feat: PartitionedOutputNode must distinguish the root of a plan #15404
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Currently when we translate a Presto plan to a Velox plan fragment, we do not have a way later on at the Velox layer to detect if a particular PartitionedOutputNode will be read by the coordinator as the final output of a plan or will be read by other workers. This patch adds a new field to PartitionedOutputNode so we can signal this during plan translation inside Prestissimo. In my particular case, This is needed to order to support exchange translation decisions in the cuDF exchange (coming in via a later PR), where the last PartitionedOutputNode has a single partition and will be read by the Presto coordinator, but all other PartitionedOutputNodes will be actually read by other workers via special, hardware accelerated Exchange. Without this change we have no way to distinguish these two at the pure Velox DriverAdapter translation level.
✅ Deploy Preview for meta-velox canceled.
|
meta-cla
bot
added
the
CLA Signed
majetideepak
changed the title
majetideepak
reviewed
Nov 5, 2025
| return outputType_; | ||
| } | ||
|
|
||
| const bool isRootFragment() const { |
Collaborator
There was a problem hiding this comment.
no need to use const since the return is a scalar value here.
bool isRootFragment() const {
Similar to below
bool isPartitioned() const {
Collaborator
|
@mbasmanova, @xiaoxmeng, any thoughts on this API change? |
Collaborator
|
On the Prestissimo side, the OutputNode is also mapped to the Velox PartitionedOutputNode. This field will help capture that information. |
mbasmanova
reviewed
Nov 5, 2025
Contributor
mbasmanova
left a comment
mbasmanova
left a comment
There was a problem hiding this comment.
@zoltan It sounds like you want to introduce a new plan node: OutputNode. That node is a special case of PartitionedOutputNode: it must be the root node and must have "gather" semantic.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Currently when we translate a Presto plan to a Velox plan fragment, we do not have a way later on at the Velox layer to detect if a particular PartitionedOutputNode will be read by the coordinator as the final output of a plan or will be read by other workers.
This patch adds a new field to PartitionedOutputNode so we can signal this during plan translation inside Prestissimo.
In my particular case, This is needed to order to support exchange translation decisions in the cuDF exchange (coming in via a later PR), where the last PartitionedOutputNode has a single partition and will be read by the Presto coordinator, but all other PartitionedOutputNodes will be actually read by other workers via special, hardware accelerated Exchange. Without this change we have no way to distinguish these two at the pure Velox DriverAdapter translation level.