Em qua., 19 de jun. de 2024 às 10:28, Ranier Vilela <ranier...@gmail.com> escreveu:
> Em qua., 19 de jun. de 2024 às 10:26, Joel Jacobson <j...@compiler.org> > escreveu: > >> Hi Ranier, >> >> Thanks for looking at this. >> >> I've double-checked the patch I sent, and it works fine. >> >> I think I know the cause of your problem: >> >> Since this is a catalog change, you need to run `make clean`, to ensure >> the catalog is rebuilt, >> followed by the usual `make && make install`. >> >> You also need to run `initdb` to create a new database cluster, with the >> new catalog version. >> >> Let me know if you need more specific instructions. >> > Sorry, sorry but I'm on Windows -> meson. > > Double checked with: > ninja clean > ninja > ninja install > Sorry for the noise, now pg_get_acl is shown in the regress test. Regarding the patch, could it be written in the following style? Datum pg_get_acl(PG_FUNCTION_ARGS) { Oid classId = PG_GETARG_OID(0); Oid objectId = PG_GETARG_OID(1); Oid catalogId; AttrNumber Anum_oid; AttrNumber Anum_acl; /* for "pinned" items in pg_depend, return null */ if (!OidIsValid(classId) && !OidIsValid(objectId)) PG_RETURN_NULL(); catalogId = (classId == LargeObjectRelationId) ? LargeObjectMetadataRelationId : classId; Anum_oid = get_object_attnum_oid(catalogId); Anum_acl = get_object_attnum_acl(catalogId); if (Anum_acl != InvalidAttrNumber) { Relation rel; HeapTuple tup; Datum datum; bool isnull; rel = table_open(catalogId, AccessShareLock); tup = get_catalog_object_by_oid(rel, Anum_oid, objectId); if (!HeapTupleIsValid(tup)) elog(ERROR, "cache lookup failed for object %u of catalog \"%s\"", objectId, RelationGetRelationName(rel)); datum = heap_getattr(tup, Anum_acl, RelationGetDescr(rel), &isnull); table_close(rel, AccessShareLock); if (!isnull) PG_RETURN_DATUM(datum); } PG_RETURN_NULL(); } best regards, Ranier Vilela