cassandra-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Ariel Weisberg (JIRA)" <>
Subject [jira] [Commented] (CASSANDRA-9626) Make C* work in all locales
Date Tue, 23 Jun 2015 23:08:42 GMT


Ariel Weisberg commented on CASSANDRA-9626:

bq. So what happens when a) someone has been relying on the old behavior, and b) they upgrade?
We put them between a rock and a hard place, they most will likely change the setting not
realizing they were even relying on the old behavior, and then they have an upgraded cluster
and aren't getting back the data they expected. No amount of NEWS.txt changes will prevent
this unfortunately.
The database will detect the mismatch and refuse to start. So when they go to do a rolling
upgrade the first node will make them aware of the configuration mismatch and then they have
an opportunity to update the configuration (either for the node or C*) to resolve it. If they
want they can run C* with the same locale as the node or they can leave it mismatched and
configure C* to ignore the inconsistency. Either way we ship a default and don't allow people
to unwittingly run with different locales all the time.

Do we want to consider the advice to use Local.ROOT instead of US or ENGLISH?

> Make C* work in all locales
> ---------------------------
>                 Key: CASSANDRA-9626
>                 URL:
>             Project: Cassandra
>          Issue Type: Improvement
>            Reporter: Robert Stupp
>            Assignee: Robert Stupp
>            Priority: Minor
>         Attachments: 9626.txt
> Default locale and default charset has immediate effect on how strings are encoded and
handles - e.g. via {{String.toLowerCase()}} or {{new String(byte[])}}.
> Problems with different default locales + charsets don't become obvious for US and most
European regional settings. But some regional OS settings will cause severe errors. Example:
{{"BILLY".toLowerCase()}} returns {{bılly}} with Locale tr_TR (take a look at the second
letter - it's an i without the dot).
> (ref:
> It's not a problem I'm currently facing, but it could become a problem for some users.
A quick fix could be to set default locale and charset in the start scripts - maybe that's
all we need.

This message was sent by Atlassian JIRA

View raw message