datahike.pg.catalog.objects — org.replikativ/pg-datahike 0.1.196

Persistent PostgreSQL catalog identities for user-defined objects.

PostgreSQL identifies an object by (catalog relation OID, object OID), with an optional sub-object number for columns. Keeping that shape here lets routines, triggers, dependencies, and future catalog work share one transactional identity model instead of inventing per-feature IDs.

address clj

(address class-oid object-oid)

(address class-oid object-oid sub-id)

A PostgreSQL object address. sub-id is zero for the object itself and positive for a sub-object such as a table column.

address-key clj

(address-key class-oid object-oid)

Stable structural serialization of a top-level catalog address.

bump-revision-tx clj

(bump-revision-tx db class-oid object-oid)

catalog-entity clj

(catalog-entity db)

catalog-key clj

catalog-version clj

column-address-key clj

(column-address-key relation-oid attnum)

column-by-attnum clj

(column-by-attnum db relation-oid attnum)

column-by-name clj

(column-by-name db relation-oid name)

column-name-key clj

(column-name-key relation-oid name)

columns-by-relation clj

(columns-by-relation db relation-oid)

(columns-by-relation db relation-oid live-only?)

All persisted column subobjects for a relation in attnum order. Dropped tombstones are included unless live-only? is true.

create-columns-tx clj

(create-columns-tx relation-oid columns)

Create durable column subobject rows. The relation may be created in the same transaction, so its lookup ref need not resolve in db yet.

create-object-tx clj

(create-object-tx db
                  {:keys [class-oid oid kind name namespace-oid owner-oid
                          revision legacy-oid? identity-tail]
                   :or {owner-oid 10 revision 1}})

Build an object-row transaction. Namespace may be nil for cluster-global objects; otherwise it must already exist. The two :db.unique/value keys are the atomic stale-reader guard: Datahike rejects duplicates and cannot upsert/merge them as it would :db.unique/identity.

creation-namespace clj

(creation-namespace db entries user-name)

The first existing explicit search_path entry. The implicit pg_catalog entry used for lookup is deliberately not a creation target.

drop-object-tx clj

(drop-object-tx db class-oid object-oid)

drop-relation-columns-tx clj

(drop-relation-columns-tx db relation-oid)

first-user-oid clj

identity-key clj

(identity-key class-oid namespace-oid name)

(identity-key class-oid namespace-oid name tail)

Stable structural serialization of a namespace-scoped object identity. tail carries kind-specific identity, notably a routine's argument types.

initialization-tx clj

(initialization-tx legacy-objects next-oid)

(initialization-tx legacy-objects legacy-columns next-oid)

initialize-catalog clj

(initialize-catalog txdb legacy-objects legacy-columns next-oid)

Transaction function for an atomic, idempotent initial migration. legacy-objects must already have deterministic, collision-free OIDs.

max-oid clj

migrate-v1-to-v2 clj

(migrate-v1-to-v2 txdb legacy-columns)

Atomic catalog-v1 upgrade that installs durable column subaddresses.

migrate-v2-to-v3 clj

(migrate-v2-to-v3 txdb tx-data expected-next next-oid)

Atomic address-completeness migration. tx-data contains preplanned generic object and column rows; allocator/version CAS operations make a stale or concurrent plan fail instead of partially merging.

namespace-by-name clj

(namespace-by-name db name)

namespace-by-oid clj

(namespace-by-oid db oid)

next-attnum clj

(next-attnum db relation-oid)

object-by-address clj

(object-by-address db class-oid object-oid)

object-by-identity clj

(object-by-identity db class-oid namespace-oid name)

(object-by-identity db class-oid namespace-oid name tail)

objects-by-kind clj

(objects-by-kind db kind)

objects-in-namespace clj

(objects-in-namespace db namespace-oid)

pg-catalog-namespace-oid clj

pg-class-oid clj

pg-namespace-oid clj

pg-proc-oid clj

pg-trigger-oid clj

pg-type-oid clj

public-namespace-oid clj

rename-column-tx clj

(rename-column-tx db relation-oid attnum new-name new-storage-ident)

Rename a live column without changing its relation-local address or physical Datahike storage ident.

rename-object-tx clj

(rename-object-tx db class-oid object-oid new-name identity-tail)

reserve-user-oid-tx clj

(reserve-user-oid-tx db)

Reserve one currently-unused user OID with a CAS operation that the caller must transact together with the object creation. Allocation is therefore rolled back with its DDL and naturally chains against a speculative DB.

reserve-user-oids-tx clj

(reserve-user-oids-tx db n)

Reserve n noncolliding OIDs with one allocator CAS. Used when PostgreSQL creates coupled catalog objects, such as a relation and its row type.

resolve-search-path clj

(resolve-search-path db entries user-name)

Resolve PostgreSQL search_path entries to existing namespace OIDs. pg_catalog is implicitly searched first unless explicitly positioned. $user contributes a schema only when it exists.

schema clj

Datahike schema for the shared user-object catalog. Address and identity keys use strict unique/value semantics: an attempted duplicate must fail, never upsert or merge two independently-created objects.

tombstone-column-tx clj

(tombstone-column-tx db relation-oid attnum)

Mark a column address dropped while retaining its attnum. The live name key is released so a later column with the same SQL name receives a fresh address. Revision guarding belongs to the surrounding relation mutation.

v1-to-v2-tx clj

(v1-to-v2-tx legacy-columns)

v2-to-v3-tx clj

(v2-to-v3-tx tx-data expected-next next-oid)

valid-user-oid? clj

(valid-user-oid? oid)

True for an allocatable PostgreSQL user OID. OID is uint32, while values below FirstNormalObjectId belong to PostgreSQL's bootstrap catalogs.