Sylvain Lebresne commented on CASSANDRA-12373:

bq. What we want/need to do re:schema is change python- and java- drivers

Right, didn't meant nothing has to be done, just that it wasn't part of this 
patch (nor C* really). I'll mark "client-impacting" in fact, so clients are 
aware they have something to do here too.

bq. Don't forget the "snapshot" schema changing code.

Well, that's kind of an interesting point because there is in fact currently 
*no* way to create a super column table from CQL. That's kind of on purpose, we 
do not want people to create them nowadays, we only want to maintain backward 
compatibility, but it does mean if you need to restore a snapshot of a super 
column table, you have to currently use thrift. So I don't known, maybe in the 
short term saying that we don't save schema for snapshot at all for super 
column table is the "most honest" solution.

But it kind of suggest to me that we should get CASSANDRA-10857 in ASAP, and 
then force users to use it on super column tables before upgrading to 4.0. It's 
not ideal, but I don't dragging super columns support forever is a better 
solution either.

> 3.0 breaks CQL compatibility with super columns families
> --------------------------------------------------------
>                 Key: CASSANDRA-12373
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-12373
>             Project: Cassandra
>          Issue Type: Bug
>          Components: CQL
>            Reporter: Sylvain Lebresne
>            Assignee: Alex Petrov
>             Fix For: 3.0.x, 3.x
> This is a follow-up to CASSANDRA-12335 to fix the CQL side of super column 
> compatibility.
> The details and a proposed solution can be found in the comments of 
> CASSANDRA-12335 but the crux of the issue is that super column famillies show 
> up differently in CQL in 3.0.x/3.x compared to 2.x, hence breaking backward 
> compatibilty.

This message was sent by Atlassian JIRA

Reply via email to