adventurous-agency-7776408/21/2023, 12:43 PM
clause in the experiment queries from Growthbook Cloud that emerged sometime over the weekend. Currently, the query compiles to
With this query, I get the error
WHERE e.experiment_id = 'hybrid_and_it2' AND e.timestamp >= '2023-08-17 15:47:00' AND e.timestamp <= '2023-08-21 12:33:48'
. The date ‘2023-08-17’ is the start date of the experiment. If I copy the query into BigQuery and edit the query above with
Could not cast literal "2023-08-17 15:47:00" to type DATE at [25:26]
, I no longer get the error.
Would someone be able to help support to solve? 🙏
WHERE e.experiment_id = 'hybrid_and_it2' AND e.timestamp >= datetime('2023-08-17 15:47:00') AND e.timestamp <= datetime('2023-08-21 12:33:48')
helpful-application-710708/21/2023, 4:09 PM
statements because it was causing sub-optimal performance when people had tables partitioned by a TIMESTAMP field rather than a DATETIME field. The error is happening because the
column you pass us is of type
and we use a string literal with hours to build the date range. We want to be able to support the case where
, especially in the metrics source, where maybe this filter is less critical and just used to improve performance, so we will work on that ASAP. One potential solution in the meantime would be to
in your Experiment Assignment query. In that case, our filters should work just as before.
CAST(timestamp AS DATETIME) as timestamp
that is of type
, you may also need to cast those to
to get it to work, at least until we land a fix from our end.
adventurous-agency-7776408/21/2023, 4:20 PM
helpful-application-710708/21/2023, 4:23 PM
fields. Because we allow you to set start and end dates for an experiment using hours (and minutes) and because you can set conversion windows down to the hour, we really are built to work with
data fields. When you work with
you lose some of that fidelity.
adventurous-agency-7776408/21/2023, 4:24 PM
agreeable-monitor-6272708/22/2023, 8:02 PM
helpful-application-710708/22/2023, 10:53 PM
blue-telephone-4043709/21/2023, 7:48 PM