apache / apache/arrow

[C++] Add generic multi-output node

Open
#31,885 2 comments 0 reactions 0 assignees View on GitHub
Component: C++ Type: task
Dominant language
C++
Stars
17.1k
Forks
4.3k
Avg merge
3d 18h
Merged PRs (30d)
91

Description

A node with multiple outputs is a bit of a tricky thing to create and deserving of its own node type. This node would be a pipeline breaker. It would need to schedule N tasks for each incoming tasks (potentially N-1 if you want to run one of the outputs on the current task). It will need to be clever about how it applies backpressure. If either downstream node asks for backpressure then the signal should be forwarded.

Another bonus is that if we can encapsulate this behavior in a generic node then we can get rid of the TeeNode because this would become a multi-output node where one output is a write node and the other output is the regular path.

**Reporter**: [Weston Pace](https://issues.apache.org/jira/browse/ARROW-16524) / @westonpace

**Note**: *This issue was originally created as [ARROW-16524](https://issues.apache.org/jira/browse/ARROW-16524). Please see the [migration documentation](https://github.com/apache/arrow/issues/14542) for further details.*

Contributor guide

Open the contributing guide

Research direction

Start by reviewing the existing TeeNode and the pipeline task-scheduling and backpressure behavior described in the issue. Define how a generic multi-output node schedules N tasks per incoming task, handles the current-task optimization, and forwards downstream backpressure; completion should support replacing TeeNode.

Written by the indexing model from the issue text.

Assessment

Tech stack
cpp
Domain
backend
Issue type
Feature
Difficulty
5/5
Estimated time
Over a week
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
25/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.