docs: stale-information audit applied to the repository Markdown - #47
Merged
Conversation
…stics brought up to the 2026-09-03 batch The 2026-09-03 batch (SingleStore, SAP HANA Express, Exasol, Altibase, Kinetica, Actian Ingres, IBM Db2 for i) left present-tense claims behind it in the Markdown. This is the audit's confirmed in-repo findings, applied: - Counts: 53 Linux / 45 macOS / 48 Windows, 24 distinct drivers, 27 upstream reports across fifteen projects (three fixed), twelve MySQL-wire entries through one Connector/ODBC, fourteen PostgreSQL-wire through psqlodbc. - docs/reference/quirks.md: a row for each of the six new drivers with the flags OdbcDetectQuirks actually sets, iiodbcdriver added to the autodetected sqllen_32bit rule, and glossary entries for wide_sql, ddl_timestamp_type_name, binary_param_as_varchar, null_decimal_param_as_char, zero_row_suffix and decimal_fixed_precision/scale. - docs/TROUBLESHOOTING.md: the six new Linux driver libraries measured with readelf -r ... R_X86_64_TPOFF64 (all zero), so the static-TLS table is now 24 rows and "exactly one has any" still holds. - prefer_param_arrays is four drivers now (MariaDB < 3.2, Vertica, Altibase, SAP HANA) and wchar_as_utf8 two more (Altibase, Ingres) — rust.md, performance.md, types.md. - bench: LANGUAGE_BENCHMARKS.md statistics recomputed from the table as it stands (47 all-five fetch rows, 40 all-five ingest); RUST_BENCHMARKS.md provenance and the ten rows with no ADBC ingest rewritten from the table; MATRIX_BENCHMARKS.md says which rows are the 2026-09-04 run; the macOS, Windows and bench/README.md headline figures reconciled. - No binding or project is called "official" any more (docs/languages/csharp.md). Dated campaign sentences are left as history throughout. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_013ZWm8DfxFwdn7LyyVAM9s5
singhpratech
added a commit
that referenced
this pull request
Sep 4, 2026
…nts, bench statistics, no "official" Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_013ZWm8DfxFwdn7LyyVAM9s5
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.
A stale-information audit over the repository's Markdown turned up 35 findings; 23 of them concern files in this repo and are applied here. The 12 about
gh-pages(index.html,bench/index.html,sitemap.xml,status.json) and the one about pkg.go.dev are left to the session that owns the site and the release tags.Almost all of them are the same shape: the 2026-09-03 batch added SingleStore, SAP HANA Express, Exasol, Altibase, Kinetica, Actian Ingres and IBM Db2 for i, and sentences written in the present tense before it never moved. Dated campaign records are untouched.
Counts reconciled to the source of truth
53 databases on Linux, 45 macOS, 48 Windows; 24 distinct drivers; 27 upstream reports across fifteen projects, three fixed by their maintainers plus unixODBC's mitigation.
README.md— the upstream bullet named three projects "so far"; it names the record now.docs/community/faq.md— the database enumeration gained the seven; the upstream paragraph the other 24 reports.docs/ROADMAP.md— "measured against all 46" → all 53 (261 of 265 Linux cells); SAP HANA dropped from the vendor-download help-wanted row (it is verified on all three OSes); the manifest anddriver="odbc"by name scoped to Windows as well, sincecmake --installwrites it there.bench/BENCHMARKS.md— psqlodbc drives fourteen PostgreSQL-wire databases, not eleven; the "eight drivers in the compat matrix" sentence put in the past tense it was written in.Driver documentation caught up with
src/odbc_driver.cdocs/reference/quirks.md— a row for each ofaltibase,libodbchdb,kinetica,iiodbcdriver,exaodbcandcwbodbc+DB2/400, listing the flagsOdbcDetectQuirksactually sets and the reason from the same block's comment;iiodbcdriveradded to the autodetectedsqllen_32bitrule; glossary rows forwide_sql,ddl_timestamp_type_name,binary_param_as_varchar,null_decimal_param_as_char,zero_row_suffixanddecimal_fixed_precision/decimal_fixed_scale.prefer_param_arraysis set for four drivers today (MariaDB Connector/ODBC before 3.2, Vertica, Altibase, SAP HANA) —docs/languages/rust.md,docs/how-it-works/performance.md, and the glossary.wchar_as_utf8gained Altibase and Actian Ingres indocs/reference/types.md, which also gained bind rows for Exasol's two parameter quirks.docs/languages/csharp.mdno longer calls the Apache .NET binding "official"; nothing in the repo's prose does.Static TLS table measured, not estimated
readelf -r <lib> | grep -c R_X86_64_TPOFF64was run on all six new Linux driver libraries (SAP HANA client, Exasol, Altibase, Kinetica, Ingres, IBM i Access). All six report 0, sodocs/TROUBLESHOOTING.mdnow has 24 rows and "exactly one has any" holds as stated. The affected library is shared by twelve matrix entries now, SingleStore included.Benchmark statistics recomputed from the tables
bench/LANGUAGE_BENCHMARKS.md— recounted from the Results table: 47 databases have an ADBC fetch in all five languages (was 42), spreads ≤1.25× on 28 and ≤1.5× on 38;adbc-driver-jniis slowest on 20 of the 47, Python next at 10; 40 rows have an ingest number in all five, Python slowest on 13, Java 9, Rust 8.bench/RUST_BENCHMARKS.md— the provenance sentence covers the seven rows taken 2026-09-04; the "four databases have no ADBC column" paragraph rewritten from the table as it stands: every row has an ADBC fetch, ten have no ADBC ingest, and the cratedb / spanner / duckdb gaps it described are closed.bench/MATRIX_BENCHMARKS.md— no rows were lost (7a6a521andmainare byte-identical), so nothing was restored; a note now says which nine rows are the 2026-09-04 run and which twelve are from 2026-08. It sits in the prose tail the generator preserves, so a re-run keeps it.bench/README.md— the matrix file is 21 databases, not 46; the macOS language file covers 44 of the 45 macOS-passing databases (Db2 for i not measured there).bench/BENCHMARKS-windows.md— the status headline keeps its dated figures and gains the third measurement: 48 of 53 databases, 219 of 240 cells.bench/LANGUAGE_BENCHMARKS-macos.md— 215 of 225 cells, with 191 kept as the count before the sixth batch.bench/BENCHMARKS-macos.md— batch 2 is not "(in progress)"; it closes in its own section.Docs only: no change under
src/,tests/or any package.🤖 Generated with Claude Code
https://claude.ai/code/session_013ZWm8DfxFwdn7LyyVAM9s5