[
https://issues.apache.org/jira/browse/FLINK-6108?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Luke Hutchison updated FLINK-6108:
----------------------------------
Description:
Currently it is possible to do
{code}
dataSet.groupBy(0).sum(1)
{code}
but not
{code}
dataSet.groupBy("id").sum("count")
{code}
groupBy() takes named fields, but sum() does not, it only works with tuples. It
would be great if sum() could take named fields too -- otherwise you have to
map to tuples to make use of easy summing.
(This would probably only work reliably with POJOs, since a new object instance
would need to be constructed to take the field containing the sum.)
Similarly, min(), max() etc. should also take named fields.
was:
Currently it is possible to do
{code}
dataSet.groupBy(0).sum(1)
{code}
but not
{code}
dataSet.groupBy("id").sum("count")
{code}
-- groupBy() takes named fields, but sum does not, it only works with tuples.
It would be great if sum() could take named fields too -- otherwise you have to
map to tuples to make use of easy summing.
(This would probably only work reliably with POJOs, since a new object instance
would need to be constructed to take the field containing the sum.)
Similarly, min, max etc. should also take named fields.
> Add sum(fieldName) to sum based on named POJO fields
> ----------------------------------------------------
>
> Key: FLINK-6108
> URL: https://issues.apache.org/jira/browse/FLINK-6108
> Project: Flink
> Issue Type: Bug
> Components: Core
> Affects Versions: 1.2.0
> Reporter: Luke Hutchison
>
> Currently it is possible to do
> {code}
> dataSet.groupBy(0).sum(1)
> {code}
> but not
> {code}
> dataSet.groupBy("id").sum("count")
> {code}
> groupBy() takes named fields, but sum() does not, it only works with tuples.
> It would be great if sum() could take named fields too -- otherwise you have
> to map to tuples to make use of easy summing.
> (This would probably only work reliably with POJOs, since a new object
> instance would need to be constructed to take the field containing the sum.)
> Similarly, min(), max() etc. should also take named fields.
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
