flink-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "godfrey he (JIRA)" <j...@apache.org>
Subject [jira] [Updated] (FLINK-12018) Add support for generating optimized logical plan for Sort and Rank
Date Mon, 01 Apr 2019 08:30:00 GMT

     [ https://issues.apache.org/jira/browse/FLINK-12018?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]

godfrey he updated FLINK-12018:
-------------------------------
    Description: 
Add support for generating optimized logical plan for Sort and Rank.

There is a summary of Sort and Rank converters:
for batch, 
1. Sort without sort fields will be converted to BatchExecLimit
2. Sort with sort fields, null-fetch and null-offset will be converted to BatchExecSort
3. otherwise, Sort will be converted to BatchExecSortLimit
4. RANK function on over with filter will be converted to BatchExecRank

for stream,
1. Sort with time-ascending-order and non-limit will be converted to StreamExecTemporalSort
2. Sort without sort fields will be converted to StreamExecLimit
3. Sort with sort fields, null-fetch and null-offset will be converted to StreamExecSort
4. otherwise, Sort will be converted to StreamExecSortLimit
5. Rank with ROW_NUMBER function which is sorted on proc-time attribute and fetches only one
record start from 0 will be converted to StreamExecDeduplicate
5. otherwise, RANK function on over with filter will be converted to StreamExecRank


  was:
Add support for generating optimized logical plan for Sort and Rank.

There is a summary of Sort and Rank converters:
for batch, 
1. Sort without sort fields, fetch and offset will be converted to BatchExecLimit
2. Sort with sort fields, null-fetch and null-offset will be converted to BatchExecSort
3. otherwise, Sort will be converted to BatchExecSortLimit
4. RANK function on over with filter will be converted to BatchExecRank

for stream,
1. Sort with time-ascending-order and non-limit will be converted to StreamExecTemporalSort
2. Sort with limit will be converted to StreamExecRank
3. Sort which is sorted on proc-time attribute and fetches only one record start from 0 will
be converted to StreamExecFirstLastRow
4. otherwise, Sort will be converted to StreamExecSort
5.  RANK function on over with filter will be converted to StreamExecRank
6. Rank with ROW_NUMBER function which is sorted on proc-time attribute and fetches only one
record start from 0 will be converted to StreamExecFirstLastRow


> Add support for generating optimized logical plan for Sort and Rank
> -------------------------------------------------------------------
>
>                 Key: FLINK-12018
>                 URL: https://issues.apache.org/jira/browse/FLINK-12018
>             Project: Flink
>          Issue Type: New Feature
>          Components: Table SQL / Planner
>            Reporter: godfrey he
>            Assignee: godfrey he
>            Priority: Major
>              Labels: pull-request-available
>             Fix For: 1.9.0
>
>          Time Spent: 20m
>  Remaining Estimate: 0h
>
> Add support for generating optimized logical plan for Sort and Rank.
> There is a summary of Sort and Rank converters:
> for batch, 
> 1. Sort without sort fields will be converted to BatchExecLimit
> 2. Sort with sort fields, null-fetch and null-offset will be converted to BatchExecSort
> 3. otherwise, Sort will be converted to BatchExecSortLimit
> 4. RANK function on over with filter will be converted to BatchExecRank
> for stream,
> 1. Sort with time-ascending-order and non-limit will be converted to StreamExecTemporalSort
> 2. Sort without sort fields will be converted to StreamExecLimit
> 3. Sort with sort fields, null-fetch and null-offset will be converted to StreamExecSort
> 4. otherwise, Sort will be converted to StreamExecSortLimit
> 5. Rank with ROW_NUMBER function which is sorted on proc-time attribute and fetches only
one record start from 0 will be converted to StreamExecDeduplicate
> 5. otherwise, RANK function on over with filter will be converted to StreamExecRank



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

Mime
View raw message