On Mon, Jul 21, 2025 at 09:40:02AM -0400, Robert Haas wrote:
> On Fri, Jul 18, 2025 at 3:17 PM Noah Misch <n...@leadboat.com> wrote:
> > + * Sort by encoding, per pg_collation_name_enc_nsp_index.
> > Wherever
> > + * this changes dump order, restoring the dump fails
> > anyway. CREATE
> > + * COLLATION can't create a tie for this to break, because
> > it imposes
> > + * restrictions to make (nspname, collname) uniquely
> > identify a
> > + * collation within a given DatabaseEncoding. While
> > + * pg_import_system_collations() can create a tie,
> > pg_dump+restore
> > + * fails after pg_import_system_collations('my_schema')
> > does so.
> > + * There's little to gain by ignoring one natural key
> > column on the
> > + * basis of those limitations elsewhere, so respect the
> > full natural
> > + * key like we do for other object types.
>
> This is also good. I suggest s/Wherever/Technically, this is not
> necessary, because wherever/ and s/There's/However, there's/.
I used that. I started to prepare the back-branch versions, but that revealed
three problems affecting the master patch:
(1) Sorting constraints segfaulted if either of a pair of equal-name
constraints was a domain constraint. Fortunately, commit da71717 added a test
case for that between when I mailed patch v1 and when I went to commit. One
can reproduce it by dumping a database containing:
CREATE DOMAIN d1 AS int CONSTRAINT dc CHECK (value > 0);
CREATE DOMAIN d2 AS int CONSTRAINT dc CHECK (value > 0);
I made pg_dump sort domain constraints of a given name before table
constraints of that name, for consistency with our decision to sort CREATE
DOMAIN before CREATE TABLE. The main alternative was to sort by parent object
name irrespective of parent object type, i.e. DOMAIN a < TABLE b < DOMAIN c.
That alternative lacked a relevant precedent. I've now audited the natural
keys of catalogs for which I'm changing sort order, and I think that was the
only one I missed.
(2) Sorting opclasses failed a new assertion when dumping a v9.2 source (and
likely 9.[345]), because getAccessMethods() doesn't read pg_am when dumping
from a version predating CREATE ACCESS METHOD. findAccessMethodByOid() found
no access methods, since pg_dump had read none. I've changed
getAccessMethods() to always read pg_am. (For pre-v9.6 sources, I've kept the
function's behavior of never marking an access method for dumping.) pg_am is
small enough for this read to incur negligible cost. The main alternative
was, for pre-v9.6, sorting access methods by pg_am.oid. That would have been
less code, but dump order would have differed between pre-v9.6 and v9.6+.
(3) pgTypeNameCompare() implied postfix operators don't exist, but master
pg_dump will support reading from pre-v14 clusters for several more years.
The code behaved fine, but I've updated the comment.
I regret missing those in v1. I've attached v2, including branch-specific
patches. I'll first need to back-patch 350e6b8, which fixed sorting of CREATE
RULE, to v17 and earlier. Since 350e6b8 is conflict-free all the way back to
v13, I'm not attaching it.
Thanks,
nm
From: Noah Misch <n...@leadboat.com>
Sort dump objects independent of OIDs, for the 7 holdout object types.
pg_dump sorts objects by their logical names, e.g. (nspname, relname,
tgname), before dependency-driven reordering. That removes one source
of logically-identical databases differing in their schema-only dumps.
In other words, it helps with schema diffing. The logical name sort
ignored essential sort keys for constraints, operators, PUBLICATION
... FOR TABLE, PUBLICATION ... FOR TABLES IN SCHEMA, operator classes,
and operator families. pg_dump's sort then depended on object OID,
yielding spurious schema diffs. After this change, OIDs affect dump
order only in the event of catalog corruption. While pg_dump also
wrongly ignored pg_collation.collencoding, CREATE COLLATION restrictions
have been keeping that imperceptible in practical use.
Use techniques like we use for object types already having full sort key
coverage. Where the pertinent queries weren't fetching the ignored sort
keys, this adds columns to those queries and stores those keys in memory
for the long term.
The ignorance of sort keys became more problematic when commit
172259afb563d35001410dc6daad78b250924038 added a schema diff test
sensitive to it. However, dump order stability isn't a new goal, and
this might avoid other dump comparison failures. Hence, back-patch to
v13 (all supported versions).
Reviewed-by: Robert Haas <robertmh...@gmail.com>
Discussion: https://postgr.es/m/20250707192654.9e.nmi...@google.com
Backpatch-through: 13
diff --git a/src/bin/pg_dump/common.c b/src/bin/pg_dump/common.c
index aa1589e..a1976fa 100644
--- a/src/bin/pg_dump/common.c
+++ b/src/bin/pg_dump/common.c
@@ -17,6 +17,7 @@
#include <ctype.h>
+#include "catalog/pg_am_d.h"
#include "catalog/pg_class_d.h"
#include "catalog/pg_collation_d.h"
#include "catalog/pg_extension_d.h"
@@ -945,6 +946,24 @@ findOprByOid(Oid oid)
}
/*
+ * findAccessMethodByOid
+ * finds the DumpableObject for the access method with the given oid
+ * returns NULL if not found
+ */
+AccessMethodInfo *
+findAccessMethodByOid(Oid oid)
+{
+ CatalogId catId;
+ DumpableObject *dobj;
+
+ catId.tableoid = AccessMethodRelationId;
+ catId.oid = oid;
+ dobj = findObjectByCatalogId(catId);
+ Assert(dobj == NULL || dobj->objType == DO_ACCESS_METHOD);
+ return (AccessMethodInfo *) dobj;
+}
+
+/*
* findCollationByOid
* finds the DumpableObject for the collation with the given oid
* returns NULL if not found
diff --git a/src/bin/pg_dump/pg_dump.c b/src/bin/pg_dump/pg_dump.c
index ede10e5..1f93e7a 100644
--- a/src/bin/pg_dump/pg_dump.c
+++ b/src/bin/pg_dump/pg_dump.c
@@ -2207,6 +2207,13 @@ selectDumpableProcLang(ProcLangInfo *plang, Archive
*fout)
static void
selectDumpableAccessMethod(AccessMethodInfo *method, Archive *fout)
{
+ /* see getAccessMethods() comment about v9.6. */
+ if (fout->remoteVersion < 90600)
+ {
+ method->dobj.dump = DUMP_COMPONENT_NONE;
+ return;
+ }
+
if (checkExtensionMembership(&method->dobj, fout))
return; /* extension membership
overrides all else */
@@ -6248,6 +6255,8 @@ getOperators(Archive *fout)
int i_oprnamespace;
int i_oprowner;
int i_oprkind;
+ int i_oprleft;
+ int i_oprright;
int i_oprcode;
/*
@@ -6259,6 +6268,8 @@ getOperators(Archive *fout)
"oprnamespace, "
"oprowner, "
"oprkind, "
+ "oprleft, "
+ "oprright, "
"oprcode::oid AS oprcode "
"FROM pg_operator");
@@ -6274,6 +6285,8 @@ getOperators(Archive *fout)
i_oprnamespace = PQfnumber(res, "oprnamespace");
i_oprowner = PQfnumber(res, "oprowner");
i_oprkind = PQfnumber(res, "oprkind");
+ i_oprleft = PQfnumber(res, "oprleft");
+ i_oprright = PQfnumber(res, "oprright");
i_oprcode = PQfnumber(res, "oprcode");
for (i = 0; i < ntups; i++)
@@ -6287,6 +6300,8 @@ getOperators(Archive *fout)
findNamespace(atooid(PQgetvalue(res, i,
i_oprnamespace)));
oprinfo[i].rolname = getRoleName(PQgetvalue(res, i,
i_oprowner));
oprinfo[i].oprkind = (PQgetvalue(res, i, i_oprkind))[0];
+ oprinfo[i].oprleft = atooid(PQgetvalue(res, i, i_oprleft));
+ oprinfo[i].oprright = atooid(PQgetvalue(res, i, i_oprright));
oprinfo[i].oprcode = atooid(PQgetvalue(res, i, i_oprcode));
/* Decide whether we want to dump it */
@@ -6315,6 +6330,7 @@ getCollations(Archive *fout)
int i_collname;
int i_collnamespace;
int i_collowner;
+ int i_collencoding;
query = createPQExpBuffer();
@@ -6325,7 +6341,8 @@ getCollations(Archive *fout)
appendPQExpBufferStr(query, "SELECT tableoid, oid, collname, "
"collnamespace, "
- "collowner "
+ "collowner, "
+ "collencoding "
"FROM pg_collation");
res = ExecuteSqlQuery(fout, query->data, PGRES_TUPLES_OK);
@@ -6339,6 +6356,7 @@ getCollations(Archive *fout)
i_collname = PQfnumber(res, "collname");
i_collnamespace = PQfnumber(res, "collnamespace");
i_collowner = PQfnumber(res, "collowner");
+ i_collencoding = PQfnumber(res, "collencoding");
for (i = 0; i < ntups; i++)
{
@@ -6350,6 +6368,7 @@ getCollations(Archive *fout)
collinfo[i].dobj.namespace =
findNamespace(atooid(PQgetvalue(res, i,
i_collnamespace)));
collinfo[i].rolname = getRoleName(PQgetvalue(res, i,
i_collowner));
+ collinfo[i].collencoding = atoi(PQgetvalue(res, i,
i_collencoding));
/* Decide whether we want to dump it */
selectDumpableObject(&(collinfo[i].dobj), fout);
@@ -6440,16 +6459,28 @@ getAccessMethods(Archive *fout)
int i_amhandler;
int i_amtype;
- /* Before 9.6, there are no user-defined access methods */
- if (fout->remoteVersion < 90600)
- return;
-
query = createPQExpBuffer();
- /* Select all access methods from pg_am table */
- appendPQExpBufferStr(query, "SELECT tableoid, oid, amname, amtype, "
- "amhandler::pg_catalog.regproc
AS amhandler "
- "FROM pg_am");
+ /*
+ * Select all access methods from pg_am table. v9.6 introduced CREATE
+ * ACCESS METHOD, so earlier versions usually have only built-in access
+ * methods. v9.6 also changed the access method API, replacing dozens
of
+ * pg_am columns with amhandler. Even if a user created an access
method
+ * by "INSERT INTO pg_am", we have no way to translate pre-v9.6 pg_am
+ * columns to a v9.6+ CREATE ACCESS METHOD. Hence, before v9.6, read
+ * pg_am just to facilitate findAccessMethodByOid() providing the
+ * OID-to-name mapping.
+ */
+ appendPQExpBufferStr(query, "SELECT tableoid, oid, amname, ");
+ if (fout->remoteVersion >= 90600)
+ appendPQExpBufferStr(query,
+ "amtype, "
+
"amhandler::pg_catalog.regproc AS amhandler ");
+ else
+ appendPQExpBufferStr(query,
+
"'i'::pg_catalog.\"char\" AS amtype, "
+
"'-'::pg_catalog.regproc AS amhandler ");
+ appendPQExpBufferStr(query, "FROM pg_am");
res = ExecuteSqlQuery(fout, query->data, PGRES_TUPLES_OK);
@@ -6498,6 +6529,7 @@ getOpclasses(Archive *fout)
OpclassInfo *opcinfo;
int i_tableoid;
int i_oid;
+ int i_opcmethod;
int i_opcname;
int i_opcnamespace;
int i_opcowner;
@@ -6507,7 +6539,7 @@ getOpclasses(Archive *fout)
* system-defined opclasses at dump-out time.
*/
- appendPQExpBufferStr(query, "SELECT tableoid, oid, opcname, "
+ appendPQExpBufferStr(query, "SELECT tableoid, oid, opcmethod, opcname, "
"opcnamespace, "
"opcowner "
"FROM pg_opclass");
@@ -6520,6 +6552,7 @@ getOpclasses(Archive *fout)
i_tableoid = PQfnumber(res, "tableoid");
i_oid = PQfnumber(res, "oid");
+ i_opcmethod = PQfnumber(res, "opcmethod");
i_opcname = PQfnumber(res, "opcname");
i_opcnamespace = PQfnumber(res, "opcnamespace");
i_opcowner = PQfnumber(res, "opcowner");
@@ -6533,6 +6566,7 @@ getOpclasses(Archive *fout)
opcinfo[i].dobj.name = pg_strdup(PQgetvalue(res, i, i_opcname));
opcinfo[i].dobj.namespace =
findNamespace(atooid(PQgetvalue(res, i,
i_opcnamespace)));
+ opcinfo[i].opcmethod = atooid(PQgetvalue(res, i, i_opcmethod));
opcinfo[i].rolname = getRoleName(PQgetvalue(res, i,
i_opcowner));
/* Decide whether we want to dump it */
@@ -6558,6 +6592,7 @@ getOpfamilies(Archive *fout)
OpfamilyInfo *opfinfo;
int i_tableoid;
int i_oid;
+ int i_opfmethod;
int i_opfname;
int i_opfnamespace;
int i_opfowner;
@@ -6569,7 +6604,7 @@ getOpfamilies(Archive *fout)
* system-defined opfamilies at dump-out time.
