[artile]

Test and demo data is hand-written today, although each column type already has an Effect Schema that knows its values. This PR adds a seed entry that makes rows from those Schemas through Effect's Schema → Arbitrary.

Closes #88.
Part of #52.

Changes

• New entry @maple-dev/effect-orm/seed with rows, insert, arbitrary, pick and SeedError. It uses Arbitrary from effect@4.0.0. There is no new dependency, and fast-check is not used at runtime.
• rows(table, { count, seed }) returns decoded rows. The same seed gives the same rows. Each column samples from its own seed (the run's seed plus the column key), so a new column does not change the values of the other columns.
• The built-in column types add the range that their SQL type stores: integer widths, float4, DateTime / DateTime64 / timestamptz years, lowercase uuid, no U+0000 in Postgres text and jsonb. No lone surrogates, no -0. A branded or refined column gets only values that its Schema accepts.
• columns overrides a column with a value, an Arbitrary (pick(ids) for a foreign key), or a function of the row index. A wrong override value is a SeedError.
• unique columns get distinct values. A Postgres table's one-column primary key and unique: true columns are added from its DDL.
• A table with a tenant column needs tenant (all rows in one tenant, so the insert compiles as "single-tenant") or crossTenant: true (the insert declares crossTenant()).
• insert(table, options) writes the rows through insertInto(table).values(...) on the Database in context, batchSize rows per statement (default 100), and returns them.
• arbitrary(table) gives whole rows to Arbitrary.checkEffect for property tests.
• Defaulted columns are left to the database unless overridden. Computed columns are never written.
• chDateTime64Literal in src/ch/types.ts is now exported inside the package (not from a public entry).
• Docs: new docs/seeding.md, rows in docs/README.md, docs/reference.md and README.md. One changeset (minor).

Left out

• Seeding a whole schema in foreign-key order (a seed(schema) over many tables). This PR seeds one table at a time; a child table picks its keys from the parent rows with pick. The graph walk can come next if you want it.
• Distinct values over several columns (composite uniques). Use a function of the row index for one of the columns.
• Realistic value generators (names, emails, cities) as in drizzle-seed. Overrides cover this.
• A effect-orm seed CLI command. The issue names it, but the library API comes first.
• SQLite, MySQL and D1 storage ranges: these dialects are not on main yet. A custom column type gets values from its Schema alone.

Validation

All commands ran on this branch, with a live ClickHouse 26.8.2.7 (EFFECT_ORM_CLICKHOUSE_URL, _USER, _PASSWORD set):

• bun run build: pass.
• bun run typecheck: pass.
• bunx vitest run --testTimeout=60000 --hookTimeout=60000: 62 files passed, 1053 tests passed, 0 skipped.
• node scripts/check-doc-citations.mjs, node scripts/check-exports-documented.mjs, node scripts/check-doc-examples.mjs: pass.
• bun scripts/check-package.ts: pass (14 entries load from the tarball under CommonJS).
• New tests: src/seed/seed.test.ts (offline rows, and inserts on PGlite that read back as seeded for every built-in Postgres type), src/seed/seed.test-d.ts (row and override types), tests/seed.clickhouse.test.ts (every built-in ClickHouse type on a live server, rows read back as seeded). I also ran the live insert tests with 30+ other seeds; all passed.

Upstream CI for a fork PR waits until a maintainer approves the run.

---

^{Need help on this PR? Tag @codesmith-bot with what you need.}

[coderabbitai]

Warning

Review limit reached

You've used all free OSS reviews for now. Wait for the free limit to reset to keep reviewing this public repository.

Next included review available in 24 minutes.

Check out review usage here.

View limit details

Limit details: You’ve used the included review currently available.

Learn how review limits work.

Review configuration:

Run configuration

• Source mode: Auto
• Repository: Unavailable
• Branch: Unavailable
• File: No configuration file from this repository applied
• Commit: Unavailable
• Layers: defaults
• Review profile: CHILL
• Plan: Advanced
• Run ID: a3070266-742c-490d-b278-f0dbcbc71e84

Commits

Reviewing files that changed from the base of the PR and between b57faf1 and 76e4514.

