TML-2808: namespace concretions address entities by coordinate#715
Merged
Conversation
Contributor
|
Important Review skippedToo many files! This PR contains 292 files, which is 142 over the limit of 150. To get a review, narrow the scope: ⚙️ Run configurationConfiguration used: Path: .coderabbit.yml Review profile: CHILL Plan: Pro Run ID: ⛔ Files ignored due to path filters (8)
📒 Files selected for processing (292)
You can disable this status message by setting the Use the checkbox below for a quick retry:
✨ Finishing Touches🧪 Generate unit tests (beta)
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. Comment |
@prisma-next/extension-author-tools
@prisma-next/mongo-runtime
@prisma-next/family-mongo
@prisma-next/sql-runtime
@prisma-next/family-sql
@prisma-next/extension-arktype-json
@prisma-next/middleware-cache
@prisma-next/mongo
@prisma-next/extension-paradedb
@prisma-next/extension-pgvector
@prisma-next/extension-postgis
@prisma-next/postgres
@prisma-next/sql-orm-client
@prisma-next/sqlite
@prisma-next/target-mongo
@prisma-next/adapter-mongo
@prisma-next/driver-mongo
@prisma-next/contract
@prisma-next/utils
@prisma-next/config
@prisma-next/errors
@prisma-next/framework-components
@prisma-next/operations
@prisma-next/ts-render
@prisma-next/contract-authoring
@prisma-next/ids
@prisma-next/psl-parser
@prisma-next/psl-printer
@prisma-next/cli
@prisma-next/cli-telemetry
@prisma-next/emitter
@prisma-next/migration-tools
prisma-next
@prisma-next/vite-plugin-contract-emit
@prisma-next/mongo-codec
@prisma-next/mongo-contract
@prisma-next/mongo-value
@prisma-next/mongo-contract-psl
@prisma-next/mongo-contract-ts
@prisma-next/mongo-emitter
@prisma-next/mongo-schema-ir
@prisma-next/mongo-query-ast
@prisma-next/mongo-orm
@prisma-next/mongo-query-builder
@prisma-next/mongo-lowering
@prisma-next/mongo-wire
@prisma-next/sql-contract
@prisma-next/sql-errors
@prisma-next/sql-operations
@prisma-next/sql-schema-ir
@prisma-next/sql-contract-psl
@prisma-next/sql-contract-ts
@prisma-next/sql-contract-emitter
@prisma-next/sql-lane-query-builder
@prisma-next/sql-relational-core
@prisma-next/sql-builder
@prisma-next/target-postgres
@prisma-next/target-sqlite
@prisma-next/adapter-postgres
@prisma-next/adapter-sqlite
@prisma-next/driver-postgres
@prisma-next/driver-sqlite
commit: |
size-limit report 📦
|
…s lift (TML-2808 D1) Move SQL/Mongo/Postgres/SQLite namespace slot maps under entries with singular keys (table, type, collection). Update framework walkers, validators, serializers, and in-scope family/target tests. Signed-off-by: Will Madden <madden@prisma.io>
…nate Signed-off-by: Will Madden <madden@prisma.io>
Rewrite downstream reads and test/fixture literals from flat namespace
slots (tables/collections/enum) to namespace.entries.{table,collection,type}.
Fold D1 review nits (storage cast removal, blindCast in serializers,
+: reject on SQL entries arktype). Refresh contract fixtures and parity
golden files for the new canonical storage shape.