*/
- appendPQExpBufferStr(query, "SELECT tableoid, oid, opfname, "
+ appendPQExpBufferStr(query, "SELECT tableoid, oid, opfmethod, opfname, "
"opfnamespace, "
"opfowner "
"FROM pg_opfamily");
@@ -6583,6 +6618,7 @@ getOpfamilies(Archive *fout)
i_tableoid = PQfnumber(res, "tableoid");
i_oid = PQfnumber(res, "oid");
i_opfname = PQfnumber(res, "opfname");
+ i_opfmethod = PQfnumber(res, "opfmethod");
i_opfnamespace = PQfnumber(res, "opfnamespace");
i_opfowner = PQfnumber(res, "opfowner");
@@ -6595,6 +6631,7 @@ getOpfamilies(Archive *fout)
opfinfo[i].dobj.name = pg_strdup(PQgetvalue(res, i, i_opfname));
opfinfo[i].dobj.namespace =
findNamespace(atooid(PQgetvalue(res, i,
i_opfnamespace)));
+ opfinfo[i].opfmethod = atooid(PQgetvalue(res, i, i_opfmethod));
opfinfo[i].rolname = getRoleName(PQgetvalue(res, i,
i_opfowner));
/* Decide whether we want to dump it */
diff --git a/src/bin/pg_dump/pg_dump.h b/src/bin/pg_dump/pg_dump.h
index 2370c98..30121af 100644
--- a/src/bin/pg_dump/pg_dump.h
+++ b/src/bin/pg_dump/pg_dump.h
@@ -260,6 +260,8 @@ typedef struct _oprInfo
DumpableObject dobj;
const char *rolname;
char oprkind;
+ Oid oprleft;
+ Oid oprright;
Oid oprcode;
} OprInfo;
@@ -273,12 +275,14 @@ typedef struct _accessMethodInfo
typedef struct _opclassInfo
{
DumpableObject dobj;
+ Oid opcmethod;
const char *rolname;
} OpclassInfo;
typedef struct _opfamilyInfo
{
DumpableObject dobj;
+ Oid opfmethod;
const char *rolname;
} OpfamilyInfo;
@@ -286,6 +290,7 @@ typedef struct _collInfo
{
DumpableObject dobj;
const char *rolname;
+ int collencoding;
} CollInfo;
typedef struct _convInfo
@@ -759,6 +764,7 @@ extern TableInfo *findTableByOid(Oid oid);
extern TypeInfo *findTypeByOid(Oid oid);
extern FuncInfo *findFuncByOid(Oid oid);
extern OprInfo *findOprByOid(Oid oid);
+extern AccessMethodInfo *findAccessMethodByOid(Oid oid);
extern CollInfo *findCollationByOid(Oid oid);
extern NamespaceInfo *findNamespaceByOid(Oid oid);
extern ExtensionInfo *findExtensionByOid(Oid oid);
diff --git a/src/bin/pg_dump/pg_dump_sort.c b/src/bin/pg_dump/pg_dump_sort.c
index f99a079..a02da3e 100644
--- a/src/bin/pg_dump/pg_dump_sort.c
+++ b/src/bin/pg_dump/pg_dump_sort.c
@@ -162,6 +162,8 @@ static DumpId postDataBoundId;
static int DOTypeNameCompare(const void *p1, const void *p2);
+static int pgTypeNameCompare(Oid typid1, Oid typid2);
+static int accessMethodNameCompare(Oid am1, Oid am2);
static bool TopoSort(DumpableObject **objs,
int numObjs,
DumpableObject **ordering,
@@ -228,12 +230,39 @@ DOTypeNameCompare(const void *p1, const void *p2)
else if (obj2->namespace)
return 1;
- /* Sort by name */
+ /*
+ * Sort by name. With a few exceptions, names here are single catalog
+ * columns. To get a fuller picture, grep pg_dump.c for "dobj.name = ".
+ * Names here don't match "Name:" in plain format output, which is a
+ * _tocEntry.tag. For example, DumpableObject.name of a constraint is
+ * pg_constraint.conname, but _tocEntry.tag of a constraint is relname
and
+ * conname joined with a space.
+ */
cmpval = strcmp(obj1->name, obj2->name);
if (cmpval != 0)
return cmpval;
- /* To have a stable sort order, break ties for some object types */
+ /*
+ * Sort by type. This helps types that share a type priority without
+ * sharing a unique name constraint, e.g. opclass and opfamily.
+ */
+ cmpval = obj1->objType - obj2->objType;
+ if (cmpval != 0)
+ return cmpval;
+
+ /*
+ * To have a stable sort order, break ties for some object types. Most
+ * catalogs have a natural key, e.g. pg_proc_proname_args_nsp_index.
Where
+ * the above "namespace" and "name" comparisons don't cover all natural
+ * key columns, compare the rest here.
+ *
+ * The natural key usually refers to other catalogs by surrogate keys.
+ * Hence, this translates each of those references to the natural key of
+ * the referenced catalog. That may descend through multiple levels of
+ * catalog references. For example, to sort by pg_proc.proargtypes,
+ * descend to each pg_type and then further to its pg_namespace, for an
+ * overall sort by (nspname, typname).
+ */
if (obj1->objType == DO_FUNC || obj1->objType == DO_AGG)
{
FuncInfo *fobj1 = *(FuncInfo *const *) p1;
@@ -246,22 +275,10 @@ DOTypeNameCompare(const void *p1, const void *p2)
return cmpval;
for (i = 0; i < fobj1->nargs; i++)
{
- TypeInfo *argtype1 =
findTypeByOid(fobj1->argtypes[i]);
- TypeInfo *argtype2 =
findTypeByOid(fobj2->argtypes[i]);
-
- if (argtype1 && argtype2)
- {
- if (argtype1->dobj.namespace &&
argtype2->dobj.namespace)
- {
- cmpval =
strcmp(argtype1->dobj.namespace->dobj.name,
-
argtype2->dobj.namespace->dobj.name);
- if (cmpval != 0)
- return cmpval;
- }
- cmpval = strcmp(argtype1->dobj.name,
argtype2->dobj.name);
- if (cmpval != 0)
- return cmpval;
- }
+ cmpval = pgTypeNameCompare(fobj1->argtypes[i],
+
fobj2->argtypes[i]);
+ if (cmpval != 0)
+ return cmpval;
}
}
else if (obj1->objType == DO_OPERATOR)
@@ -273,6 +290,57 @@ DOTypeNameCompare(const void *p1, const void *p2)
cmpval = (oobj2->oprkind - oobj1->oprkind);
if (cmpval != 0)
return cmpval;
+ /* Within an oprkind, sort by argument type names */
+ cmpval = pgTypeNameCompare(oobj1->oprleft, oobj2->oprleft);
+ if (cmpval != 0)
+ return cmpval;
+ cmpval = pgTypeNameCompare(oobj1->oprright, oobj2->oprright);
+ if (cmpval != 0)
+ return cmpval;
+ }
+ else if (obj1->objType == DO_OPCLASS)
+ {
+ OpclassInfo *opcobj1 = *(OpclassInfo *const *) p1;
+ OpclassInfo *opcobj2 = *(OpclassInfo *const *) p2;
+
+ /* Sort by access method name, per pg_opclass_am_name_nsp_index
*/
+ cmpval = accessMethodNameCompare(opcobj1->opcmethod,
+
opcobj2->opcmethod);
+ if (cmpval != 0)
+ return cmpval;
+ }
+ else if (obj1->objType == DO_OPFAMILY)
+ {
+ OpfamilyInfo *opfobj1 = *(OpfamilyInfo *const *) p1;
+ OpfamilyInfo *opfobj2 = *(OpfamilyInfo *const *) p2;
+
+ /* Sort by access method name, per
pg_opfamily_am_name_nsp_index */
+ cmpval = accessMethodNameCompare(opfobj1->opfmethod,
+
opfobj2->opfmethod);
+ if (cmpval != 0)
+ return cmpval;
+ }
+ else if (obj1->objType == DO_COLLATION)
+ {
+ CollInfo *cobj1 = *(CollInfo *const *) p1;
+ CollInfo *cobj2 = *(CollInfo *const *) p2;
+
+ /*
+ * Sort by encoding, per pg_collation_name_enc_nsp_index.
Technically,
+ * this is not necessary, because wherever this changes dump
order,
+ * restoring the dump fails anyway. CREATE COLLATION can't
create a
+ * tie for this to break, because it imposes restrictions to
make
+ * (nspname, collname) uniquely identify a collation within a
given
+ * DatabaseEncoding. While pg_import_system_collations() can
create a
+ * tie, pg_dump+restore fails after
+ * pg_import_system_collations('my_schema') does so. However,
there's
+ * little to gain by ignoring one natural key column on the
basis of
+ * those limitations elsewhere, so respect the full natural key
like
+ * we do for other object types.
+ */
+ cmpval = cobj1->collencoding - cobj2->collencoding;
+ if (cmpval != 0)
+ return cmpval;
}
else if (obj1->objType == DO_ATTRDEF)
{
@@ -317,11 +385,143 @@ DOTypeNameCompare(const void *p1, const void *p2)
if (cmpval != 0)
return cmpval;
}
+ else if (obj1->objType == DO_CONSTRAINT)
+ {
+ ConstraintInfo *robj1 = *(ConstraintInfo *const *) p1;
+ ConstraintInfo *robj2 = *(ConstraintInfo *const *) p2;
- /* Usually shouldn't get here, but if we do, sort by OID */
+ /*
+ * Sort domain constraints before table constraints, for
consistency
+ * with our decision to sort CREATE DOMAIN before CREATE TABLE.
+ */
+ if (robj1->condomain)
+ {
+ if (robj2->condomain)
+ {
+ /* Sort by domain name (domain namespace was
considered) */
+ cmpval = strcmp(robj1->condomain->dobj.name,
+
robj2->condomain->dobj.name);
+ if (cmpval != 0)
+ return cmpval;
+ }
+ else
+ return PRIO_TYPE - PRIO_TABLE;
+ }
+ else if (robj2->condomain)
+ return PRIO_TABLE - PRIO_TYPE;
+ else
+ {
+ /* Sort by table name (table namespace was considered
already) */
+ cmpval = strcmp(robj1->contable->dobj.name,
+
robj2->contable->dobj.name);
+ if (cmpval != 0)
+ return cmpval;
+ }
+ }
+ else if (obj1->objType == DO_PUBLICATION_REL)
+ {
+ PublicationRelInfo *probj1 = *(PublicationRelInfo *const *) p1;
+ PublicationRelInfo *probj2 = *(PublicationRelInfo *const *) p2;
+
+ /* Sort by publication name, since (namespace, name) match the
rel */
+ cmpval = strcmp(probj1->publication->dobj.name,
+ probj2->publication->dobj.name);
+ if (cmpval != 0)
+ return cmpval;
+ }
+ else if (obj1->objType == DO_PUBLICATION_TABLE_IN_SCHEMA)
+ {
+ PublicationSchemaInfo *psobj1 = *(PublicationSchemaInfo *const
*) p1;
+ PublicationSchemaInfo *psobj2 = *(PublicationSchemaInfo *const
*) p2;
+
+ /* Sort by publication name, since ->name is just nspname */
+ cmpval = strcmp(psobj1->publication->dobj.name,
+ psobj2->publication->dobj.name);
+ if (cmpval != 0)
+ return cmpval;
+ }
+
+ /*
+ * Shouldn't get here except after catalog corruption, but if we do,
sort
+ * by OID. This may make logically-identical databases differ in the
+ * order of objects in dump output. Users will get spurious schema
diffs.
+ * Expect flaky failures of 002_pg_upgrade.pl test 'dump outputs from
+ * original and restored regression databases match' if the regression
+ * database contains objects allowing that test to reach here. That's a
+ * consequence of the test using "pg_restore -j", which doesn't fully
+ * constrain OID assignment order.
