[FLINK-729] HashReduce based on java-util-hash-map

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Invalid
• Fix Version/s: pre-apache
• Labels: github-import
• Assignee: Unassigned
• Reporter: GitHub Import
• Created: 09/Jun/14 12:50
• Updated: 18/Jun/14 15:44
• Resolved: 18/Jun/14 15:44

Description

This is a first step for the hash-aggregator, corresponding issue: https://github.com/stratosphere/stratosphere/issues/82

This implementation is a ReduceOperator based on a hash-table, instead of sorting the elements.
It is just a prototype because it is still using the java-util-hash-map, and furthermore the CostEstimation is just 0 for now.

This PR is based on the PR for dividing the reduce and group reduce operators(https://github.com/stratosphere/stratosphere/pull/726), only commit bb5640a is the HashReduce code. I just opened the PR to get feedback on the current code and how I should proceed with a stratosphere-hash-map implementation and adjusting the CostEstimation.

---------------- Imported from GitHub ----------------
Url: https://github.com/stratosphere/stratosphere/pull/729
Created by: filiphaase
Labels:
Created at: Mon Apr 28 12:23:25 CEST 2014
State: open

Issue Links

• is related to: FLINK-82 (Hash) Aggregators (fold style)