[
https://issues.apache.org/jira/browse/CALCITE-3218?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16897619#comment-16897619
]
Julian Hyde commented on CALCITE-3218:
--------------------------------------
[~lindseycat] wrote:{quote} Is the babel parser different from the one we were
looking at yesterday?{quote}
Yes, kind of. It's an extension to the core parser. You "extend" it by adding
definitions to text files read by the templating system. See what I did to add
the "DATE()" function in
[f1cad57|https://github.com/apache/calcite/commit/f1cad57035db0f94190433aa6db2352228a641e0].
That said, you should start by getting it working in the core parser, and add
tests. Then create a PR. Then it's straightforward to move the code from
{{Parser.jj}} to {{parserImpls.ftl}} and {{config.fmpp}}.
> Syntax error while parsing DATEADD function (which is valid on Redshift)
> ------------------------------------------------------------------------
>
> Key: CALCITE-3218
> URL: https://issues.apache.org/jira/browse/CALCITE-3218
> Project: Calcite
> Issue Type: Bug
> Reporter: Lindsey Meyer
> Assignee: Julian Hyde
> Priority: Major
>
> Syntax error while parsing the {{DATEADD}} function:
> {code:java}
> SELECT
> DATE(CONVERT_TIMEZONE('UTC', 'America/Los_Angeles', events.event_date
> )) AS "events.event_date",
> COALESCE(SUM(events.daily_user_count ), 0) AS
> "events.daily_active_users",
> COALESCE(SUM(events.monthly_user_count ), 0) AS
> "events.monthly_active_users"
> FROM public.events_proto AS events
> WHERE
> (((events.event_date ) >= ((CONVERT_TIMEZONE('America/Los_Angeles',
> 'UTC', DATEADD(day,-364, DATE_TRUNC('day',CONVERT_TIMEZONE('UTC',
> 'America/Los_Angeles', GETDATE())) )))) AND (events.event_date ) <
> ((CONVERT_TIMEZONE('America/Los_Angeles', 'UTC', DATEADD(day,365,
> DATEADD(day,-364, DATE_TRUNC('day',CONVERT_TIMEZONE('UTC',
> 'America/Los_Angeles', GETDATE())) ) ))))))
> GROUP BY 1
> HAVING
> NOT (COALESCE(SUM(events.monthly_user_count ), 0) = 0)
> ORDER BY 1 DESC
> LIMIT 500{code}
> throws
> {noformat}
> `Column 'year' not found in any table`
> `DATEADD(year,1,...`
> {noformat}
> This query is valid on Redshift. Redshift's parser understands that the first
> argument ({{DAY}}) is intended to be a time unit, not an identifier.
--
This message was sent by Atlassian JIRA
(v7.6.14#76016)