+ */
+ Assert(false);
return oidcmp(obj1->catId.oid, obj2->catId.oid);
}
+/* Compare two OID-identified pg_type values by nspname, then by typname. */
+static int
+pgTypeNameCompare(Oid typid1, Oid typid2)
+{
+ TypeInfo *typobj1;
+ TypeInfo *typobj2;
+ int cmpval;
+
+ if (typid1 == typid2)
+ return 0;
+
+ typobj1 = findTypeByOid(typid1);
+ typobj2 = findTypeByOid(typid2);
+
+ if (!typobj1 || !typobj2)
+ {
+ /*
+ * getTypes() didn't find some OID. Assume catalog corruption,
e.g.
+ * an oprright value without the corresponding OID in a pg_type
row.
+ * Report as "equal", so the caller uses the next available
basis for
+ * comparison, e.g. the next function argument.
+ *
+ * Unary operators have InvalidOid in oprleft (if oprkind='r')
or in
+ * oprright (if oprkind='l'). Caller already sorted by oprkind,
+ * calling us only for like-kind operators. Hence, "typid1 ==
typid2"
+ * took care of InvalidOid. (v14 removed postfix operator
support.
+ * Hence, when dumping from v14+, only oprleft can be
InvalidOid.)
+ */
+ Assert(false);
+ return 0;
+ }
+
+ if (!typobj1->dobj.namespace || !typobj2->dobj.namespace)
+ Assert(false); /* catalog corruption */
+ else
+ {
+ cmpval = strcmp(typobj1->dobj.namespace->dobj.name,
+
typobj2->dobj.namespace->dobj.name);
+ if (cmpval != 0)
+ return cmpval;
+ }
+ return strcmp(typobj1->dobj.name, typobj2->dobj.name);
+}
+
+/* Compare two OID-identified pg_am values by amname. */
+static int
+accessMethodNameCompare(Oid am1, Oid am2)
+{
+ AccessMethodInfo *amobj1;
+ AccessMethodInfo *amobj2;
+
+ if (am1 == am2)
+ return 0;
+
+ amobj1 = findAccessMethodByOid(am1);
+ amobj2 = findAccessMethodByOid(am2);
+
+ if (!amobj1 || !amobj2)
+ {
+ /* catalog corruption: handle like pgTypeNameCompare() does */
+ Assert(false);
+ return 0;
+ }
+
+ return strcmp(amobj1->dobj.name, amobj2->dobj.name);
+}
+
/*
* Sort the given objects into a safe dump order using dependency
diff --git a/src/test/regress/expected/publication.out
b/src/test/regress/expected/publication.out
index 3a2eacd..1ec3fa3 100644
--- a/src/test/regress/expected/publication.out
+++ b/src/test/regress/expected/publication.out
@@ -1934,3 +1934,24 @@ RESET client_min_messages;
RESET SESSION AUTHORIZATION;
DROP ROLE regress_publication_user, regress_publication_user2;
DROP ROLE regress_publication_user_dummy;
+-- stage objects for pg_dump tests
+CREATE SCHEMA pubme CREATE TABLE t0 (c int, d int) CREATE TABLE t1 (c int);
+CREATE SCHEMA pubme2 CREATE TABLE t0 (c int, d int);
+SET client_min_messages = 'ERROR';
+CREATE PUBLICATION dump_pub_qual_1ct FOR
+ TABLE ONLY pubme.t0 (c, d) WHERE (c > 0);
+CREATE PUBLICATION dump_pub_qual_2ct FOR
+ TABLE ONLY pubme.t0 (c) WHERE (c > 0),
+ TABLE ONLY pubme.t1 (c);
+CREATE PUBLICATION dump_pub_nsp_1ct FOR
+ TABLES IN SCHEMA pubme;
+CREATE PUBLICATION dump_pub_nsp_2ct FOR
+ TABLES IN SCHEMA pubme,
+ TABLES IN SCHEMA pubme2;
+CREATE PUBLICATION dump_pub_all FOR
+ TABLE ONLY pubme.t0,
+ TABLE ONLY pubme.t1 WHERE (c < 0),
+ TABLES IN SCHEMA pubme,
+ TABLES IN SCHEMA pubme2
+ WITH (publish_via_partition_root = true);
+RESET client_min_messages;
diff --git a/src/test/regress/sql/publication.sql
b/src/test/regress/sql/publication.sql
index c9e3091..2585f08 100644
--- a/src/test/regress/sql/publication.sql
+++ b/src/test/regress/sql/publication.sql
@@ -1229,3 +1229,25 @@ RESET client_min_messages;
RESET SESSION AUTHORIZATION;
DROP ROLE regress_publication_user, regress_publication_user2;
DROP ROLE regress_publication_user_dummy;
+
+-- stage objects for pg_dump tests
+CREATE SCHEMA pubme CREATE TABLE t0 (c int, d int) CREATE TABLE t1 (c int);
+CREATE SCHEMA pubme2 CREATE TABLE t0 (c int, d int);
+SET client_min_messages = 'ERROR';
+CREATE PUBLICATION dump_pub_qual_1ct FOR
+ TABLE ONLY pubme.t0 (c, d) WHERE (c > 0);
+CREATE PUBLICATION dump_pub_qual_2ct FOR
+ TABLE ONLY pubme.t0 (c) WHERE (c > 0),
+ TABLE ONLY pubme.t1 (c);
+CREATE PUBLICATION dump_pub_nsp_1ct FOR
+ TABLES IN SCHEMA pubme;
+CREATE PUBLICATION dump_pub_nsp_2ct FOR
+ TABLES IN SCHEMA pubme,
+ TABLES IN SCHEMA pubme2;
+CREATE PUBLICATION dump_pub_all FOR
+ TABLE ONLY pubme.t0,
+ TABLE ONLY pubme.t1 WHERE (c < 0),
+ TABLES IN SCHEMA pubme,
+ TABLES IN SCHEMA pubme2
+ WITH (publish_via_partition_root = true);
+RESET client_min_messages;
commit 4d82220 (HEAD, zzy_test-commit-REL_17_STABLE)
Author: Noah Misch <n...@leadboat.com>
AuthorDate: Thu Jul 24 17:21:35 2025 -0700
Commit: Noah Misch <n...@leadboat.com>
CommitDate: Thu Jul 24 17:23:44 2025 -0700
Sort dump objects independent of OIDs, for the 7 holdout object types.
pg_dump sorts objects by their logical names, e.g. (nspname, relname,
tgname), before dependency-driven reordering. That removes one source
of logically-identical databases differing in their schema-only dumps.
In other words, it helps with schema diffing. The logical name sort
ignored essential sort keys for constraints, operators, PUBLICATION
... FOR TABLE, PUBLICATION ... FOR TABLES IN SCHEMA, operator classes,
and operator families. pg_dump's sort then depended on object OID,
yielding spurious schema diffs. After this change, OIDs affect dump
order only in the event of catalog corruption. While pg_dump also
wrongly ignored pg_collation.collencoding, CREATE COLLATION restrictions
have been keeping that imperceptible in practical use.
Use techniques like we use for object types already having full sort key
coverage. Where the pertinent queries weren't fetching the ignored sort
keys, this adds columns to those queries and stores those keys in memory
for the long term.
The ignorance of sort keys became more problematic when commit
172259afb563d35001410dc6daad78b250924038 added a schema diff test
sensitive to it. However, dump order stability isn't a new goal, and
this might avoid other dump comparison failures. Hence, back-patch to
v13 (all supported versions).
Reviewed-by: Robert Haas <robertmh...@gmail.com>
Discussion: https://postgr.es/m/20250707192654.9e.nmi...@google.com
Backpatch-through: 13
Conflicts:
src/bin/pg_dump/pg_dump.c
---
src/bin/pg_dump/common.c | 19 +++
src/bin/pg_dump/pg_dump.c | 62 ++++++--
src/bin/pg_dump/pg_dump.h | 6 +
src/bin/pg_dump/pg_dump_sort.c | 238 +++++++++++++++++++++++++++---
src/test/regress/expected/publication.out | 21 +++
src/test/regress/sql/publication.sql | 22 +++
6 files changed, 335 insertions(+), 33 deletions(-)
diff --git a/src/bin/pg_dump/common.c b/src/bin/pg_dump/common.c
index 64e7dc8..74bbea7 100644
--- a/src/bin/pg_dump/common.c
+++ b/src/bin/pg_dump/common.c
@@ -17,6 +17,7 @@
#include <ctype.h>
+#include "catalog/pg_am_d.h"
#include "catalog/pg_class_d.h"
#include "catalog/pg_collation_d.h"
#include "catalog/pg_extension_d.h"
@@ -934,6 +935,24 @@ findOprByOid(Oid oid)
}
/*
+ * findAccessMethodByOid
+ * finds the DumpableObject for the access method with the given oid
+ * returns NULL if not found
+ */
+AccessMethodInfo *
+findAccessMethodByOid(Oid oid)
+{
+ CatalogId catId;
+ DumpableObject *dobj;
+
+ catId.tableoid = AccessMethodRelationId;
+ catId.oid = oid;
+ dobj = findObjectByCatalogId(catId);
+ Assert(dobj == NULL || dobj->objType == DO_ACCESS_METHOD);
+ return (AccessMethodInfo *) dobj;
+}
+
+/*
* findCollationByOid
* finds the DumpableObject for the collation with the given oid
* returns NULL if not found
diff --git a/src/bin/pg_dump/pg_dump.c b/src/bin/pg_dump/pg_dump.c
index 2626dd2..0f26b01 100644
--- a/src/bin/pg_dump/pg_dump.c
+++ b/src/bin/pg_dump/pg_dump.c
@@ -2012,6 +2012,13 @@ selectDumpableProcLang(ProcLangInfo *plang, Archive
*fout)
static void
selectDumpableAccessMethod(AccessMethodInfo *method, Archive *fout)
{
+ /* see getAccessMethods() comment about v9.6. */
+ if (fout->remoteVersion < 90600)
+ {
+ method->dobj.dump = DUMP_COMPONENT_NONE;
+ return;
+ }
+
if (checkExtensionMembership(&method->dobj, fout))
return; /* extension membership
overrides all else */
@@ -5997,6 +6004,8 @@ getOperators(Archive *fout, int *numOprs)
int i_oprnamespace;
int i_oprowner;
int i_oprkind;
+ int i_oprleft;
+ int i_oprright;
int i_oprcode;
/*
@@ -6008,6 +6017,8 @@ getOperators(Archive *fout, int *numOprs)
"oprnamespace, "
"oprowner, "
"oprkind, "
+ "oprleft, "
+ "oprright, "
"oprcode::oid AS oprcode "
"FROM pg_operator");
@@ -6024,6 +6035,8 @@ getOperators(Archive *fout, int *numOprs)
i_oprnamespace = PQfnumber(res, "oprnamespace");
i_oprowner = PQfnumber(res, "oprowner");
i_oprkind = PQfnumber(res, "oprkind");
+ i_oprleft = PQfnumber(res, "oprleft");
+ i_oprright = PQfnumber(res, "oprright");
i_oprcode = PQfnumber(res, "oprcode");
for (i = 0; i < ntups; i++)
@@ -6037,6 +6050,8 @@ getOperators(Archive *fout, int *numOprs)
findNamespace(atooid(PQgetvalue(res, i,
i_oprnamespace)));
oprinfo[i].rolname = getRoleName(PQgetvalue(res, i,
i_oprowner));
oprinfo[i].oprkind = (PQgetvalue(res, i, i_oprkind))[0];
+ oprinfo[i].oprleft = atooid(PQgetvalue(res, i, i_oprleft));
+ oprinfo[i].oprright = atooid(PQgetvalue(res, i, i_oprright));
oprinfo[i].oprcode = atooid(PQgetvalue(res, i, i_oprcode));
/* Decide whether we want to dump it */
@@ -6070,6 +6085,7 @@ getCollations(Archive *fout, int *numCollations)
int i_collname;
int i_collnamespace;
int i_collowner;
+ int i_collencoding;
query = createPQExpBuffer();
@@ -6080,7 +6096,8 @@ getCollations(Archive *fout, int *numCollations)
appendPQExpBufferStr(query, "SELECT tableoid, oid, collname, "
"collnamespace, "
- "collowner "
+ "collowner, "
+ "collencoding "
"FROM pg_collation");
res = ExecuteSqlQuery(fout, query->data, PGRES_TUPLES_OK);
@@ -6095,6 +6112,7 @@ getCollations(Archive *fout, int *numCollations)
i_collname = PQfnumber(res, "collname");
i_collnamespace = PQfnumber(res, "collnamespace");
i_collowner = PQfnumber(res, "collowner");
+ i_collencoding = PQfnumber(res, "collencoding");
for (i = 0; i < ntups; i++)
{
@@ -6106,6 +6124,7 @@ getCollations(Archive *fout, int *numCollations)
collinfo[i].dobj.namespace =
findNamespace(atooid(PQgetvalue(res, i,
i_collnamespace)));
collinfo[i].rolname = getRoleName(PQgetvalue(res, i,
i_collowner));
+ collinfo[i].collencoding = atoi(PQgetvalue(res, i,
i_collencoding));
/* Decide whether we want to dump it */
selectDumpableObject(&(collinfo[i].dobj), fout);
@@ -6207,19 +6226,28 @@ getAccessMethods(Archive *fout, int *numAccessMethods)
int i_amhandler;
int i_amtype;
- /* Before 9.6, there are no user-defined access methods */
- if (fout->remoteVersion < 90600)
- {
- *numAccessMethods = 0;
- return NULL;
- }
-
query = createPQExpBuffer();
- /* Select all access methods from pg_am table */
- appendPQExpBufferStr(query, "SELECT tableoid, oid, amname, amtype, "
- "amhandler::pg_catalog.regproc
AS amhandler "
- "FROM pg_am");
+ /*
+ * Select all access methods from pg_am table. v9.6 introduced CREATE
+ * ACCESS METHOD, so earlier versions usually have only built-in access
+ * methods. v9.6 also changed the access method API, replacing dozens
of
+ * pg_am columns with amhandler. Even if a user created an access
method
+ * by "INSERT INTO pg_am", we have no way to translate pre-v9.6 pg_am
+ * columns to a v9.6+ CREATE ACCESS METHOD. Hence, before v9.6, read
+ * pg_am just to facilitate findAccessMethodByOid() providing the
+ * OID-to-name mapping.