Signed-off-by: Will Madden <madden@prisma.io>
Complete D2 verification fixes for entries-lifted storage shape. Signed-off-by: Will Madden <madden@prisma.io>
…ML-2808 D3) Rename kind to sqlite-namespace, install it non-enumerably, and wire sqliteCreateNamespace through TS and PSL authoring so storage namespaces are SqliteDatabase instances instead of generic sql-namespace literals. Signed-off-by: Will Madden <madden@prisma.io>
Consume familyInstance.deserializeContract return value instead of the pre-hydration enriched envelope, with a regression test pinning the flow. Signed-off-by: Will Madden <madden@prisma.io>
Mirror executeContractEmit: emit artifacts from deserializeContract output, not the pre-hydration enriched envelope. Signed-off-by: Will Madden <madden@prisma.io>
Regenerate integration contract fixtures and update tests that still mutated or asserted flat namespace.tables slots. Signed-off-by: Will Madden <madden@prisma.io>
Move sqliteAuthoringDefaults above overload declarations; narrow CLI emit regression test family mock for exactOptionalPropertyTypes. Signed-off-by: Will Madden <madden@prisma.io>
Signed-off-by: Will Madden <madden@prisma.io>
Main merged control-policy tests and postgres issue resolution still used flat namespace slots; switch to entries.collection / entries.type and entries.table in CLI migration fixtures. Signed-off-by: Will Madden <madden@prisma.io>
wmadden-electric
force-pushed
the
tml-2808-substrate-finishing
branch
from
8ffd2aa to
a955581
Compare
Re-emit contracts after namespace.entries reshape and align extension headRef hashes with descriptor self-consistency recomputation. Signed-off-by: Will Madden <madden@prisma.io>
Refresh end-contract snapshots and head refs after namespace.entries reshape so descriptor self-consistency and db-init graph-walk agree on storage identity. Signed-off-by: Will Madden <madden@prisma.io>
Re-emit paradedb, pgvector, and postgis extension contracts and align headRef / migration end-contract hashes with descriptor self-consistency. Signed-off-by: Will Madden <madden@prisma.io>
Signed-off-by: Will Madden <madden@prisma.io>
Refresh emitted contract fixtures, fix mongo migration/test stubs to use entries.collection, and update db-verify and react-router expectations for the namespace coordinate lift. Signed-off-by: Will Madden <madden@prisma.io>
wmadden
requested changes
Jun 4, 2026
D2 replaced query-builder ordering with no-ops or in-memory sort/slice. Reinstate collection helpers, route chaining, PostGIS tie-breaker, and e2e assertions that exercise SQL ORDER BY / LIMIT / OFFSET. Signed-off-by: Will Madden <madden@prisma.io>
Signed-off-by: Will Madden <madden@prisma.io>
Use where().orderBy() instead of forUser().newestFirst() so the narrowed collection type keeps orderBy; apply sqlite import sort from biome. Signed-off-by: Will Madden <madden@prisma.io>
wmadden
requested changes
Jun 4, 2026
Contributor
wmadden
left a comment
wmadden
left a comment
There was a problem hiding this comment.
Wtf is SqlNamespaceTablesInput? That looks like a holdover from the old structure that should have been stripped out
…ntries check Replace the hand-rolled length+index guard in omitDefaults with a named STORAGE_NAMESPACE_ENTRIES_PATTERN constant passed to matchesPathPattern, matching the style of the adjacent DOMAIN_MODEL_STORAGE_PATTERN. Signed-off-by: Will Madden <madden@prisma.io>
…teNamespaceEntrySchema The arktype schema already has `"+": "reject"` which rejects undeclared keys like `tables`, `enum`, and `collections` before the narrow predicate ever runs. The manual Object.hasOwn check is therefore dead code; remove it. Signed-off-by: Will Madden <madden@prisma.io>
…ables.test.ts The file only contains test utilities (fixture builders) consumed by other test files; renaming it to *.test.ts makes vitest discover it as part of the test suite and aligns it with the project naming convention. Signed-off-by: Will Madden <madden@prisma.io>
mongo-emitter: remove redundant `as Readonly<Record<string, MongoCollection>>`
cast on ns.entries.collection — the MongoStorage.namespaces type already
carries the correct Readonly<Record<string,MongoCollection>> shape.
sql-emitter: remove ad-hoc isPostgresSchemaNamespace predicate that cast
`ns.kind` via `{kind?: unknown}`. Namespace already exposes `kind: string`
directly, so the function simplifies to a `ns.kind === "schema"` guard at
the call site; the type slot is narrowed with a documented blindCast that
names the reason (the authoring entity registry materializes the concrete
instances).
