chore: project setup — README (archive + living fork), ROADMAP, CI matrix, OCR review, standard files - #13
Merged
Merged
Conversation
- README.md: archive + living-fork framing; branches/tags + how to diff; SSI
as first new feature.
- ROADMAP.md: 10 prioritized features to match/beat InnoDB & WiredTiger,
weighted to multicore/NUMA scalability.
- .github/ocr/{litellm.yaml,rule.json}: OCR config (rule.json adapted to
Berkeley DB C/region/lock/MVCC conventions).
- .github/{CODE_OF_CONDUCT,CONTRIBUTING,SECURITY}.md + PR template.
- .editorconfig (BDB tab conventions), .gitattributes, .git-blame-ignore-revs.
(CI + OCR workflow YAML land in a follow-up commit requiring a workflow-scoped
push token.)
.github/workflows/ci.yml (Linux/macOS x gcc/clang x configure variants, 32-bit, ASan/UBSan, Windows MSBuild, TCL tests) and ocr-review.yml (AI PR review via Bedrock/LiteLLM + OIDC).
gburd
added a commit
that referenced
this pull request
Jun 16, 2026
Add docs/design/lsm.md comparing three log-structured implementations against the Amethyst adaptive-compaction paper, and refine ROADMAP #9. Findings: - aether (Rust): HanoiDB levels + SuRF/Bloom + STRUCTURE-level adaptation (SingleIndex/Hybrid/MultiLevel via rolling metrics + hysteresis) + work-budget merge strategies. No leveled/tiered choice. - Amethyst (paper): SEGMENT-level adaptive compaction (per-segment leveled/tiered FSM + counters + cooldowns) -- orthogonal to aether, same mechanism. - libxtc (C): Bitcask append-only log + in-memory hash index = the LSM-HASH config. - noxu (Rust): JE-style B+tree-in-WAL + cleaner + VLSN/replication = the index-in-WAL config. Synthesis: a two-axis adaptive LSM (aether structure axis + Amethyst segment policy axis) over one shared rolling-counter+cooldown controller; HanoiDB layout + SuRF/Bloom; and #9 (LSM) / #13 (HASH) / #14 (index-in-WAL) share one log-structured core (Bitcask for HASH, JE cleaner for B-tree).
gburd
added a commit
that referenced
this pull request
Jun 16, 2026
/#14 lab/lsm: standalone prototype of the two-axis adaptive controller from docs/design/lsm.md -- one rolling-window + cooldown core driving the structure axis (SINGLE/HYBRID/MULTILEVEL) and the per-segment policy axis (LEVELED/ TIERED). Deterministic (virtual tick). test_adaptive.c runs an Amethyst-style phase-shifting workload; all assertions pass (spawn/collapse, anti-flap, and per-segment convergence with no churn). 'make check' to run. ROADMAP: #13 now ties the HASH-algorithm review to the Bitcask/LSM-HASH in-memory directory (Ctrie/HAMT); #14 ties the JE index-in-WAL + cleaner to the shared log-structured core (cleaner == compactor) and the adaptive controller. Both reference docs/design/lsm.md.
gburd
added a commit
that referenced
this pull request
Jul 31, 2026
chore: project setup — README (archive + living fork), ROADMAP, CI matrix, OCR review, standard files
gburd
added a commit
that referenced
this pull request
Jul 31, 2026
Add docs/design/lsm.md comparing three log-structured implementations against the Amethyst adaptive-compaction paper, and refine ROADMAP #9. Findings: - aether (Rust): HanoiDB levels + SuRF/Bloom + STRUCTURE-level adaptation (SingleIndex/Hybrid/MultiLevel via rolling metrics + hysteresis) + work-budget merge strategies. No leveled/tiered choice. - Amethyst (paper): SEGMENT-level adaptive compaction (per-segment leveled/tiered FSM + counters + cooldowns) -- orthogonal to aether, same mechanism. - libxtc (C): Bitcask append-only log + in-memory hash index = the LSM-HASH config. - noxu (Rust): JE-style B+tree-in-WAL + cleaner + VLSN/replication = the index-in-WAL config. Synthesis: a two-axis adaptive LSM (aether structure axis + Amethyst segment policy axis) over one shared rolling-counter+cooldown controller; HanoiDB layout + SuRF/Bloom; and #9 (LSM) / #13 (HASH) / #14 (index-in-WAL) share one log-structured core (Bitcask for HASH, JE cleaner for B-tree).
gburd
added a commit
that referenced
this pull request
Jul 31, 2026
/#14 lab/lsm: standalone prototype of the two-axis adaptive controller from docs/design/lsm.md -- one rolling-window + cooldown core driving the structure axis (SINGLE/HYBRID/MULTILEVEL) and the per-segment policy axis (LEVELED/ TIERED). Deterministic (virtual tick). test_adaptive.c runs an Amethyst-style phase-shifting workload; all assertions pass (spawn/collapse, anti-flap, and per-segment convergence with no churn). 'make check' to run. ROADMAP: #13 now ties the HASH-algorithm review to the Bitcask/LSM-HASH in-memory directory (Ctrie/HAMT); #14 ties the JE index-in-WAL + cleaner to the shared log-structured core (cleaner == compactor) and the adaptive controller. Both reference docs/design/lsm.md.
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.
Living-fork project scaffolding on
master.README — reframes the repo as both a historical Berkeley DB archive (
historicalbranch + all version tags) and a living fork onmaster; documents branches/tags + how togit diffthe archive; calls out SSI (PR #12) as the first new feature.ROADMAP — 10 prioritized features to make BDB competitive with InnoDB and WiredTiger, weighted to multicore/NUMA scalability (NUMA-aware sharded buffer pool, latch-free buffer lookup, group-commit WAL, low-contention lock manager, concurrent MVCC GC, sharded eviction, false-sharing elimination, compression/LSM, adaptive indexing + async I/O, measured benchmark/perf-regression harness).
OCR —
.github/ocr/{litellm.yaml,rule.json}for the Open Code Review AI reviewer (rule.json rewritten for Berkeley DB conventions). Requires repo variablesAWS_ROLE_ARN/AWS_REGION/OCR_BEDROCK_MODEL(OIDC, no secrets).Standard files — CODE_OF_CONDUCT / CONTRIBUTING / SECURITY, PR template, .editorconfig (BDB tab conventions), .gitattributes, .git-blame-ignore-revs.
NOTE: the two workflow YAMLs (
.github/workflows/ci.ymlbroad build matrix +ocr-review.yml) are committed locally but could not be pushed — the push token lacks the GitHubworkflowOAuth scope. Rungh auth refresh -h github.com -s workflow(or push with a PAT that hasworkflow) and they'll go up as a follow-up commit on this branch.