+ */
+ appendPQExpBufferStr(query, "SELECT tableoid, oid, amname, ");
+ if (fout->remoteVersion >= 90600)
+ appendPQExpBufferStr(query,
+ "amtype, "
+
"amhandler::pg_catalog.regproc AS amhandler ");
+ else
+ appendPQExpBufferStr(query,
+
"'i'::pg_catalog.\"char\" AS amtype, "
+
"'-'::pg_catalog.regproc AS amhandler ");
+ appendPQExpBufferStr(query, "FROM pg_am");
res = ExecuteSqlQuery(fout, query->data, PGRES_TUPLES_OK);
@@ -6274,6 +6302,7 @@ getOpclasses(Archive *fout, int *numOpclasses)
OpclassInfo *opcinfo;
int i_tableoid;
int i_oid;
+ int i_opcmethod;
int i_opcname;
int i_opcnamespace;
int i_opcowner;
@@ -6283,7 +6312,7 @@ getOpclasses(Archive *fout, int *numOpclasses)
* system-defined opclasses at dump-out time.
*/
- appendPQExpBufferStr(query, "SELECT tableoid, oid, opcname, "
+ appendPQExpBufferStr(query, "SELECT tableoid, oid, opcmethod, opcname, "
"opcnamespace, "
"opcowner "
"FROM pg_opclass");
@@ -6297,6 +6326,7 @@ getOpclasses(Archive *fout, int *numOpclasses)
i_tableoid = PQfnumber(res, "tableoid");
i_oid = PQfnumber(res, "oid");
+ i_opcmethod = PQfnumber(res, "opcmethod");
i_opcname = PQfnumber(res, "opcname");
i_opcnamespace = PQfnumber(res, "opcnamespace");
i_opcowner = PQfnumber(res, "opcowner");
@@ -6310,6 +6340,7 @@ getOpclasses(Archive *fout, int *numOpclasses)
opcinfo[i].dobj.name = pg_strdup(PQgetvalue(res, i, i_opcname));
opcinfo[i].dobj.namespace =
findNamespace(atooid(PQgetvalue(res, i,
i_opcnamespace)));
+ opcinfo[i].opcmethod = atooid(PQgetvalue(res, i, i_opcmethod));
opcinfo[i].rolname = getRoleName(PQgetvalue(res, i,
i_opcowner));
/* Decide whether we want to dump it */
@@ -6340,6 +6371,7 @@ getOpfamilies(Archive *fout, int *numOpfamilies)
OpfamilyInfo *opfinfo;
int i_tableoid;
int i_oid;
+ int i_opfmethod;
int i_opfname;
int i_opfnamespace;
int i_opfowner;
@@ -6351,7 +6383,7 @@ getOpfamilies(Archive *fout, int *numOpfamilies)
* system-defined opfamilies at dump-out time.
*/
- appendPQExpBufferStr(query, "SELECT tableoid, oid, opfname, "
+ appendPQExpBufferStr(query, "SELECT tableoid, oid, opfmethod, opfname, "
"opfnamespace, "
"opfowner "
"FROM pg_opfamily");
@@ -6366,6 +6398,7 @@ getOpfamilies(Archive *fout, int *numOpfamilies)
i_tableoid = PQfnumber(res, "tableoid");
i_oid = PQfnumber(res, "oid");
i_opfname = PQfnumber(res, "opfname");
+ i_opfmethod = PQfnumber(res, "opfmethod");
i_opfnamespace = PQfnumber(res, "opfnamespace");
i_opfowner = PQfnumber(res, "opfowner");
@@ -6378,6 +6411,7 @@ getOpfamilies(Archive *fout, int *numOpfamilies)
opfinfo[i].dobj.name = pg_strdup(PQgetvalue(res, i, i_opfname));
opfinfo[i].dobj.namespace =
findNamespace(atooid(PQgetvalue(res, i,
i_opfnamespace)));
+ opfinfo[i].opfmethod = atooid(PQgetvalue(res, i, i_opfmethod));
opfinfo[i].rolname = getRoleName(PQgetvalue(res, i,
i_opfowner));
/* Decide whether we want to dump it */
diff --git a/src/bin/pg_dump/pg_dump.h b/src/bin/pg_dump/pg_dump.h
index 2439423..2de5afd 100644
--- a/src/bin/pg_dump/pg_dump.h
+++ b/src/bin/pg_dump/pg_dump.h
@@ -253,6 +253,8 @@ typedef struct _oprInfo
DumpableObject dobj;
const char *rolname;
char oprkind;
+ Oid oprleft;
+ Oid oprright;
Oid oprcode;
} OprInfo;
@@ -266,12 +268,14 @@ typedef struct _accessMethodInfo
typedef struct _opclassInfo
{
DumpableObject dobj;
+ Oid opcmethod;
const char *rolname;
} OpclassInfo;
typedef struct _opfamilyInfo
{
DumpableObject dobj;
+ Oid opfmethod;
const char *rolname;
} OpfamilyInfo;
@@ -279,6 +283,7 @@ typedef struct _collInfo
{
DumpableObject dobj;
const char *rolname;
+ int collencoding;
} CollInfo;
typedef struct _convInfo
@@ -723,6 +728,7 @@ extern TableInfo *findTableByOid(Oid oid);
extern TypeInfo *findTypeByOid(Oid oid);
extern FuncInfo *findFuncByOid(Oid oid);
extern OprInfo *findOprByOid(Oid oid);
+extern AccessMethodInfo *findAccessMethodByOid(Oid oid);
extern CollInfo *findCollationByOid(Oid oid);
extern NamespaceInfo *findNamespaceByOid(Oid oid);
extern ExtensionInfo *findExtensionByOid(Oid oid);
diff --git a/src/bin/pg_dump/pg_dump_sort.c b/src/bin/pg_dump/pg_dump_sort.c
index bb31c15..62e27b8 100644
--- a/src/bin/pg_dump/pg_dump_sort.c
+++ b/src/bin/pg_dump/pg_dump_sort.c
@@ -161,6 +161,8 @@ static DumpId postDataBoundId;
static int DOTypeNameCompare(const void *p1, const void *p2);
+static int pgTypeNameCompare(Oid typid1, Oid typid2);
+static int accessMethodNameCompare(Oid am1, Oid am2);
static bool TopoSort(DumpableObject **objs,
int numObjs,
DumpableObject **ordering,
@@ -227,12 +229,39 @@ DOTypeNameCompare(const void *p1, const void *p2)
else if (obj2->namespace)
return 1;
- /* Sort by name */
+ /*
+ * Sort by name. With a few exceptions, names here are single catalog
+ * columns. To get a fuller picture, grep pg_dump.c for "dobj.name = ".
+ * Names here don't match "Name:" in plain format output, which is a
+ * _tocEntry.tag. For example, DumpableObject.name of a constraint is
+ * pg_constraint.conname, but _tocEntry.tag of a constraint is relname
and
+ * conname joined with a space.
+ */
cmpval = strcmp(obj1->name, obj2->name);
if (cmpval != 0)
return cmpval;
- /* To have a stable sort order, break ties for some object types */
+ /*
+ * Sort by type. This helps types that share a type priority without
+ * sharing a unique name constraint, e.g. opclass and opfamily.
+ */
+ cmpval = obj1->objType - obj2->objType;
+ if (cmpval != 0)
+ return cmpval;
+
+ /*
+ * To have a stable sort order, break ties for some object types. Most
+ * catalogs have a natural key, e.g. pg_proc_proname_args_nsp_index.
Where
+ * the above "namespace" and "name" comparisons don't cover all natural
+ * key columns, compare the rest here.
+ *
+ * The natural key usually refers to other catalogs by surrogate keys.
+ * Hence, this translates each of those references to the natural key of
+ * the referenced catalog. That may descend through multiple levels of
+ * catalog references. For example, to sort by pg_proc.proargtypes,
+ * descend to each pg_type and then further to its pg_namespace, for an
+ * overall sort by (nspname, typname).
+ */
if (obj1->objType == DO_FUNC || obj1->objType == DO_AGG)
{
FuncInfo *fobj1 = *(FuncInfo *const *) p1;
@@ -245,22 +274,10 @@ DOTypeNameCompare(const void *p1, const void *p2)
return cmpval;
for (i = 0; i < fobj1->nargs; i++)
{
- TypeInfo *argtype1 =
findTypeByOid(fobj1->argtypes[i]);
- TypeInfo *argtype2 =
findTypeByOid(fobj2->argtypes[i]);
-
- if (argtype1 && argtype2)
- {
- if (argtype1->dobj.namespace &&
argtype2->dobj.namespace)
- {
- cmpval =
strcmp(argtype1->dobj.namespace->dobj.name,
-
argtype2->dobj.namespace->dobj.name);
- if (cmpval != 0)
- return cmpval;
- }
- cmpval = strcmp(argtype1->dobj.name,
argtype2->dobj.name);
- if (cmpval != 0)
- return cmpval;
- }
+ cmpval = pgTypeNameCompare(fobj1->argtypes[i],
+
fobj2->argtypes[i]);
+ if (cmpval != 0)
+ return cmpval;
}
}
else if (obj1->objType == DO_OPERATOR)
@@ -272,6 +289,57 @@ DOTypeNameCompare(const void *p1, const void *p2)
cmpval = (oobj2->oprkind - oobj1->oprkind);
if (cmpval != 0)
return cmpval;
+ /* Within an oprkind, sort by argument type names */
+ cmpval = pgTypeNameCompare(oobj1->oprleft, oobj2->oprleft);
+ if (cmpval != 0)
+ return cmpval;
+ cmpval = pgTypeNameCompare(oobj1->oprright, oobj2->oprright);
+ if (cmpval != 0)
+ return cmpval;
+ }
+ else if (obj1->objType == DO_OPCLASS)
+ {
+ OpclassInfo *opcobj1 = *(OpclassInfo *const *) p1;
+ OpclassInfo *opcobj2 = *(OpclassInfo *const *) p2;
+
+ /* Sort by access method name, per pg_opclass_am_name_nsp_index
*/
+ cmpval = accessMethodNameCompare(opcobj1->opcmethod,
+
opcobj2->opcmethod);
+ if (cmpval != 0)
+ return cmpval;
+ }
+ else if (obj1->objType == DO_OPFAMILY)
+ {
+ OpfamilyInfo *opfobj1 = *(OpfamilyInfo *const *) p1;
+ OpfamilyInfo *opfobj2 = *(OpfamilyInfo *const *) p2;
+
+ /* Sort by access method name, per
pg_opfamily_am_name_nsp_index */
+ cmpval = accessMethodNameCompare(opfobj1->opfmethod,
+
opfobj2->opfmethod);
+ if (cmpval != 0)
+ return cmpval;
+ }
+ else if (obj1->objType == DO_COLLATION)
+ {
+ CollInfo *cobj1 = *(CollInfo *const *) p1;
+ CollInfo *cobj2 = *(CollInfo *const *) p2;
+
+ /*
+ * Sort by encoding, per pg_collation_name_enc_nsp_index.
