Repository navigation
[pull] main from MaterializeInc:main - #1420
Merged
Merged
Conversation
## Summary
Automated daily refresh of `doc/developer/generated/` to reflect source
code changes.
This PR was generated by the `update-generated-docs` workflow running
`.claude/commands/update-docs.md`.
## Scope
Only files under `doc/developer/generated/` are modified.
## Review checklist
- [ ] No unrelated rewording or restructuring of existing text
- [ ] No changelog language ("now", "added", "previously", etc.)
- [ ] Changes correspond to actual source code changes
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
### Motivation A materialized view or index that computes `MIN` or `MAX` over an input with updates or deletes kept an extra copy of its input in memory. The first stage of the hierarchical reduction arranged every input row, where it should arrange only the rows its buckets reject. With the default bucketing almost every row wins its first-stage bucket, so that arrangement should be nearly empty. For a `MAX` over 100,000 rows in 1,000 groups: | | before | after | |---|---|---| | first-stage reduction, default bucketing | 100,000 records, 4.00 MB | 0 records | | all arrangements of the view, default bucketing | 16.8 MB | 12.8 MB | | first-stage reduction, `AGGREGATE INPUT GROUP SIZE = 100` | 100,000 records, 4.25 MB | 84,017 records, 2.84 MB | | all arrangements of the view, `AGGREGATE INPUT GROUP SIZE = 100` | 5.4 MB | 3.9 MB | The saving applies where a bucket's winner is one of its input rows, which holds for a single `MIN` or `MAX`. `ReduceReduction` keeps all hierarchical aggregates of a `Reduce` in one hierarchy, so for `MIN(a), MAX(b)` the winner combines values from different rows, matches no input row, and the stage arranges its input as before the change. ### Description Each stage maintains, as its reduction's output, its winners minus its input, and emits `input + output`, which is the winners alone. The reduction pushed each winner with `Diff::MINUS_ONE`, so a stage emitted its winners at -1, and the first stage's reduction held the winner at -2 and every other row at -1. Later stages, fed -1 inputs, still held only their rejected rows, which is why only the first stage grew. The sign has been off since #27085 moved the negation of the inputs into the reduction. Results were unaffected: min and max count each row once whatever its multiplicity, and no check reads a stage's output. The first stage validates its raw input, and the final `ReduceMinsMaxes` check runs only when there are no stages. The winner is now emitted with `Diff::ONE`, so stages emit their winners at +1 and a winning input row cancels against itself in its stage's arrangement. TopK's corresponding stage, `build_topk_negated_stage`, already uses this sign. ### Verification Adds a section to `test/testdrive/introspection-sources.td`: a `MAX` over 1,000 groups of one row each. Every row wins its bucket at every stage however the rows hash, so the test asserts that no stage's reduction arranges a record. Without this change the first stage holds all 1,000 rows. A second view computes `MIN(v), MAX(v)` over the same table to cover value rows with one column per aggregate, where the winner of a single-row group still equals its input. 🤖 Generated with [Claude Code](https://claude.com/claude-code) --------- Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> Co-authored-by: Moritz Hoffmann <antiguru@gmail.com>
Nightly release builds on x86_64 and aarch64 fail since #39339 bumped tikv-jemalloc-sys to 0.7.1 (jemalloc 5.3.1). jemalloc's configure aborts with `cannot determine return type of strerror_r`. jemalloc 5.3.1 runs its `strerror_r` link checks with `-Werror`. The release LDFLAGS contain `-static-libstdc++`, which clang reports as unused when linking a C program, so both checks fail. This change adds `-Wno-unused-command-line-argument` to the release LDFLAGS. Running configure in the ci-builder image with clang-22 and the nightly's flags reproduces the failure, and with this flag configure succeeds and detects `JEMALLOC_STRERROR_R_RETURNS_CHAR_WITH_GNU_SOURCE`. Posted by Claude Code. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
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 subscribe to this conversation on GitHub.
Already have an account?
Sign in.
2 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.
See Commits and Changes for more details.
Created by
pull[bot] (v2.0.0-alpha.4)
Can you help keep this open source service alive? 💖 Please sponsor : )