Files selected for processing: 16

• .changeset/seed-from-schemas.md
• README.md
• docs/README.md
• docs/reference.md
• docs/seeding.md
• package.json
• src/ch/types.ts
• src/docs-examples.test.ts
• src/seed.ts
• src/seed/seed.sqlite.test.ts
• src/seed/seed.test-d.ts
• src/seed/seed.test.ts
• src/seed/seed.ts
• tests/package-consumer.mts
• tests/seed.clickhouse.test.ts
• tsdown.config.ts

• Autofix · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

Share

• X
• Mastodon
• Reddit
• LinkedIn

[Makisuo]

Merge-readiness pass (merge ccac6bd, then 20e55bf, 0646d34 and 76e4514, no rebase):

• Merged current main with no conflicts.
• Bug fix for SQLite column types, which main now has:
  • SQLite.integer columns always failed. Their Schema is a Number with a safe-integer check, so Arbitrary.schema could only reach integers by filtering and gave up. Seed.rows returned SeedError: the column's arbitrary rejected too many values for any table with an integer column. They now sample the safe-integer range (±2^53−1, what node:sqlite and D1 read back), as int8 does. SQLite.boolean is also an integer column: it decodes to booleans, so it keeps its own Schema.
  • SQLite.timestamp (stored as text) now samples the full 0000–9999 range its text covers. Before, it clustered within milliseconds of 1970.
  • declaredUnique now also reads SQLite DDL, so a SQLite table's one-column primary key and unique columns get distinct values, as on Postgres.
• New test src/seed/seed.sqlite.test.ts:
  • Offline checks of the ranges: safe ints, 0000–9999 with values far from 1970, and distinct PK/unique.
  • Every built-in SQLite type (integer, text, nullable text, real, boolean, timestamp, nullable timestamp, blob, json, plus a defaulted column) inserted through Seed.insert. The table is created from its own DDL on node:sqlite (5 seeds) and on a local Miniflare D1 binding. Every row reads back equal to what was seeded.
  • I also ran the node:sqlite round trip once with 40 extra seeds; all passed. (The commit message calls that run "off-line"; it ran on node:sqlite.)
• CI fix: tests/seed.clickhouse.test.ts failed every time on ClickHouse 26.2. I reproduced it on a local 26.2 server: its INSERT ... VALUES parser reads some Float64 values one ulp off, even with precise_float_parsing=1 (26.8 reads them exactly). The test now compares Ratio within a few ulps and every other column exactly. seeding.md notes this.
• Docs (seeding.md range table and unique section) and the changeset now cover SQLite.

Verification (local, ClickHouse 26.8):

• bun run build, bun run typecheck: pass.
• bun run test: 1456/1456 tests passed, plus the doc checks. An earlier run of the same suite hit the known Postgres "division by zero" generated-fixtures fuzz flake (test(fuzz): fix three flaky generated-data properties #131).
• Live tests/seed.clickhouse.test.ts: passed on 26.8 and on a local 26.2.
• bun scripts/check-package.ts: pass (15 CJS entries).
• FUZZ_SCALE=0.3 bunx vitest run tests/fuzz: 269/269 passed.

CI run 41476791340 is green on both ClickHouse jobs. The first attempt failed on 26.8 only with the known fuzz flakes (#131) and a 5 s timeout in src/migrate/pg-enums-uniques.test.ts (PGlite start-up, a file this PR does not touch). That timeout also hit an earlier run on this branch, so it may be load-sensitive with the extra PGlite seed test.

Interaction note: if #109 (casing) and #111 (client defaults) merge, the seed still maps uniques through sqlName, so casing is fine. A defaultFn column, however, would be generated by the seed rather than left to the client default, because table.defaults lists only database defaults. That is worth a follow-up once both are on main.