Technically,
+ * this is not necessary, because wherever this changes dump
order,
+ * restoring the dump fails anyway. CREATE COLLATION can't
create a
+ * tie for this to break, because it imposes restrictions to
make
+ * (nspname, collname) uniquely identify a collation within a
given
+ * DatabaseEncoding. While pg_import_system_collations() can
create a
+ * tie, pg_dump+restore fails after
+ * pg_import_system_collations('my_schema') does so. However,
there's
+ * little to gain by ignoring one natural key column on the
basis of
+ * those limitations elsewhere, so respect the full natural key
like
+ * we do for other object types.
+ */
+ cmpval = cobj1->collencoding - cobj2->collencoding;
+ if (cmpval != 0)
+ return cmpval;
}
else if (obj1->objType == DO_ATTRDEF)
{
@@ -316,11 +384,143 @@ DOTypeNameCompare(const void *p1, const void *p2)
if (cmpval != 0)
return cmpval;
}
+ else if (obj1->objType == DO_CONSTRAINT)
+ {
+ ConstraintInfo *robj1 = *(ConstraintInfo *const *) p1;
+ ConstraintInfo *robj2 = *(ConstraintInfo *const *) p2;
- /* Usually shouldn't get here, but if we do, sort by OID */
+ /*
+ * Sort domain constraints before table constraints, for
consistency
+ * with our decision to sort CREATE DOMAIN before CREATE TABLE.
+ */
+ if (robj1->condomain)
+ {
+ if (robj2->condomain)
+ {
+ /* Sort by domain name (domain namespace was
considered) */
+ cmpval = strcmp(robj1->condomain->dobj.name,
+
robj2->condomain->dobj.name);
+ if (cmpval != 0)
+ return cmpval;
+ }
+ else
+ return PRIO_TYPE - PRIO_TABLE;
+ }
+ else if (robj2->condomain)
+ return PRIO_TABLE - PRIO_TYPE;
+ else
+ {
+ /* Sort by table name (table namespace was considered
already) */
+ cmpval = strcmp(robj1->contable->dobj.name,
+
robj2->contable->dobj.name);
+ if (cmpval != 0)
+ return cmpval;
+ }
+ }
+ else if (obj1->objType == DO_PUBLICATION_REL)
+ {
+ PublicationRelInfo *probj1 = *(PublicationRelInfo *const *) p1;
+ PublicationRelInfo *probj2 = *(PublicationRelInfo *const *) p2;
+
+ /* Sort by publication name, since (namespace, name) match the
rel */
+ cmpval = strcmp(probj1->publication->dobj.name,
+ probj2->publication->dobj.name);
+ if (cmpval != 0)
+ return cmpval;
+ }
+ else if (obj1->objType == DO_PUBLICATION_TABLE_IN_SCHEMA)
+ {
+ PublicationSchemaInfo *psobj1 = *(PublicationSchemaInfo *const
*) p1;
+ PublicationSchemaInfo *psobj2 = *(PublicationSchemaInfo *const
*) p2;
+
+ /* Sort by publication name, since ->name is just nspname */
+ cmpval = strcmp(psobj1->publication->dobj.name,
+ psobj2->publication->dobj.name);
+ if (cmpval != 0)
+ return cmpval;
+ }
+
+ /*
+ * Shouldn't get here except after catalog corruption, but if we do,
sort
+ * by OID. This may make logically-identical databases differ in the
+ * order of objects in dump output. Users will get spurious schema
diffs.
+ * Expect flaky failures of 002_pg_upgrade.pl test 'dump outputs from
+ * original and restored regression databases match' if the regression
+ * database contains objects allowing that test to reach here. That's a
+ * consequence of the test using "pg_restore -j", which doesn't fully
+ * constrain OID assignment order.
+ */
+ Assert(false);
return oidcmp(obj1->catId.oid, obj2->catId.oid);
}
+/* Compare two OID-identified pg_type values by nspname, then by typname. */
+static int
+pgTypeNameCompare(Oid typid1, Oid typid2)
+{
+ TypeInfo *typobj1;
+ TypeInfo *typobj2;
+ int cmpval;
+
+ if (typid1 == typid2)
+ return 0;
+
+ typobj1 = findTypeByOid(typid1);
+ typobj2 = findTypeByOid(typid2);
+
+ if (!typobj1 || !typobj2)
+ {
+ /*
+ * getTypes() didn't find some OID. Assume catalog corruption,
e.g.
+ * an oprright value without the corresponding OID in a pg_type
row.
+ * Report as "equal", so the caller uses the next available
basis for
+ * comparison, e.g. the next function argument.
+ *
+ * Unary operators have InvalidOid in oprleft (if oprkind='r')
or in
+ * oprright (if oprkind='l'). Caller already sorted by oprkind,
+ * calling us only for like-kind operators. Hence, "typid1 ==
typid2"
+ * took care of InvalidOid. (v14 removed postfix operator
support.
+ * Hence, when dumping from v14+, only oprleft can be
InvalidOid.)
+ */
+ Assert(false);
+ return 0;
+ }
+
+ if (!typobj1->dobj.namespace || !typobj2->dobj.namespace)
+ Assert(false); /* catalog corruption */
+ else
+ {
+ cmpval = strcmp(typobj1->dobj.namespace->dobj.name,
+
typobj2->dobj.namespace->dobj.name);
+ if (cmpval != 0)
+ return cmpval;
+ }
+ return strcmp(typobj1->dobj.name, typobj2->dobj.name);
+}
+
+/* Compare two OID-identified pg_am values by amname. */
+static int
+accessMethodNameCompare(Oid am1, Oid am2)
+{
+ AccessMethodInfo *amobj1;
+ AccessMethodInfo *amobj2;
+
+ if (am1 == am2)
+ return 0;
+
+ amobj1 = findAccessMethodByOid(am1);
+ amobj2 = findAccessMethodByOid(am2);
+
+ if (!amobj1 || !amobj2)
+ {
+ /* catalog corruption: handle like pgTypeNameCompare() does */
+ Assert(false);
+ return 0;
+ }
+
+ return strcmp(amobj1->dobj.name, amobj2->dobj.name);
+}
+
/*
* Sort the given objects into a safe dump order using dependency
diff --git a/src/test/regress/expected/publication.out
b/src/test/regress/expected/publication.out
index 30b6371..3edf0be 100644
--- a/src/test/regress/expected/publication.out
+++ b/src/test/regress/expected/publication.out
@@ -1745,3 +1745,24 @@ DROP SCHEMA sch2 cascade;
RESET SESSION AUTHORIZATION;
DROP ROLE regress_publication_user, regress_publication_user2;
DROP ROLE regress_publication_user_dummy;
+-- stage objects for pg_dump tests
+CREATE SCHEMA pubme CREATE TABLE t0 (c int, d int) CREATE TABLE t1 (c int);
+CREATE SCHEMA pubme2 CREATE TABLE t0 (c int, d int);
+SET client_min_messages = 'ERROR';
+CREATE PUBLICATION dump_pub_qual_1ct FOR
+ TABLE ONLY pubme.t0 (c, d) WHERE (c > 0);
+CREATE PUBLICATION dump_pub_qual_2ct FOR
+ TABLE ONLY pubme.t0 (c) WHERE (c > 0),
+ TABLE ONLY pubme.t1 (c);
+CREATE PUBLICATION dump_pub_nsp_1ct FOR
+ TABLES IN SCHEMA pubme;
+CREATE PUBLICATION dump_pub_nsp_2ct FOR
+ TABLES IN SCHEMA pubme,
+ TABLES IN SCHEMA pubme2;
+CREATE PUBLICATION dump_pub_all FOR
+ TABLE ONLY pubme.t0,
+ TABLE ONLY pubme.t1 WHERE (c < 0),
+ TABLES IN SCHEMA pubme,
+ TABLES IN SCHEMA pubme2
+ WITH (publish_via_partition_root = true);
+RESET client_min_messages;
diff --git a/src/test/regress/sql/publication.sql
b/src/test/regress/sql/publication.sql
index 479d4f3..c4f12d4 100644
--- a/src/test/regress/sql/publication.sql
+++ b/src/test/regress/sql/publication.sql
@@ -1109,3 +1109,25 @@ DROP SCHEMA sch2 cascade;
RESET SESSION AUTHORIZATION;
DROP ROLE regress_publication_user, regress_publication_user2;
DROP ROLE regress_publication_user_dummy;
+
+-- stage objects for pg_dump tests
+CREATE SCHEMA pubme CREATE TABLE t0 (c int, d int) CREATE TABLE t1 (c int);
+CREATE SCHEMA pubme2 CREATE TABLE t0 (c int, d int);
+SET client_min_messages = 'ERROR';
+CREATE PUBLICATION dump_pub_qual_1ct FOR
+ TABLE ONLY pubme.t0 (c, d) WHERE (c > 0);
+CREATE PUBLICATION dump_pub_qual_2ct FOR
+ TABLE ONLY pubme.t0 (c) WHERE (c > 0),
+ TABLE ONLY pubme.t1 (c);
+CREATE PUBLICATION dump_pub_nsp_1ct FOR
+ TABLES IN SCHEMA pubme;
+CREATE PUBLICATION dump_pub_nsp_2ct FOR
+ TABLES IN SCHEMA pubme,
+ TABLES IN SCHEMA pubme2;
+CREATE PUBLICATION dump_pub_all FOR
+ TABLE ONLY pubme.t0,
+ TABLE ONLY pubme.t1 WHERE (c < 0),
+ TABLES IN SCHEMA pubme,
+ TABLES IN SCHEMA pubme2
+ WITH (publish_via_partition_root = true);
+RESET client_min_messages;
commit 4ca0d70 (HEAD, zzy_test-commit-REL_15_STABLE)
Author: Noah Misch <n...@leadboat.com>
AuthorDate: Thu Jul 24 17:25:43 2025 -0700
Commit: Noah Misch <n...@leadboat.com>
CommitDate: Thu Jul 24 17:26:18 2025 -0700
Sort dump objects independent of OIDs, for the 7 holdout object types.
pg_dump sorts objects by their logical names, e.g. (nspname, relname,
tgname), before dependency-driven reordering. That removes one source
of logically-identical databases differing in their schema-only dumps.
In other words, it helps with schema diffing. The logical name sort
ignored essential sort keys for constraints, operators, PUBLICATION
... FOR TABLE, PUBLICATION ... FOR TABLES IN SCHEMA, operator classes,
and operator families. pg_dump's sort then depended on object OID,
yielding spurious schema diffs. After this change, OIDs affect dump
order only in the event of catalog corruption. While pg_dump also
wrongly ignored pg_collation.collencoding, CREATE COLLATION restrictions
have been keeping that imperceptible in practical use.
Use techniques like we use for object types already having full sort key
coverage. Where the pertinent queries weren't fetching the ignored sort
keys, this adds columns to those queries and stores those keys in memory
for the long term.
The ignorance of sort keys became more problematic when commit
172259afb563d35001410dc6daad78b250924038 added a schema diff test
sensitive to it. However, dump order stability isn't a new goal, and
this might avoid other dump comparison failures. Hence, back-patch to
v13 (all supported versions).
