@crvouga/mockingbird-service-postgres 3.5.0 → 3.7.0

package/CHANGELOG.md

@@ -1,5 +1,17 @@
 # Changelog — @crvouga/mockingbird-service-postgres
 
+## 3.7.0 (2026-10-17)
+
+### Features
+
+- resolve fleet and emulator issue batch ([26d2f8c](https://github.com/crvouga/mockingbird/commit/26d2f8cee99fefed93192204bac44fb1be35c1da))
+
+## 3.6.0 (2026-10-15)
+
+### Features
+
+- resolve SQL, routing and writer gaps ([73f97d0](https://github.com/crvouga/mockingbird/commit/73f97d0d2379f44b9131180498f4fab76648fed9))
+
 ## 3.5.0 (2026-10-15)
 
 ### Features

package/COMPATIBILITY.md

@@ -96,6 +96,9 @@ The oracle exposes **2787 builtin functions** and **74 operators** in `pg_catalo
 | `COMMENT ON INDEX`, `pg_description`, `obj_description` | VERIFIED | Stored on the index: replace, clear with `NULL` or `''`, removed with the index / table / schema, kept through renames, transactional, in snapshots, isolated per wire transaction; `42P01` / `3F000` / `42809` for a missing index, a missing schema, a relation that is not an index. Comments on every other kind of object are still accepted and not stored (divergence `comment-on-not-stored`) (`tests/contract/catalogs/index-comments.test.ts`) |
 | `regnamespace`, `to_regnamespace` | VERIFIED | Input (`-`, an all-digit oid, or exactly one identifier, case-folded unless quoted), output (quoted like `quote_ident`, numeric when no schema has the oid, `-` for zero), casts from text / name / oid / integers, implicit use as `oid` in comparisons, `pg_toast` / `pg_catalog` / `information_schema`; `3F000`, `42602` and `22003` as PostgreSQL raises them. `public` has a generated oid, not 2200 (`tests/contract/catalogs/regnamespace.test.ts`) |
 | `pg_stat_user_tables`, `pg_stat_user_indexes`, `ANALYZE` | PARTIALLY VERIFIED | PostgreSQL 18's columns and types; one row per user table and materialized view (and per index), `relid` joins `pg_class.oid`, follows create / drop / rename / schema moves, restored with snapshots. `n_live_tup` is the exact row count, immediately; `ANALYZE` validates its targets (`42P01`, `42703`, `42701`, `42601`) and records `analyze_count` / `last_analyze` from the database clock. Not tracked (read zero / NULL): scans, tuples inserted / updated / deleted, dead tuples, vacuum; `idx_scan` counts only the unique-key lookups made for `SELECT`. `VACUUM ANALYZE` does not count. Temporary tables are listed under the schema they were created in, not `pg_temp` (divergence `pg-stat-counters-immediate`) (`tests/contract/catalogs/table-statistics.test.ts`) |
+| Storage-default session settings | VERIFIED | `default_tablespace` (empty/database default or `pg_default`) and `default_table_access_method` (`heap`), SHOW/current_setting/set_config/RESET and transaction-local scope. Unknown methods/tablespaces fail with `22023`. Physical storage placement is not modelled (`tests/contract/misc/storage-defaults.test.ts`) |
+| Materialized view indexes | PARTIALLY VERIFIED | Ordinary/partial/unique indexes, creation before population, catalog links, unique enforcement and atomic failure on refresh, snapshot roundtrip. Physical access methods and concurrent refresh are not modelled (`tests/contract/views/materialized-indexes.test.ts`) |
+| Constraint triggers | PARTIALLY VERIFIED | AFTER ROW multi-event triggers, DEFERRABLE INITIALLY DEFERRED/IMMEDIATE, final-state execution at commit, failed-commit rollback, SET CONSTRAINTS and savepoints; definitions survive snapshots, pending events cannot be exported. Uses the existing plpgsql-lite subset (`tests/contract/triggers/deferred.test.ts`, `tests/wire/deferred.test.ts`) |
 | Triggers (row-level, LANGUAGE sql-expressible) | PARTIALLY VERIFIED | Creation-order firing, `UPDATE OF` ignored, no INSTEAD OF (documented) |
 | CREATE FUNCTION LANGUAGE sql / plpgsql-lite | PARTIALLY VERIFIED | Scalar + set-returning; plpgsql subset: DECLARE, EXCEPTION WHEN others, CASE, FOR-IN-SELECT, RETURN NEXT |
 | Text search (tsvector / tsquery / @@ / ts_rank) | PARTIALLY VERIFIED | `simple`-style config; no ispell/synonym dictionaries |
@@ -138,3 +141,34 @@ PostgreSQL; clock profiles are supported. This adds fleet orchestration around e
 snapshot semantics, not a broader SQL compatibility claim. Acceptance evidence is
 `packages/adapters/node/fleet.management.test.ts` using the unmodified `pg` wire client.
 See [Fleets](../../../docs/FLEETS.md).
+
+## Schema preflight and row-version metadata
+
+`analyzeSchema(sql, options)` (the package's main export) checks plain schema SQL in a discarded
+in-memory database and returns a JSON-serializable report, including the installed build version,
+statement offsets/lines/columns, object names, SQLSTATEs, independent failures and identifiable
+missing prerequisites. Defaults cap input at 1 MiB, statements at 1,000 and diagnostics at 100;
+configurable hard caps are 4 MiB, 10,000 and 1,000. It does not execute data statements, contact an
+oracle or change an existing database. Accepted statements remain explicitly **unknown** for full
+semantic compatibility, including deferred SQL/PL/pgSQL bodies and dynamic SQL. A detected gap sets
+`compatible: false`; otherwise it stays `null`, never a claim of a verified import. Source-tree
+execution identifies the version as `source-tree`; published builds embed their package version.
+Lexical errors that prevent reliable statement recovery terminate the scan.
+
+SQL and PL/pgSQL function definitions retain comma-separated `SET` lists (including `search_path`),
+apply them during calls and restore the caller's settings on return or error. Portable snapshots
+retain these settings. Sequence functions accept explicit `regclass` input, and
+`pg_get_serial_sequence` finds owned serial, identity and explicitly owned sequences using exact
+column-name case; ownership follows sequence/table/column renames and snapshots.
+
+Stored tables expose `xmin` and `alias.xmin` as hidden `xid` system columns. INSERT, UPDATE, upsert,
+COPY and referential-action writes stamp row versions with a local transaction identity; explicit
+transactions share that identity, including wire workspace replay and commit. Rollback restores
+committed row metadata. Snapshots preserve identities and continue allocation above retained IDs;
+legacy PGMM snapshots without metadata expose frozen xid `2`. Each database/namespace owns its
+allocator. These identities are version markers, not permanent globally unique row identifiers.
+The implementation does not add full MVCC, VACUUM/freeze maintenance, physical tuple metadata or
+xid wraparound management; it fails when the 32-bit allocation space is exhausted.
+
+Evidence: `tests/contract/regressions/issue-queue.test.ts` (PostgreSQL 18.3 differential contracts),
+`tests/contract/api/schema-report.test.ts`, and `tests/wire/issue-queue.test.ts`.

