Hello everyone,
we have recently had an issue for a change in Avatica which was not working
for all timezones that almost slipped through (see CALCITE-4600
<https://issues.apache.org/jira/browse/CALCITE-4600>) if it was not for a
local run which happened to be in a different timezone.

As for Calcite, also Avatica has code paths which are susceptible to depend
on the local user timezone (notably the time-related data types), in the
current state CI won't necessarily catch misbehaviour related to that.

I think it would be worth adding at least an additional CI check using a
US/Asia timezone. Avatica tests are quick to run, so I don't think it would
be a big issue.

What do you think?

Best regards,
Alessandro

Reply via email to