syntaqlite — Competitive Comparison

SQLite SQL tooling landscape.

Parser Comparison

Per-statement SQLite SQL parsing accuracy, validated against sqlite3 as ground truth.

Ground Truth

Validating all test statements against sqlite3:

40/40 statements validated by sqlite3.

Parser Accuracy

Legend: PASS = correctly parses valid SQL, FAIL = rejects valid SQL, FP = accepts invalid SQL

Scoreboard

Parse Speed

• bench.sql: 117 lines, 3,545 bytes
• bench_30x.sql: 3510 lines, 106,350 bytes

bench.sql (1x)

bench_30x.sql (30x)

Formatter Comparison

Round-trip correctness (format then validate with sqlite3) and speed.

Ground Truth

40/40 statements validated by sqlite3.

Round-Trip Validation

For each formatter: does the formatted output still pass real SQLite?

Scoreboard

Corruption Details

Format Speed

• bench.sql: 117 lines, 3,545 bytes
• bench_30x.sql: 3510 lines, 106,350 bytes

bench.sql (1x)

bench_30x.sql (30x)

Slow Tools (single timed run)

Validator Comparison

Error detection accuracy and diagnostic quality.

Diagnostic Quality

A realistic query with subtle errors — how does each tool report them?

Query (2 errors: CTE declares 3 columns but SELECT produces 2; typo ROUDN):

WITH
  monthly_stats(month, revenue, order_count) AS (
    SELECT
      STRFTIME('%Y-%m', o.created_at) AS month,
      SUM(o.total) AS revenue
    FROM orders o
    WHERE o.status = 'completed'
    GROUP BY STRFTIME('%Y-%m', o.created_at)
  )
SELECT
  ms.month,
  ms.revenue,
  ms.order_count,
  ROUDN(ms.revenue / ms.order_count, 2) AS avg_order
FROM monthly_stats ms
ORDER BY ms.month DESC
LIMIT 12;

syntaqlite

Static semantic analysis — offline, no database needed. Finds both errors in one pass:

error: table 'monthly_stats' has 2 values for 3 columns
  --> /var/folders/rx/t6_rqmqx0f15l7kgp7yjhcbc0000gn/T/tmpimbv0__j.sql:29:3
   |
29 |   monthly_stats(month, revenue, order_count) AS (
   |   ^~~~~~~~~~~~~
warning: unknown function 'ROUDN'
  --> /var/folders/rx/t6_rqmqx0f15l7kgp7yjhcbc0000gn/T/tmpimbv0__j.sql:41:3
   |
41 |   ROUDN(ms.revenue / ms.order_count, 2) AS avg_order
   |   ^~~~~
   = help: did you mean 'round'?

sqlite3

Runtime execution — stops at first error:

Error: in prepare, table monthly_stats has 2 values for 3 columns

sqlite-runner-lsp

Runtime via LSP — wraps sqlite3, same single error:

28:1 error Parse error: table monthly_stats has 2 values for 3 columns

sql-lint

Structural checks only:

/var/folders/rx/t6_rqmqx0f15l7kgp7yjhcbc0000gn/T/tmpum1hr0xz.sql:1 sql-lint was unable to lint the following query "WITH...

Error Detection Accuracy

Schema: users, orders, products, order_items. Ground truth: sqlite3.

Scoreboard

Validation Speed

• bench.sql: 117 lines, 3,545 bytes (+ schema preamble)
• bench_30x.sql: 3510 lines, 106,350 bytes (+ schema preamble)

bench.sql (1x)

bench_30x.sql (30x)

LSP Comparison

Feature testing for SQLite-aware language servers.

Tested Capabilities

Each server is started, sent a test file, and probed for completion, hover, diagnostics, and formatting. Results are from actual LSP responses.

Diagnostic Detail

What each server reports for SELEC * FROM users; (syntax error):

syntaqlite

1:1 error syntax error near 'SELEC'

sqls

(no diagnostics)

sql-language-server

1:2 error Expected "$", "(", "--", "/*", "ALTER", "CREATE TABLE", "CREATE", "DELETE", "DROP TABLE", "DROP VIEW", "DROP", "INSERT", "REPLACE", "SELECT", "UPDATE", "WITH", "return", [ \t\n\r], or end of input but "S" found.

LSP Startup + Response Speed

Time to start server, send document, receive diagnostics, and exit: