[
https://issues.apache.org/jira/browse/CASSANDRA-10784?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15061734#comment-15061734
]
Robert Stupp commented on CASSANDRA-10784:
------------------------------------------
[~jasonstack], the problem with time zone (and locales) is that the reference
data needs to be in sync in the whole cluster. But (unfortunately) time zones
(and also locales & currencies) change. Companies that have to deal with TZs,
locales and/or currencies in their application build their own data set (or buy
that from vendors) to be both up-to-date and consistent - where "up-to-date"
also means that you get the correct information for any specific timestamp for
any location.
IMO that's not something Cassandra as a database can or should provide. If we
would go that route, it would open a can of worms.
Keep in mind that borders change and political decisions can change everything
- if such changes occur, your master data may become worthless if not
completely wrong. The only source or truth we have is the JRE (or any other
library). Keeping that information in sync in a distributed system with
x-hundred nodes is hard if no impossible (upgrading the JRE on all nodes at
once... no way). Just look at the Java release notes - these things change even
between minor Java releases.
Obviously, you can store time zone information in a separate column, use UDTs,
whatever. But still, IMO it can only be handled correctly in the application -
probably storing the reference data in C*.
Don't get me wrong - I'm not saying that having time zone (or locales or
currencies) is bad - just want to say that IMO it's extremely hard to implement
and maintain right.
> ZonedDateTime as new CQL data type
> ----------------------------------
>
> Key: CASSANDRA-10784
> URL: https://issues.apache.org/jira/browse/CASSANDRA-10784
> Project: Cassandra
> Issue Type: New Feature
> Components: CQL, Documentation and Website
> Reporter: ZhaoYang
> Assignee: ZhaoYang
> Fix For: 3.x
>
>
> For globalization, ZonedDateTime is very important. Timestamp is not enough
> and Zone info should be included. So Cassandra should support ZonedDateTime
> as native data type and ordering using ZonedDateTime's utc value.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