Signed-off-by: Will Madden <madden@prisma.io>
…y namespace layer The `entries.type` slot in SqlNamespaceTablesInput / SqlNamespace was a postgres-specific pass-through baked into family-layer types. This commit lifts it to the postgres target: - SqlStorageTypeEntry drops PostgresEnumStorageEntry; SqlStorage.types now only holds StorageTypeInstance. normaliseTypeEntry no longer accepts the postgres-enum case. - SqlNamespaceTablesInput.entries and entries.table become required. The type slot is kept as entries.type?: unknown for target pass-through; the SQL family registry handles only the table slot. - SqlContractSerializerBase no longer owns the type-slot hydration loop; entityTypeRegistry is promoted to protected so postgres can access it. - PostgresContractSerializer extracts the type slot directly from the raw JSON envelope in its hydrateSqlNamespaceEntry override, bypassing the family-layer slot. The serializer also registers the postgres-enum factory. - planner-sql-checks: replace postgresCreateNamespace helper calls with direct PostgresSchema construction (no entries needed for schema-name qualification). - planner-strategies / contract-to-schema-ir / verify-sql-schema: use bracket access for the postgres-specific type slot so the family type annotation stays slot-agnostic. Signed-off-by: Will Madden <madden@prisma.io>
…nality
Make entries and entries.collection required on MongoNamespaceCollectionsInput
(parallel to the SQL family tightening). Drop every defensive ?.collection ?? {}
chain at consumption sites — buildMongoNamespaceMap, contractToMongoSchemaIR,
MongoContractSerializerBase — now that the type contract guarantees the slot.
The deserializer supplies { collection: {} } when raw JSON omits the entries
slot, so the required field is always satisfied before the IR is constructed.
Also fix contract-builder.per-model-namespace.test.ts: entries.table is now
required on SqlNamespaceTablesInput so the optional-chain on the stub accessor
was stale.
Signed-off-by: Will Madden <madden@prisma.io>
Postgres namespaces now emit __unbound__ explicitly (entries.table and entries.type required on PostgresSchemaInput), and field ordering within the namespace object changed (entries before id/kind). SQLite/Mongo fixtures pick up the equivalent ordering normalization. Storage hashes updated to reflect the entries-required shape. Signed-off-by: Will Madden <madden@prisma.io>
…ries.table?. optional chains Signed-off-by: Will Madden <madden@prisma.io>
wmadden
reviewed
Jun 4, 2026
…lizeAndFreezeEntries, drop postgres type slot from SQL family, tighten PostgresSchemaInput, inject postgresCreateNamespace
Theme A: delete the over-engineered materializeAndFreezeEntries helper and
inline deep-freeze logic directly in SqlBoundNamespace, PostgresSchema, and
SqliteUnboundDatabase constructors.
Theme B1: remove the type? slot from SqlNamespaceTablesInput.entries and
SqlNamespace.entries — postgres-specific shape does not belong in family code.
Theme B2: remove PostgresEnumStorageEntry from the family-level BuiltContract
type in contract-types.ts.
Theme B3: make PostgresSchemaInput.entries.type required (not optional);
update postgresCreateNamespace to always supply it (defaulting to {}).
Theme B4: remove defensive entries? optional chains, rewrite PSL enum tests
to capture enumTypes via a buildEnumCapturingFactory helper, fix family-level
schema-verify test helper to use a TestEnumNamespace that correctly stores
the type slot, and add isPostgresSchema narrowing where TS4111 fired.
Also inject createNamespace: postgresCreateNamespace in the postgres extension
define-contract.ts and define-config.ts (and PSL parity config templates /
fixtures) so PostgresSchema instances are produced consistently across both TS
and PSL authoring paths — fixing the fixtures:check discrepancy where enum
type data was silently dropped by the family-level buildSqlNamespace fallback.
