This is an expected behavior.

We learned this issue/feature at the current site (we use Dse 5.08)

Subroto 

> On Jun 18, 2017, at 10:29 PM, Zhongxiang Zheng <zzh...@yahoo-corp.jp> wrote:
> 
> Hi all,
> 
> I have a question about a behavior when insert a list with specifying 
> timestamp.
> 
> It is documented that "An INSERT will always replace the entire list."
> https://github.com/apache/cassandra/blob/trunk/doc/cql3/CQL.textile#lists
> 
> However, When a list is inserted multiple times using same timestamp,
> it will not be replaced, but will be added as follows.
> 
> cqlsh> CREATE TABLE test.test (k int PRIMARY KEY , v list<int>);              
>                                                                               
>                                                                         
> cqlsh> INSERT INTO test.test (k , v ) VALUES ( 1 ,[1]) USING TIMESTAMP 1000 ; 
>                                                                               
>                                                                         
> cqlsh> INSERT INTO test.test (k , v ) VALUES ( 1 ,[1]) USING TIMESTAMP 1000 ;
> cqlsh> SELECT * FROM test.test ;
> 
> k | v
> ---+--------
> 1 | [1, 1]
> 
> I confirmed this behavior is reproduced in 3.0.13 and 3.10.
> I'd like to ask whether this behavior is a expected behavior or a bug?
> 
> In our use case, CQL statements with same values and timestamp will be issued 
> multiple times
> to retry inserting under the assumption that insert is idempotent.
> So, I expect that the entire list will be replace even if insert a list 
> multiple times with same timestamp.
> 
> Thanks,
> 
> Zhongxiang
> 
> 
> ТÒÒÒÒÒÒÒÒÒÒÒÒÒÒÒÒÒÒÒÒÒÒÒÒÒÒÒÒÒÒÒÒÒÒÒÒÒÒÒÒÒÒÒÒÒÒÒÒÒÒÒÒÒÒÒÒÒÒÒÒÒÒÒÒÒÒÒÒÐÐ¥FòVç7V'67&–&RÂRÖÖ–âW6W"×Vç7V'67&–&T676æG&æ6†Ræ÷&pФf÷"FF—F–öæÂ6öÖÖæG2ÂRÖÖ–âW6W"Ö†VÇ676æG&æ6†Ræ÷&pÐ


---------------------------------------------------------------------
To unsubscribe, e-mail: user-unsubscr...@cassandra.apache.org
For additional commands, e-mail: user-h...@cassandra.apache.org

Reply via email to