[ 
https://issues.apache.org/jira/browse/CASSANDRA-1472?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12983248#action_12983248
 ] 

Gary Dusbabek commented on CASSANDRA-1472:
------------------------------------------

bq. would it be reasonable to commit this patch with Avro's file format, and 
then to discuss replacing the file format in a separate issue
Not a good idea, imo. 

bq. writing our own file format is an optimization
I don't think it is an optimization at all.  It's a way of protecting us from 
ourselves.  The only way to do that and still use avro is to create our own 
readers according to the spec, no?

bq.  This decision needs to be made for technical reasons and not grounded in 
NIH.
Poppycock.  The reasons are technical.  The fact is that the way we use avro is 
constantly breaking on us (I'm referring to migrations).  Some of this is 
probably self-inflicted, but either way--we should avoid chopping down the 
forest until we can operate the chainsaw without cutting off our fingers.  We 
don't have much experience yet with what happens when records change slightly 
or radically over time and how avro copes with that.  The experience we do have 
leads me to believe that unless we are extremely cautious, we'll end up in a 
hole.  See CASSANDRA-2001 for an example.  Two seemly innocuous changes have 
both managed to break migration deserialization.

> Add bitmap secondary indexes
> ----------------------------
>
>                 Key: CASSANDRA-1472
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-1472
>             Project: Cassandra
>          Issue Type: Improvement
>          Components: Core
>            Reporter: Stu Hood
>            Assignee: Stu Hood
>             Fix For: 0.7.1
>
>         Attachments: 0.7-1472-v5.tgz, 0.7-1472-v6.tgz, 
> 0001-CASSANDRA-1472-rebased-to-0.7-branch.txt, 
> 0019-Rename-bugfixes-and-fileclose.txt, 1472-v3.tgz, 1472-v4.tgz, 
> 1472-v5.tgz, anatomy.png, v4-bench-c32.txt
>
>
> Bitmap indexes are a very efficient structure for dealing with immutable 
> data. We can take advantage of the fact that SSTables are immutable by 
> attaching them directly to SSTables as a new component (supported by 
> CASSANDRA-1471).

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.

Reply via email to