Signed-off-by: wmadden-electric <286902546+wmadden-electric@users.noreply.github.com>
Signed-off-by: Will Madden <madden@prisma.io>
…Namespace; enum type slot now present Signed-off-by: wmadden-electric <286902546+wmadden-electric@users.noreply.github.com> Signed-off-by: Will Madden <madden@prisma.io>
…anner test helper Signed-off-by: Will Madden <madden@prisma.io>
Merge origin/main (21 commits) into tml-2808-substrate-finishing. Resolve provider.ts by keeping both createNamespace (substrate) and defaultControlPolicy (control-policy); accept main's deletion of projects/control-policy/plan.md. Signed-off-by: Will Madden <madden@prisma.io>
Signed-off-by: Will Madden <madden@prisma.io>
After merging main: fix unbound-tables import in control-policy test, format contract-factories.ts for biome, and regen paradedb/pgvector/postgis migration metadata to match substrate storage hashes. Signed-off-by: Will Madden <madden@prisma.io>
Signed-off-by: Will Madden <madden@prisma.io>
Emit deserializes contracts through the SQL family serializer, which injects the empty __unbound__ namespace before serializing to JSON. Authoring previously hashed storage without that slot, so extension contractSpace descriptors failed self-consistency checks and blocked emit for any config loading extension packs. Signed-off-by: Will Madden <madden@prisma.io>
Regenerate extension contract spaces, migration head refs, and downstream fixtures so published storageHash matches descriptor self-consistency checks. Signed-off-by: Will Madden <madden@prisma.io>
Signed-off-by: Will Madden <madden@prisma.io>
Signed-off-by: Will Madden <madden@prisma.io>
createNamespace may return a target namespace concretion; blindCast keeps build-contract assignable while preserving runtime SqlNamespace shape. Update namespace builder tests to expect the materialised __unbound__ slot. Signed-off-by: Will Madden <madden@prisma.io>
Postgres enum types live under namespace entries.type after the namespace substrate change; codecRefForStorageColumn still looked up the retired ns.enum slot, leaving enum ParamRefs without codecs during migration emit. Signed-off-by: Will Madden <madden@prisma.io>
Signed-off-by: Will Madden <madden@prisma.io>
Establishes that contract.json / contract.d.ts / expected.contract.json are emitter-owned: agents must regenerate via `pnpm fixtures:emit` rather than write or patch these files by hand. Captured after a babysit subagent attempted to hand-craft fixtures and produced drift the next emit run reverted. Signed-off-by: Will Madden <madden@prisma.io>
CI was failing on two fronts:
1. `test/unbound-tables.test.ts` matched vitest discovery but contained
zero suites — it is a helper module consumed by 5 sibling tests.
Treating it as a discoverable test file was the wrong fix for the
filename-convention complaint that prompted the rename; the helper
is renamed back to `unbound-tables.ts` and the 5 imports updated.
2. Five bare `as` casts in `src/` failed the `no-bare-cast` plugin:
- `interpreter.ts:1711` widened `ContractModelBase`→`ContractModel`;
since the latter only restates `fields`, the cast is purely
structural and removable.
- `psl-column-resolution.ts` x3 walked `unknown` after a
`typeof/null/Array.isArray` guard; replaced with `blindCast`
carrying a reason that names the preceding narrow.
- `psl-relation-resolution.ts` widened a string lookup into a
`keyof typeof REFERENTIAL_ACTION_MAP`; the map is now typed
`Record<string, ReferentialAction | undefined>` at declaration so
the lookup carries the right type without an `as`.
`pnpm lint:casts` confirms the global cast count fell by 26.