package/README.md

@@ -529,6 +529,11 @@ search, `ON CONFLICT DO NOTHING/UPDATE`, `RETURNING`, `PREPARE`/`EXECUTE`/`DEALL
 
 ## API
 
+`analyzeSchema(sql, options?)` returns bounded, JSON-serializable schema compatibility diagnostics
+without changing an application database. Accepted statements remain semantically unverified;
+see [schema preflight and row-version metadata](COMPATIBILITY.md#schema-preflight-and-row-version-metadata)
+for limits, the report contract and `xmin` support.
+
 Stable runtime exports of the main entry:
 
 | Export | Description |

package/compat/sections/trg.ts

@@ -4,6 +4,8 @@ const D = "documented_divergence" as const;
 const REPORT = ["tests/contract/_reports/session-system.md"];
 
 export const TRG_SECTION: CatalogSection = section("TRG", "Triggers", true, [
+  ["deferred-01", "deferred constraint trigger observes final transaction state"],
+  ["deferred-02", "SET CONSTRAINTS IMMEDIATE flushes pending events"],
   ["bi-01", "BEFORE INSERT trigger assigns a NEW column"],
   ["bi-02", "BEFORE INSERT trigger computes from another NEW column"],
   ["bi-03", "BEFORE INSERT returning NULL suppresses the row"],