Skip to content

build(deps): pin SQLQuack 353861ea, after its layout refactor - #48

Merged
kvlonge merged 1 commit into
mainfrom
build/sqlquack-layout
Nov 9, 2026
Merged

kvlonge merged 1 commit into
mainfrom
build/sqlquack-layout

Conversation

@kvlonge

@kvlonge kvlonge commented Nov 9, 2026

Copy link
Copy Markdown
Contributor

Why

SQLQuack is about to restart its history from a clean root, so SQLClean's current pin, 804864d2, will stop being reachable. This moves SQLClean to SQLQuack main 353861ea, which is after the layout refactor, with no behaviour change beyond what that main brings.

Changes

  • Pin every SQLQuack crate at 353861ea. Cargo.lock resolves all 11 SQLQuack packages to that one commit; there is no polyglot.
  • SQLQuack now builds each dialect only with its feature, so SQLClean enables the ones it reads:
    • lint: bigquery, databricks, duckdb, postgres, snowflake, sqlite;
    • format: bigquery, duckdb, postgres, snowflake, sqlserver (Databricks stays unavailable before SQLQuack is called);
    • fingerprint: bigquery, duckdb (which also builds generic), postgres, snowflake, sqlite, sqlserver.
  • The shared lossless tokenizer moved out of sqlquack-syntax; it is now reached through sqlquack-parse, with no features.
  • A test checks that every dialect the configuration page names is checked, formatted and fingerprinted as it says, and never refused as not built.

Verification

  • Corpus diff against 0.6.2, on the same machine: all 1,066 format-corpus queries formatted at no width and at width 80 (2,132 responses), their fingerprints (1,066), and lint (1,340 runs: each query in its dialect, the PostgreSQL ones also as generic and sqlite) are byte-identical, 4,538 of 4,538 lines.
  • The new dialect test fails when any one feature is left out: lint without sqlite, format without sqlserver, and fingerprint without bigquery each fail it, with SQLQuack's "built without this dialect's feature" message.
  • make verify: fensu 0 faults; cargo 41 + 8 passed; pytest 72 passed.

SQLQuack now builds each dialect only with its feature, so SQLClean
enables the ones it reads: lint for BigQuery, Databricks, DuckDB,
PostgreSQL, Snowflake and SQLite; format for BigQuery, DuckDB,
PostgreSQL, Snowflake and SQL Server; fingerprints for every dialect
they name. The shared lossless tokenizer is now reached through
sqlquack-parse. A test checks that every dialect the configuration page
names is checked, formatted and fingerprinted as it says.

Behaviour is unchanged: across the 1,066-query format corpus, formatting
at no width and at 80, fingerprints, and lint on every dialect are
byte-identical to 0.6.2.
@cloudflare-workers-and-pages

Copy link
Copy Markdown

Deploying sqlclean with  Cloudflare Pages  Cloudflare Pages

Latest commit: bf7148e
Status: ✅  Deploy successful!
Preview URL: https://3dfff7b9.sqlclean.pages.dev
Branch Preview URL: https://build-sqlquack-layout.sqlclean.pages.dev

View logs

@kvlonge
kvlonge merged commit ff5ea5a into main Nov 9, 2026
14 checks passed
This was referenced Nov 9, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant