compute: Take the linear-join accumulator off the collection edge - #37788
Draft
antiguru wants to merge 3 commits into
Draft
compute: Take the linear-join accumulator off the collection edge#37788antiguru wants to merge 3 commits into
antiguru wants to merge 3 commits into
Conversation
antiguru
force-pushed
the
columnar-tb-join-accumulator
branch
from
July 22, 2026 08:41
7323a80 to
083529c
Compare
antiguru
force-pushed
the
columnar-tb-join-accumulator
branch
2 times, most recently
from
July 22, 2026 17:50
d299f39 to
9a534cc
Compare
antiguru
force-pushed
the
columnar-tb-join-accumulator
branch
from
August 19, 2026 12:01
9a534cc to
5f6e9e6
Compare
antiguru
force-pushed
the
columnar-tb-join-accumulator
branch
2 times, most recently
from
August 20, 2026 08:49
6eca1cf to
ea1f3f4
Compare
antiguru
force-pushed
the
columnar-tb-join-accumulator
branch
from
August 20, 2026 09:13
ea1f3f4 to
8808ba0
Compare
`mz_join_core` is `Vec`-internal, so the linear join accumulates across stages in a `VecCollection`. That intra-operator accumulator was wrapped in the `CollectionEdge` enum (`JoinedFlavor::Collection(CollectionEdge)`, whose `Vec` arm was the live accumulator), which conflated the inter-node edge type with a purely local `Vec`. Split `JoinedFlavor` so the collection edge appears only where it should: `Edge(CollectionEdge)` is the stage-1 source input (a columnar edge from an upstream producer), and `Collection(VecCollection)` is the bare multi-stage accumulator. Stage 1 keys its source off the edge via `arrange_join_input` (unchanged, both arms); subsequent stages arrange the bare accumulator via the new `arrange_join_collection`. The two share the `Vec` keying (`key_join_input_vec`) and the arrange tail (`arrange_keyed_join_input`), so the source `Vec` arm and the accumulator path cannot drift. Finalization (node P7) still encodes the output edge. Behavior-preserving: the accumulator was already `Vec` at runtime, just wrapped in `CollectionEdge::Vec`; this drops the wrapper. `mz_join_core`, the source columnar path, and the output encoding are untouched, and all operator names are preserved. After this, `JoinedFlavor::Collection` no longer mentions `CollectionEdge`, so the enum collapse leaves the join accumulator alone. Adds `arrange_join_collection_matches_vec_edge` to guard the accumulator wiring against the `Vec` edge arm. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Comment-only, no behavior change. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The doc named the migration plan's node for the borrowed-push pattern. State the property instead, which is what a caller needs to know.
antiguru
force-pushed
the
columnar-tb-join-accumulator
branch
from
September 6, 2026 18:23
8808ba0 to
b829158
Compare
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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.
Move the linear-join multi-stage accumulator off the
CollectionEdgeonto a bareVecCollection, encoding to the columnar edge only at the join output. Removes an internal live Vec use of the edge.Columnar dataflow-edge migration. Design doc:
doc/developer/design/20260720_columnar_dataflow_edges.md(#37744).Part of CPU-51.