to add two BigQuery-specific functions,
ARRAY_ITEM_BASE_0 and ARRAY_ITEM_BASE_1, which generate OFFSET and ORDINAL
respectively. You would need to transform ITEM to ARRAY_ITEM_BASE_1 in
RelToSqlConverter.
> Array element accessing in BigQuery should invoke `ORDINAL` operator
> --------------------------------------------------------------------
>
> Key: CALCITE-5245
> URL: https://issues.apache.org/jira/browse/CALCITE-5245
> Project: Calcite
> Issue Type: Bug
> Components: core
> Affects Versions: 1.31.0
> Reporter: Jia-Xuan Liu
> Priority: Minor
> Labels: pull-request-available
> Time Spent: 1h 10m
> Remaining Estimate: 0h
>
> Access an element of array in BigQuery should invoke `ORDINAL`(1-based) or
> `OFFSET`(0-based) operator for the index. See [BigQuery Accessing array
> elements|https://cloud.google.com/bigquery/docs/reference/standard-sql/arrays#accessing_array_elements].
> For example:
> {code:sql}
> SELECT ARRAY[1,2,3][ORDINAL(1)]
> {code}
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