Reviewed-by: Robert Haas <robertmh...@gmail.com>
Discussion: https://postgr.es/m/20250707192654.9e.nmi...@google.com
Backpatch-through: 13
Conflicts:
src/bin/pg_dump/pg_dump.c
---
src/bin/pg_dump/common.c | 19 +++
src/bin/pg_dump/pg_dump.c | 62 ++++++--
src/bin/pg_dump/pg_dump.h | 6 +
src/bin/pg_dump/pg_dump_sort.c | 238 +++++++++++++++++++++++++++---
src/test/regress/expected/publication.out | 21 +++
src/test/regress/sql/publication.sql | 22 +++
6 files changed, 335 insertions(+), 33 deletions(-)
diff --git a/src/bin/pg_dump/common.c b/src/bin/pg_dump/common.c
index a64d37e..95f8980 100644
--- a/src/bin/pg_dump/common.c
+++ b/src/bin/pg_dump/common.c
@@ -17,6 +17,7 @@
#include <ctype.h>
+#include "catalog/pg_am_d.h"
#include "catalog/pg_class_d.h"
#include "catalog/pg_collation_d.h"
#include "catalog/pg_extension_d.h"
@@ -852,6 +853,24 @@ findOprByOid(Oid oid)
}
/*
+ * findAccessMethodByOid
+ * finds the DumpableObject for the access method with the given oid
+ * returns NULL if not found
+ */
+AccessMethodInfo *
+findAccessMethodByOid(Oid oid)
+{
+ CatalogId catId;
+ DumpableObject *dobj;
+
+ catId.tableoid = AccessMethodRelationId;
+ catId.oid = oid;
+ dobj = findObjectByCatalogId(catId);
+ Assert(dobj == NULL || dobj->objType == DO_ACCESS_METHOD);
+ return (AccessMethodInfo *) dobj;
+}
+
+/*
* findCollationByOid
* finds the DumpableObject for the collation with the given oid
* returns NULL if not found
diff --git a/src/bin/pg_dump/pg_dump.c b/src/bin/pg_dump/pg_dump.c
index e85f220..54f07b9 100644
--- a/src/bin/pg_dump/pg_dump.c
+++ b/src/bin/pg_dump/pg_dump.c
@@ -1874,6 +1874,13 @@ selectDumpableProcLang(ProcLangInfo *plang, Archive
*fout)
static void
selectDumpableAccessMethod(AccessMethodInfo *method, Archive *fout)
{
+ /* see getAccessMethods() comment about v9.6. */
+ if (fout->remoteVersion < 90600)
+ {
+ method->dobj.dump = DUMP_COMPONENT_NONE;
+ return;
+ }
+
if (checkExtensionMembership(&method->dobj, fout))
return; /* extension membership
overrides all else */
@@ -5496,6 +5503,8 @@ getOperators(Archive *fout, int *numOprs)
int i_oprnamespace;
int i_oprowner;
int i_oprkind;
+ int i_oprleft;
+ int i_oprright;
int i_oprcode;
/*
@@ -5507,6 +5516,8 @@ getOperators(Archive *fout, int *numOprs)
"oprnamespace, "
"oprowner, "
"oprkind, "
+ "oprleft, "
+ "oprright, "
"oprcode::oid AS oprcode "
"FROM pg_operator");
@@ -5523,6 +5534,8 @@ getOperators(Archive *fout, int *numOprs)
i_oprnamespace = PQfnumber(res, "oprnamespace");
i_oprowner = PQfnumber(res, "oprowner");
i_oprkind = PQfnumber(res, "oprkind");
+ i_oprleft = PQfnumber(res, "oprleft");
+ i_oprright = PQfnumber(res, "oprright");
i_oprcode = PQfnumber(res, "oprcode");
for (i = 0; i < ntups; i++)
@@ -5536,6 +5549,8 @@ getOperators(Archive *fout, int *numOprs)
findNamespace(atooid(PQgetvalue(res, i,
i_oprnamespace)));
oprinfo[i].rolname = getRoleName(PQgetvalue(res, i,
i_oprowner));
oprinfo[i].oprkind = (PQgetvalue(res, i, i_oprkind))[0];
+ oprinfo[i].oprleft = atooid(PQgetvalue(res, i, i_oprleft));
+ oprinfo[i].oprright = atooid(PQgetvalue(res, i, i_oprright));
oprinfo[i].oprcode = atooid(PQgetvalue(res, i, i_oprcode));
/* Decide whether we want to dump it */
@@ -5569,6 +5584,7 @@ getCollations(Archive *fout, int *numCollations)
int i_collname;
int i_collnamespace;
int i_collowner;
+ int i_collencoding;
query = createPQExpBuffer();
@@ -5579,7 +5595,8 @@ getCollations(Archive *fout, int *numCollations)
appendPQExpBuffer(query, "SELECT tableoid, oid, collname, "
"collnamespace, "
- "collowner "
+ "collowner, "
+ "collencoding "
"FROM pg_collation");
res = ExecuteSqlQuery(fout, query->data, PGRES_TUPLES_OK);
@@ -5594,6 +5611,7 @@ getCollations(Archive *fout, int *numCollations)
i_collname = PQfnumber(res, "collname");
i_collnamespace = PQfnumber(res, "collnamespace");
i_collowner = PQfnumber(res, "collowner");
+ i_collencoding = PQfnumber(res, "collencoding");
for (i = 0; i < ntups; i++)
{
@@ -5605,6 +5623,7 @@ getCollations(Archive *fout, int *numCollations)
collinfo[i].dobj.namespace =
findNamespace(atooid(PQgetvalue(res, i,
i_collnamespace)));
collinfo[i].rolname = getRoleName(PQgetvalue(res, i,
i_collowner));
+ collinfo[i].collencoding = atoi(PQgetvalue(res, i,
i_collencoding));
/* Decide whether we want to dump it */
selectDumpableObject(&(collinfo[i].dobj), fout);
@@ -5706,19 +5725,28 @@ getAccessMethods(Archive *fout, int *numAccessMethods)
int i_amhandler;
int i_amtype;
- /* Before 9.6, there are no user-defined access methods */
- if (fout->remoteVersion < 90600)
- {
- *numAccessMethods = 0;
- return NULL;
- }
-
query = createPQExpBuffer();
- /* Select all access methods from pg_am table */
- appendPQExpBufferStr(query, "SELECT tableoid, oid, amname, amtype, "
- "amhandler::pg_catalog.regproc
AS amhandler "
- "FROM pg_am");
+ /*
+ * Select all access methods from pg_am table. v9.6 introduced CREATE
+ * ACCESS METHOD, so earlier versions usually have only built-in access
+ * methods. v9.6 also changed the access method API, replacing dozens
of
+ * pg_am columns with amhandler. Even if a user created an access
method
+ * by "INSERT INTO pg_am", we have no way to translate pre-v9.6 pg_am
+ * columns to a v9.6+ CREATE ACCESS METHOD. Hence, before v9.6, read
+ * pg_am just to facilitate findAccessMethodByOid() providing the
+ * OID-to-name mapping.
+ */
+ appendPQExpBufferStr(query, "SELECT tableoid, oid, amname, ");
+ if (fout->remoteVersion >= 90600)
+ appendPQExpBufferStr(query,
+ "amtype, "
+
"amhandler::pg_catalog.regproc AS amhandler ");
+ else
+ appendPQExpBufferStr(query,
+
"'i'::pg_catalog.\"char\" AS amtype, "
+
"'-'::pg_catalog.regproc AS amhandler ");
+ appendPQExpBufferStr(query, "FROM pg_am");
res = ExecuteSqlQuery(fout, query->data, PGRES_TUPLES_OK);
@@ -5773,6 +5801,7 @@ getOpclasses(Archive *fout, int *numOpclasses)
OpclassInfo *opcinfo;
int i_tableoid;
int i_oid;
+ int i_opcmethod;
int i_opcname;
int i_opcnamespace;
int i_opcowner;
@@ -5782,7 +5811,7 @@ getOpclasses(Archive *fout, int *numOpclasses)
* system-defined opclasses at dump-out time.
*/
- appendPQExpBuffer(query, "SELECT tableoid, oid, opcname, "
+ appendPQExpBuffer(query, "SELECT tableoid, oid, opcmethod, opcname, "
"opcnamespace, "
"opcowner "
"FROM pg_opclass");
@@ -5796,6 +5825,7 @@ getOpclasses(Archive *fout, int *numOpclasses)
i_tableoid = PQfnumber(res, "tableoid");
i_oid = PQfnumber(res, "oid");
+ i_opcmethod = PQfnumber(res, "opcmethod");
i_opcname = PQfnumber(res, "opcname");
i_opcnamespace = PQfnumber(res, "opcnamespace");
i_opcowner = PQfnumber(res, "opcowner");
@@ -5809,6 +5839,7 @@ getOpclasses(Archive *fout, int *numOpclasses)
opcinfo[i].dobj.name = pg_strdup(PQgetvalue(res, i, i_opcname));
opcinfo[i].dobj.namespace =
findNamespace(atooid(PQgetvalue(res, i,
i_opcnamespace)));
+ opcinfo[i].opcmethod = atooid(PQgetvalue(res, i, i_opcmethod));
opcinfo[i].rolname = getRoleName(PQgetvalue(res, i,
i_opcowner));
/* Decide whether we want to dump it */
@@ -5839,6 +5870,7 @@ getOpfamilies(Archive *fout, int *numOpfamilies)
OpfamilyInfo *opfinfo;
int i_tableoid;
int i_oid;
+ int i_opfmethod;
int i_opfname;
int i_opfnamespace;
int i_opfowner;
@@ -5850,7 +5882,7 @@ getOpfamilies(Archive *fout, int *numOpfamilies)
* system-defined opfamilies at dump-out time.
*/
- appendPQExpBuffer(query, "SELECT tableoid, oid, opfname, "
+ appendPQExpBuffer(query, "SELECT tableoid, oid, opfmethod, opfname, "
"opfnamespace, "
"opfowner "
"FROM pg_opfamily");
@@ -5865,6 +5897,7 @@ getOpfamilies(Archive *fout, int *numOpfamilies)
i_tableoid = PQfnumber(res, "tableoid");
i_oid = PQfnumber(res, "oid");
i_opfname = PQfnumber(res, "opfname");
+ i_opfmethod = PQfnumber(res, "opfmethod");
i_opfnamespace = PQfnumber(res, "opfnamespace");
i_opfowner = PQfnumber(res, "opfowner");
@@ -5877,6 +5910,7 @@ getOpfamilies(Archive *fout, int *numOpfamilies)
opfinfo[i].dobj.name = pg_strdup(PQgetvalue(res, i, i_opfname));
opfinfo[i].dobj.namespace =
findNamespace(atooid(PQgetvalue(res, i,
i_opfnamespace)));
+ opfinfo[i].opfmethod = atooid(PQgetvalue(res, i, i_opfmethod));
opfinfo[i].rolname = getRoleName(PQgetvalue(res, i,
i_opfowner));
/* Decide whether we want to dump it */
diff --git a/src/bin/pg_dump/pg_dump.h b/src/bin/pg_dump/pg_dump.h
index 774ecb2..a64859a7 100644
--- a/src/bin/pg_dump/pg_dump.h
+++ b/src/bin/pg_dump/pg_dump.h
@@ -250,6 +250,8 @@ typedef struct _oprInfo
DumpableObject dobj;
const char *rolname;
char oprkind;
+ Oid oprleft;
+ Oid oprright;
Oid oprcode;
} OprInfo;
@@ -263,12 +265,14 @@ typedef struct _accessMethodInfo
typedef struct _opclassInfo
{
DumpableObject dobj;
+ Oid opcmethod;
const char *rolname;
} OpclassInfo;
typedef struct _opfamilyInfo
{
DumpableObject dobj;
+ Oid opfmethod;
const char *rolname;
} OpfamilyInfo;
@@ -276,6 +280,7 @@ typedef struct _collInfo
{
DumpableObject dobj;
const char *rolname;
+ int collencoding;
} CollInfo;
typedef struct _convInfo
@@ -694,6 +699,7 @@ extern TableInfo *findTableByOid(Oid oid);
extern TypeInfo *findTypeByOid(Oid oid);
extern FuncInfo *findFuncByOid(Oid oid);
extern OprInfo *findOprByOid(Oid oid);
+extern AccessMethodInfo *findAccessMethodByOid(Oid oid);
extern CollInfo *findCollationByOid(Oid oid);
extern NamespaceInfo *findNamespaceByOid(Oid oid);
extern ExtensionInfo *findExtensionByOid(Oid oid);
diff --git a/src/bin/pg_dump/pg_dump_sort.c b/src/bin/pg_dump/pg_dump_sort.c
index f5152ec..38ad6aa 100644
--- a/src/bin/pg_dump/pg_dump_sort.c
+++ b/src/bin/pg_dump/pg_dump_sort.c
@@ -157,6 +157,8 @@ static DumpId postDataBoundId;
static int DOTypeNameCompare(const void *p1, const void *p2);
+static int pgTypeNameCompare(Oid typid1, Oid typid2);
+static int accessMethodNameCompare(Oid am1, Oid am2);
static bool TopoSort(DumpableObject **objs,
int numObjs,
DumpableObject **ordering,
@@ -224,12 +226,39 @@ DOTypeNameCompare(const void *p1, const void *p2)
else if (obj2->namespace)
return 1;
- /* Sort by name */
+ /*
+ * Sort by name. With a few exceptions, names here are single catalog
+ * columns. To get a fuller picture, grep pg_dump.c for "dobj.name = ".
+ * Names here don't match "Name:" in plain format output, which is a
+ * _tocEntry.tag. For example, DumpableObject.name of a constraint is
+ * pg_constraint.conname, but _tocEntry.tag of a constraint is relname
and
+ * conname joined with a space.
+ */
cmpval = strcmp(obj1->name, obj2->name);
if (cmpval != 0)
return cmpval;
- /* To have a stable sort order, break ties for some object types */
+ /*
+ * Sort by type. This helps types that share a type priority without
+ * sharing a unique name constraint, e.g. opclass and opfamily.