Signed-off-by: Will Madden <madden@prisma.io>
The previous substrate cascade added `entries: { table: { ... } }`
wrappers around table literals in test files but did not reflow the
inner indentation. Biome `--error-on-warnings` rejects the resulting
mis-indented bodies in CI. This commit is the pure auto-format catch-up
(no semantic changes — `pnpm exec biome check --write` output).
Signed-off-by: Will Madden <madden@prisma.io>
CI lint:rules rejected the rule for missing the description+alwaysApply frontmatter block that the validator requires of every .mdc rule. Signed-off-by: Will Madden <madden@prisma.io>
…t shape
The substrate move from `base: "ModelName"` / `relations.X.to: "ModelName"`
strings to `{ namespace, model }` objects (alongside the namespace
envelope reshape that landed earlier) left the backfill migration’s
frozen contract snapshots in the old shape. The MongoContractSerializer
now rejects them on load:
Contract structural validation failed:
domain.namespaces.__unbound__.models.AddToCartEvent.base must be an
object (was a string)
...
Programmatically lifted every `models.<M>.base` string and every
`models.<M>.relations.<R>.to` string in both `start-contract.json` and
`end-contract.json` to `{ namespace: "__unbound__", model: "<M>" }`.
The transform touches `domain` only — `storageHash` derivation excludes
`domain` (`hashContract` zeroes it before canonicalising), so all
migration / chain hashes stay valid and no downstream consumer needs an
update.
A general regen-example-migrations tool is the proper fix for the next
substrate shape change — filed as a follow-up rather than adding to the
substrate PR.
Signed-off-by: Will Madden <madden@prisma.io>
The new never-hand-edit-contract-fixtures rule pushed three rules footprint metrics over the previously-pinned thresholds. Trimmed the rule (60→39 lines) to its load-bearing content — the prohibition, the correct tool, what counts as forbidden, the commit pattern, and the narrow escape hatch — and bumped the thresholds by ~5% so the addition fits without reverting other rules. Signed-off-by: Will Madden <madden@prisma.io>
PR touches examples/ and packages/3-extensions/ via fixture/format churn from the substrate reshape (namespace.entries envelope + CrossReference object shape). No consumer code change is required — re-emit produces the new shape directly. Record explicit changes: [] in both upgrade skills with an inline note so the per-pr-declaration check sees the intentional incidental classification. Signed-off-by: Will Madden <madden@prisma.io>
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
2 participants
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.
Summary
Every table, enum, and collection in a Prisma contract has an address:
(schema, kind, name). The framework already hands you these addresses when you walk a contract. Until now you could not use the address directly to find the thing — you had to translatetable→.tables,enum→.enum, and so on. Every walker carried that translation table.This PR fixes that. Entity-kind buckets use singular essence names (
table,type,collection) and live under oneentriescontainer on each namespace concretion, so resolvingnamespaces.public.entries.table.Useris the direct path. Theentrieswrapper is a structural fence: namespace metadata can grow beside entity buckets without colliding with them, and extension-contributed kinds land underentries.<kind>without re-planning the namespace surface.contract.jsonon disk follows the same shape.While touching this surface we also:
SqliteDatabase/SqliteUnboundDatabaseclass instances with non-enumerablekind: 'sqlite-namespace'(parity with Mongo/Postgres).deserializeContract's hydrated contract reachesemit, not the pre-hydration plain envelope (executeContractEmitandControlClient.emit).The diff is wide (~200+ files) but almost entirely mechanical rewrites of the same pattern.
Design rationale: ADR 224.
Test plan
pnpm typecheck(135/135)pnpm test:packages(721/723 files;@prisma-next/cli-telemetryfails locally —prisma-nextnot on PATH in exampleemitscripts; infra wart)pnpm test:integration(172/179 files; remaining failures are mostly DB/e2e journeys, pgvector contract-space, and timeouts — not namespace-shape regressions)pnpm lint:depspnpm fixtures:check— blocked by knownprisma-next: command not foundin rootfixtures:emit; integrationpnpm emitused as workaround