[
https://issues.apache.org/jira/browse/CASSANDRA-12417?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15545395#comment-15545395
]
Benjamin Lerer commented on CASSANDRA-12417:
--------------------------------------------
If my understanding is right, both functions (sum and avg) can end up returning
different results, even in normal situation, due to the fact that both use now
the Kahan's algorithm. The way overflow is handled is also not the same.
I do not deny that overall it is much better but it is clearly a change of
behaviour, which is why I would have prefered that the change goes only in 3.X.
Now, it is true that it might also not be a problem for anybody, so feel free
to do as you prefer.
> Built-in AVG aggregate is much less useful than it should be
> ------------------------------------------------------------
>
> Key: CASSANDRA-12417
> URL: https://issues.apache.org/jira/browse/CASSANDRA-12417
> Project: Cassandra
> Issue Type: Bug
> Components: CQL
> Reporter: Branimir Lambov
> Assignee: Alex Petrov
>
> For fixed-size integer types overflow is all but guaranteed to happen,
> yielding incorrect result. While for sum it is somewhat acceptable as the
> result cannot fit the type, this is not the case for average.
> As the result of average is always within the scope of the source type,
> failing to produce it only signifies a bad implementation. Yes, one can solve
> this by type-casting, but do we really want to always have to be telling
> people that the correct spelling of the average function is
> {{cast(avg(cast(value as bigint))) as int)}}, especially if this is so
> trivial to fix?
> Additionally, the straightforward addition we use for floating point versions
> is not a good choice numerically for larger numbers of values. We should
> switch to a more stable version, e.g. iterative mean using {{avg = avg +
> (value - avg) / count}}.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