+ */
+ cmpval = obj1->objType - obj2->objType;
+ if (cmpval != 0)
+ return cmpval;
+
+ /*
+ * To have a stable sort order, break ties for some object types. Most
+ * catalogs have a natural key, e.g. pg_proc_proname_args_nsp_index.
Where
+ * the above "namespace" and "name" comparisons don't cover all natural
+ * key columns, compare the rest here.
+ *
+ * The natural key usually refers to other catalogs by surrogate keys.
+ * Hence, this translates each of those references to the natural key of
+ * the referenced catalog. That may descend through multiple levels of
+ * catalog references. For example, to sort by pg_proc.proargtypes,
+ * descend to each pg_type and then further to its pg_namespace, for an
+ * overall sort by (nspname, typname).
+ */
if (obj1->objType == DO_FUNC || obj1->objType == DO_AGG)
{
FuncInfo *fobj1 = *(FuncInfo *const *) p1;
@@ -242,22 +271,10 @@ DOTypeNameCompare(const void *p1, const void *p2)
return cmpval;
for (i = 0; i < fobj1->nargs; i++)
{
- TypeInfo *argtype1 =
findTypeByOid(fobj1->argtypes[i]);
- TypeInfo *argtype2 =
findTypeByOid(fobj2->argtypes[i]);
-
- if (argtype1 && argtype2)
- {
- if (argtype1->dobj.namespace &&
argtype2->dobj.namespace)
- {
- cmpval =
strcmp(argtype1->dobj.namespace->dobj.name,
-
argtype2->dobj.namespace->dobj.name);
- if (cmpval != 0)
- return cmpval;
- }
- cmpval = strcmp(argtype1->dobj.name,
argtype2->dobj.name);
- if (cmpval != 0)
- return cmpval;
- }
+ cmpval = pgTypeNameCompare(fobj1->argtypes[i],
+
fobj2->argtypes[i]);
+ if (cmpval != 0)
+ return cmpval;
}
}
else if (obj1->objType == DO_OPERATOR)
@@ -269,6 +286,57 @@ DOTypeNameCompare(const void *p1, const void *p2)
cmpval = (oobj2->oprkind - oobj1->oprkind);
if (cmpval != 0)
return cmpval;
+ /* Within an oprkind, sort by argument type names */
+ cmpval = pgTypeNameCompare(oobj1->oprleft, oobj2->oprleft);
+ if (cmpval != 0)
+ return cmpval;
+ cmpval = pgTypeNameCompare(oobj1->oprright, oobj2->oprright);
+ if (cmpval != 0)
+ return cmpval;
+ }
+ else if (obj1->objType == DO_OPCLASS)
+ {
+ OpclassInfo *opcobj1 = *(OpclassInfo *const *) p1;
+ OpclassInfo *opcobj2 = *(OpclassInfo *const *) p2;
+
+ /* Sort by access method name, per pg_opclass_am_name_nsp_index
*/
+ cmpval = accessMethodNameCompare(opcobj1->opcmethod,
+
opcobj2->opcmethod);
+ if (cmpval != 0)
+ return cmpval;
+ }
+ else if (obj1->objType == DO_OPFAMILY)
+ {
+ OpfamilyInfo *opfobj1 = *(OpfamilyInfo *const *) p1;
+ OpfamilyInfo *opfobj2 = *(OpfamilyInfo *const *) p2;
+
+ /* Sort by access method name, per
pg_opfamily_am_name_nsp_index */
+ cmpval = accessMethodNameCompare(opfobj1->opfmethod,
+
opfobj2->opfmethod);
+ if (cmpval != 0)
+ return cmpval;
+ }
+ else if (obj1->objType == DO_COLLATION)
+ {
+ CollInfo *cobj1 = *(CollInfo *const *) p1;
+ CollInfo *cobj2 = *(CollInfo *const *) p2;
+
+ /*
+ * Sort by encoding, per pg_collation_name_enc_nsp_index.
Technically,
+ * this is not necessary, because wherever this changes dump
order,
+ * restoring the dump fails anyway. CREATE COLLATION can't
create a
+ * tie for this to break, because it imposes restrictions to
make
+ * (nspname, collname) uniquely identify a collation within a
given
+ * DatabaseEncoding. While pg_import_system_collations() can
create a
+ * tie, pg_dump+restore fails after
+ * pg_import_system_collations('my_schema') does so. However,
there's
+ * little to gain by ignoring one natural key column on the
basis of
+ * those limitations elsewhere, so respect the full natural key
like
+ * we do for other object types.
+ */
+ cmpval = cobj1->collencoding - cobj2->collencoding;
+ if (cmpval != 0)
+ return cmpval;
}
else if (obj1->objType == DO_ATTRDEF)
{
@@ -313,11 +381,143 @@ DOTypeNameCompare(const void *p1, const void *p2)
if (cmpval != 0)
return cmpval;
}
+ else if (obj1->objType == DO_CONSTRAINT)
+ {
+ ConstraintInfo *robj1 = *(ConstraintInfo *const *) p1;
+ ConstraintInfo *robj2 = *(ConstraintInfo *const *) p2;
- /* Usually shouldn't get here, but if we do, sort by OID */
+ /*
+ * Sort domain constraints before table constraints, for
consistency
+ * with our decision to sort CREATE DOMAIN before CREATE TABLE.
+ */
+ if (robj1->condomain)
+ {
+ if (robj2->condomain)
+ {
+ /* Sort by domain name (domain namespace was
considered) */
+ cmpval = strcmp(robj1->condomain->dobj.name,
+
robj2->condomain->dobj.name);
+ if (cmpval != 0)
+ return cmpval;
+ }
+ else
+ return PRIO_TYPE - PRIO_TABLE;
+ }
+ else if (robj2->condomain)
+ return PRIO_TABLE - PRIO_TYPE;
+ else
+ {
+ /* Sort by table name (table namespace was considered
already) */
+ cmpval = strcmp(robj1->contable->dobj.name,
+
robj2->contable->dobj.name);
+ if (cmpval != 0)
+ return cmpval;
+ }
+ }
+ else if (obj1->objType == DO_PUBLICATION_REL)
+ {
+ PublicationRelInfo *probj1 = *(PublicationRelInfo *const *) p1;
+ PublicationRelInfo *probj2 = *(PublicationRelInfo *const *) p2;
+
+ /* Sort by publication name, since (namespace, name) match the
rel */
+ cmpval = strcmp(probj1->publication->dobj.name,
+ probj2->publication->dobj.name);
+ if (cmpval != 0)
+ return cmpval;
+ }
+ else if (obj1->objType == DO_PUBLICATION_TABLE_IN_SCHEMA)
+ {
+ PublicationSchemaInfo *psobj1 = *(PublicationSchemaInfo *const
*) p1;
+ PublicationSchemaInfo *psobj2 = *(PublicationSchemaInfo *const
*) p2;
+
+ /* Sort by publication name, since ->name is just nspname */
+ cmpval = strcmp(psobj1->publication->dobj.name,
+ psobj2->publication->dobj.name);
+ if (cmpval != 0)
+ return cmpval;
+ }
+
+ /*
+ * Shouldn't get here except after catalog corruption, but if we do,
sort
+ * by OID. This may make logically-identical databases differ in the
+ * order of objects in dump output. Users will get spurious schema
diffs.
+ * Expect flaky failures of 002_pg_upgrade.pl test 'dump outputs from
+ * original and restored regression databases match' if the regression
+ * database contains objects allowing that test to reach here. That's a
+ * consequence of the test using "pg_restore -j", which doesn't fully
+ * constrain OID assignment order.
+ */
+ Assert(false);
return oidcmp(obj1->catId.oid, obj2->catId.oid);
}
+/* Compare two OID-identified pg_type values by nspname, then by typname. */
+static int
+pgTypeNameCompare(Oid typid1, Oid typid2)
+{
+ TypeInfo *typobj1;
+ TypeInfo *typobj2;
+ int cmpval;
+
+ if (typid1 == typid2)
+ return 0;
+
+ typobj1 = findTypeByOid(typid1);
+ typobj2 = findTypeByOid(typid2);
+
+ if (!typobj1 || !typobj2)
+ {
+ /*
+ * getTypes() didn't find some OID. Assume catalog corruption,
e.g.
+ * an oprright value without the corresponding OID in a pg_type
row.
+ * Report as "equal", so the caller uses the next available
basis for
+ * comparison, e.g. the next function argument.
+ *
+ * Unary operators have InvalidOid in oprleft (if oprkind='r')
or in
+ * oprright (if oprkind='l'). Caller already sorted by oprkind,
+ * calling us only for like-kind operators. Hence, "typid1 ==
typid2"
+ * took care of InvalidOid. (v14 removed postfix operator
support.
+ * Hence, when dumping from v14+, only oprleft can be
InvalidOid.)
+ */
+ Assert(false);
+ return 0;
+ }
+
+ if (!typobj1->dobj.namespace || !typobj2->dobj.namespace)
+ Assert(false); /* catalog corruption */
+ else
+ {
+ cmpval = strcmp(typobj1->dobj.namespace->dobj.name,
+
typobj2->dobj.namespace->dobj.name);
+ if (cmpval != 0)
+ return cmpval;
+ }
+ return strcmp(typobj1->dobj.name, typobj2->dobj.name);
+}
+
+/* Compare two OID-identified pg_am values by amname. */
+static int
+accessMethodNameCompare(Oid am1, Oid am2)
+{
+ AccessMethodInfo *amobj1;
+ AccessMethodInfo *amobj2;
+
+ if (am1 == am2)
+ return 0;
+
+ amobj1 = findAccessMethodByOid(am1);
+ amobj2 = findAccessMethodByOid(am2);
+
+ if (!amobj1 || !amobj2)
+ {
+ /* catalog corruption: handle like pgTypeNameCompare() does */
+ Assert(false);
+ return 0;
+ }
+
+ return strcmp(amobj1->dobj.name, amobj2->dobj.name);
+}
+
/*
* Sort the given objects into a safe dump order using dependency
diff --git a/src/test/regress/expected/publication.out
b/src/test/regress/expected/publication.out
index 69dc6cf..e8d907c 100644
--- a/src/test/regress/expected/publication.out
+++ b/src/test/regress/expected/publication.out
@@ -1735,3 +1735,24 @@ DROP SCHEMA sch2 cascade;
RESET SESSION AUTHORIZATION;
DROP ROLE regress_publication_user, regress_publication_user2;
DROP ROLE regress_publication_user_dummy;
+-- stage objects for pg_dump tests
+CREATE SCHEMA pubme CREATE TABLE t0 (c int, d int) CREATE TABLE t1 (c int);
+CREATE SCHEMA pubme2 CREATE TABLE t0 (c int, d int);
+SET client_min_messages = 'ERROR';
+CREATE PUBLICATION dump_pub_qual_1ct FOR
+ TABLE ONLY pubme.t0 (c, d) WHERE (c > 0);
+CREATE PUBLICATION dump_pub_qual_2ct FOR
+ TABLE ONLY pubme.t0 (c) WHERE (c > 0),
+ TABLE ONLY pubme.t1 (c);
+CREATE PUBLICATION dump_pub_nsp_1ct FOR
+ TABLES IN SCHEMA pubme;
+CREATE PUBLICATION dump_pub_nsp_2ct FOR
+ TABLES IN SCHEMA pubme,
+ TABLES IN SCHEMA pubme2;
+CREATE PUBLICATION dump_pub_all FOR
+ TABLE ONLY pubme.t0,
+ TABLE ONLY pubme.t1 WHERE (c < 0),
+ TABLES IN SCHEMA pubme,
+ TABLES IN SCHEMA pubme2
+ WITH (publish_via_partition_root = true);
+RESET client_min_messages;
diff --git a/src/test/regress/sql/publication.sql
b/src/test/regress/sql/publication.sql
index d5051a5..46e1489 100644
--- a/src/test/regress/sql/publication.sql
+++ b/src/test/regress/sql/publication.sql
@@ -1100,3 +1100,25 @@ DROP SCHEMA sch2 cascade;
RESET SESSION AUTHORIZATION;
DROP ROLE regress_publication_user, regress_publication_user2;
DROP ROLE regress_publication_user_dummy;
+
+-- stage objects for pg_dump tests
+CREATE SCHEMA pubme CREATE TABLE t0 (c int, d int) CREATE TABLE t1 (c int);
+CREATE SCHEMA pubme2 CREATE TABLE t0 (c int, d int);
+SET client_min_messages = 'ERROR';
+CREATE PUBLICATION dump_pub_qual_1ct FOR
+ TABLE ONLY pubme.t0 (c, d) WHERE (c > 0);
+CREATE PUBLICATION dump_pub_qual_2ct FOR
+ TABLE ONLY pubme.t0 (c) WHERE (c > 0),
+ TABLE ONLY pubme.t1 (c);
+CREATE PUBLICATION dump_pub_nsp_1ct FOR
+ TABLES IN SCHEMA pubme;
+CREATE PUBLICATION dump_pub_nsp_2ct FOR
+ TABLES IN SCHEMA pubme,
+ TABLES IN SCHEMA pubme2;
+CREATE PUBLICATION dump_pub_all FOR
+ TABLE ONLY pubme.t0,
+ TABLE ONLY pubme.t1 WHERE (c < 0),
+ TABLES IN SCHEMA pubme,
+ TABLES IN SCHEMA pubme2
+ WITH (publish_via_partition_root = true);
+RESET client_min_messages;
commit 57c5fae (HEAD, zzy_test-commit-REL_14_STABLE)
Author: Noah Misch <n...@leadboat.com>
AuthorDate: Thu Jul 24 17:26:41 2025 -0700
Commit: Noah Misch <n...@leadboat.com>
CommitDate: Thu Jul 24 17:30:33 2025 -0700
Sort dump objects independent of OIDs, for the 7 holdout object types.
pg_dump sorts objects by their logical names, e.g. (nspname, relname,
tgname), before dependency-driven reordering. That removes one source
of logically-identical databases differing in their schema-only dumps.
