Repository navigation
benchmark docs: match the code, retire stage history, owed work and the roster decisions in section 7 (#66) - #67
Merged
Merged
Conversation
…section 7 (sweep 2026-10-02) Match the code: the runner lock path and its startup refusal, the dev-build admission rule (_commit_matches), the heap ratios from HEAP_BY_SCALE, the QuestDB and mongot heap exceptions, the launch exports a commit-pinned campaign needs, MongoDB's graph arm as chained $lookup only, pgvector's sparse index built after its load inside one timer, the int8 dense arms at 9.99M only, what page_check runs now (MAPPING and _resolve are gone), the missing live overrides in PROTOCOL section 7 (Memgraph, FalkorDB, LadybugDB, the PostgreSQL family's memory, Neo4j's checkpoint, ArangoDB, MongoDB, the l4 settle), Memgraph and FalkorDB in F10, the AGE version string the rows carry, the Python-cost table's real source, and the in-use libraries. Retire: the September chain and extension, qCAL, the October plan and uplift narrative, retired L1 tabular rows, struck-through comparator entries, stale version numbers, a rule cited to a decision that does not hold it, and hand-carried counts. Rules are kept; live stage state is left in place. Owed work into CAMPAIGN section 7: rows 26 to 36 (PG+AGE on the graph tables, the $graphLookup check and QUERY_LANGUAGE, pycost, QuestDB and mongot heaps, the F13/F11 header, BUGS C4, planned tables, graph_gav, figures, the sparse one-timer sentence), the memory sweep into row 7, the fixes expected to move cells into row 9, and row 20's thread and memory-pool audits. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01WWcDr7TwvcfMsQkeYQVHZr
…MPARATORS Section 7: row 26 (PostgreSQL + AGE on the graph tables) runs inside the 26.10.1 measurement, not as an October extension; the MongoDB rows record the $graphLookup rewrite and the Python-cost row the re-measure; rows 37-44 add the new dense arms (Elasticsearch fp32 and int8, Memgraph, FalkorDB, LadybugDB), the cross-model arms (Memgraph, LadybugDB, DuckDB tested), the lifecycle and plain-PostgreSQL time-series arms, the quantization survey, PG+AGE in the multi-model table, and the dense ingest/index split for DuckDB VSS and ArangoDB. COMPARATORS states the roster rule and records Chroma on sparse and FalkorDB on cross-model as not added, with the reasons checked 2026-10-02. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01WWcDr7TwvcfMsQkeYQVHZr
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.
Closes #66.
Markdown only: no code changes while the October chain runs (it pulls
mainbefore every stage).page_check.MAPPINGand_resolvegates are no longer documented, andpage_checkis described as it runs; heap ratios are stated fromHEAP_BY_SCALEagainstMEM_BY_SCALE; dev and SNAPSHOT rows are admitted only when the commit matches the pin; the runner lock's path; QuestDB and mongot run at their own default heap; pgvector's sparse index is built after its load; MongoDB's graph arm runs chained$lookup; the overrides inventory gains the live overrides it lacked; the retired synthetic document lane is no longer described as live; the launch block exports what a commit-pinned campaign needs.Checks:
structure_check.pygates and stage chain pass (its later stages need the bench host's results files and run at landing);version_pin_check.py0 disagreements; every table row keeps its header's column count; no em dashes added.🤖 Generated with Claude Code
https://claude.ai/code/session_01WWcDr7TwvcfMsQkeYQVHZr