In other words, it helps with schema diffing. The logical name sort
ignored essential sort keys for constraints, operators, PUBLICATION
... FOR TABLE, PUBLICATION ... FOR TABLES IN SCHEMA, operator classes,
and operator families. pg_dump's sort then depended on object OID,
yielding spurious schema diffs. After this change, OIDs affect dump
order only in the event of catalog corruption. While pg_dump also
wrongly ignored pg_collation.collencoding, CREATE COLLATION restrictions
have been keeping that imperceptible in practical use.
Use techniques like we use for object types already having full sort key
coverage. Where the pertinent queries weren't fetching the ignored sort
keys, this adds columns to those queries and stores those keys in memory
for the long term.
The ignorance of sort keys became more problematic when commit
172259afb563d35001410dc6daad78b250924038 added a schema diff test
sensitive to it. However, dump order stability isn't a new goal, and
this might avoid other dump comparison failures. Hence, back-patch to
v13 (all supported versions).
Reviewed-by: Robert Haas <robertmh...@gmail.com>
Discussion: https://postgr.es/m/20250707192654.9e.nmi...@google.com
Backpatch-through: 13
Conflicts:
src/bin/pg_dump/pg_dump.c
src/bin/pg_dump/pg_dump.h
---
src/bin/pg_dump/common.c | 19 +++
src/bin/pg_dump/pg_dump.c | 62 ++++++--
src/bin/pg_dump/pg_dump.h | 6 +
src/bin/pg_dump/pg_dump_sort.c | 227 +++++++++++++++++++++++++++---
src/test/regress/expected/publication.out | 8 ++
src/test/regress/sql/publication.sql | 9 ++
6 files changed, 298 insertions(+), 33 deletions(-)
diff --git a/src/bin/pg_dump/common.c b/src/bin/pg_dump/common.c
index 3cabd82..7c59c91 100644
--- a/src/bin/pg_dump/common.c
+++ b/src/bin/pg_dump/common.c
@@ -17,6 +17,7 @@
#include <ctype.h>
+#include "catalog/pg_am_d.h"
#include "catalog/pg_class_d.h"
#include "fe_utils/string_utils.h"
#include "pg_backup_archiver.h"
@@ -931,6 +932,24 @@ findOprByOid(Oid oid)
}
/*
+ * findAccessMethodByOid
+ * finds the DumpableObject for the access method with the given oid
+ * returns NULL if not found
+ */
+AccessMethodInfo *
+findAccessMethodByOid(Oid oid)
+{
+ CatalogId catId;
+ DumpableObject *dobj;
+
+ catId.tableoid = AccessMethodRelationId;
+ catId.oid = oid;
+ dobj = findObjectByCatalogId(catId);
+ Assert(dobj == NULL || dobj->objType == DO_ACCESS_METHOD);
+ return (AccessMethodInfo *) dobj;
+}
+
+/*
* findCollationByOid
* finds the entry (in collinfo) of the collation with the given oid
* returns NULL if not found
diff --git a/src/bin/pg_dump/pg_dump.c b/src/bin/pg_dump/pg_dump.c
index f5a6578..a255cc6 100644
--- a/src/bin/pg_dump/pg_dump.c
+++ b/src/bin/pg_dump/pg_dump.c
@@ -1899,6 +1899,13 @@ selectDumpableProcLang(ProcLangInfo *plang, Archive
*fout)
static void
selectDumpableAccessMethod(AccessMethodInfo *method, Archive *fout)
{
+ /* see getAccessMethods() comment about v9.6. */
+ if (fout->remoteVersion < 90600)
+ {
+ method->dobj.dump = DUMP_COMPONENT_NONE;
+ return;
+ }
+
if (checkExtensionMembership(&method->dobj, fout))
return; /* extension membership
overrides all else */
@@ -5525,6 +5532,8 @@ getOperators(Archive *fout, int *numOprs)
int i_oprnamespace;
int i_rolname;
int i_oprkind;
+ int i_oprleft;
+ int i_oprright;
int i_oprcode;
/*
@@ -5536,6 +5545,8 @@ getOperators(Archive *fout, int *numOprs)
"oprnamespace, "
"(%s oprowner) AS rolname, "
"oprkind, "
+ "oprleft, "
+ "oprright, "
"oprcode::oid AS oprcode "
"FROM pg_operator",
username_subquery);
@@ -5553,6 +5564,8 @@ getOperators(Archive *fout, int *numOprs)
i_oprnamespace = PQfnumber(res, "oprnamespace");
i_rolname = PQfnumber(res, "rolname");
i_oprkind = PQfnumber(res, "oprkind");
+ i_oprleft = PQfnumber(res, "oprleft");
+ i_oprright = PQfnumber(res, "oprright");
i_oprcode = PQfnumber(res, "oprcode");
for (i = 0; i < ntups; i++)
@@ -5566,6 +5579,8 @@ getOperators(Archive *fout, int *numOprs)
findNamespace(atooid(PQgetvalue(res, i,
i_oprnamespace)));
oprinfo[i].rolname = pg_strdup(PQgetvalue(res, i, i_rolname));
oprinfo[i].oprkind = (PQgetvalue(res, i, i_oprkind))[0];
+ oprinfo[i].oprleft = atooid(PQgetvalue(res, i, i_oprleft));
+ oprinfo[i].oprright = atooid(PQgetvalue(res, i, i_oprright));
oprinfo[i].oprcode = atooid(PQgetvalue(res, i, i_oprcode));
/* Decide whether we want to dump it */
@@ -5606,6 +5621,7 @@ getCollations(Archive *fout, int *numCollations)
int i_collname;
int i_collnamespace;
int i_rolname;
+ int i_collencoding;
/* Collations didn't exist pre-9.1 */
if (fout->remoteVersion < 90100)
@@ -5623,7 +5639,8 @@ getCollations(Archive *fout, int *numCollations)
appendPQExpBuffer(query, "SELECT tableoid, oid, collname, "
"collnamespace, "
- "(%s collowner) AS rolname "
+ "(%s collowner) AS rolname, "
+ "collencoding "
"FROM pg_collation",
username_subquery);
@@ -5639,6 +5656,7 @@ getCollations(Archive *fout, int *numCollations)
i_collname = PQfnumber(res, "collname");
i_collnamespace = PQfnumber(res, "collnamespace");
i_rolname = PQfnumber(res, "rolname");
+ i_collencoding = PQfnumber(res, "collencoding");
for (i = 0; i < ntups; i++)
{
@@ -5650,6 +5668,7 @@ getCollations(Archive *fout, int *numCollations)
collinfo[i].dobj.namespace =
findNamespace(atooid(PQgetvalue(res, i,
i_collnamespace)));
collinfo[i].rolname = pg_strdup(PQgetvalue(res, i, i_rolname));
+ collinfo[i].collencoding = atoi(PQgetvalue(res, i,
i_collencoding));
/* Decide whether we want to dump it */
selectDumpableObject(&(collinfo[i].dobj), fout);
@@ -5758,19 +5777,28 @@ getAccessMethods(Archive *fout, int *numAccessMethods)
int i_amhandler;
int i_amtype;
- /* Before 9.6, there are no user-defined access methods */
- if (fout->remoteVersion < 90600)
- {
- *numAccessMethods = 0;
- return NULL;
- }
-
query = createPQExpBuffer();
- /* Select all access methods from pg_am table */
- appendPQExpBufferStr(query, "SELECT tableoid, oid, amname, amtype, "
- "amhandler::pg_catalog.regproc
AS amhandler "
- "FROM pg_am");
+ /*
+ * Select all access methods from pg_am table. v9.6 introduced CREATE
+ * ACCESS METHOD, so earlier versions usually have only built-in access
+ * methods. v9.6 also changed the access method API, replacing dozens
of
+ * pg_am columns with amhandler. Even if a user created an access
method
+ * by "INSERT INTO pg_am", we have no way to translate pre-v9.6 pg_am
+ * columns to a v9.6+ CREATE ACCESS METHOD. Hence, before v9.6, read
+ * pg_am just to facilitate findAccessMethodByOid() providing the
+ * OID-to-name mapping.
+ */
+ appendPQExpBufferStr(query, "SELECT tableoid, oid, amname, ");
+ if (fout->remoteVersion >= 90600)
+ appendPQExpBufferStr(query,
+ "amtype, "
+
"amhandler::pg_catalog.regproc AS amhandler ");
+ else
+ appendPQExpBufferStr(query,
+
"'i'::pg_catalog.\"char\" AS amtype, "
+
"'-'::pg_catalog.regproc AS amhandler ");
+ appendPQExpBufferStr(query, "FROM pg_am");
res = ExecuteSqlQuery(fout, query->data, PGRES_TUPLES_OK);
@@ -5828,6 +5856,7 @@ getOpclasses(Archive *fout, int *numOpclasses)
OpclassInfo *opcinfo;
int i_tableoid;
int i_oid;
+ int i_opcmethod;
int i_opcname;
int i_opcnamespace;
int i_rolname;
@@ -5837,7 +5866,7 @@ getOpclasses(Archive *fout, int *numOpclasses)
* system-defined opclasses at dump-out time.
*/
- appendPQExpBuffer(query, "SELECT tableoid, oid, opcname, "
+ appendPQExpBuffer(query, "SELECT tableoid, oid, opcmethod, opcname, "
"opcnamespace, "
"(%s opcowner) AS rolname "
"FROM pg_opclass",
@@ -5852,6 +5881,7 @@ getOpclasses(Archive *fout, int *numOpclasses)
i_tableoid = PQfnumber(res, "tableoid");
i_oid = PQfnumber(res, "oid");
+ i_opcmethod = PQfnumber(res, "opcmethod");
i_opcname = PQfnumber(res, "opcname");
i_opcnamespace = PQfnumber(res, "opcnamespace");
i_rolname = PQfnumber(res, "rolname");
@@ -5865,6 +5895,7 @@ getOpclasses(Archive *fout, int *numOpclasses)
opcinfo[i].dobj.name = pg_strdup(PQgetvalue(res, i, i_opcname));
opcinfo[i].dobj.namespace =
findNamespace(atooid(PQgetvalue(res, i,
i_opcnamespace)));
+ opcinfo[i].opcmethod = atooid(PQgetvalue(res, i, i_opcmethod));
opcinfo[i].rolname = pg_strdup(PQgetvalue(res, i, i_rolname));
/* Decide whether we want to dump it */
@@ -5902,6 +5933,7 @@ getOpfamilies(Archive *fout, int *numOpfamilies)
OpfamilyInfo *opfinfo;
int i_tableoid;
int i_oid;
+ int i_opfmethod;
int i_opfname;
int i_opfnamespace;
int i_rolname;
@@ -5920,7 +5952,7 @@ getOpfamilies(Archive *fout, int *numOpfamilies)
* system-defined opfamilies at dump-out time.
*/
- appendPQExpBuffer(query, "SELECT tableoid, oid, opfname, "
+ appendPQExpBuffer(query, "SELECT tableoid, oid, opfmethod, opfname, "
"opfnamespace, "
"(%s opfowner) AS rolname "
"FROM pg_opfamily",
@@ -5936,6 +5968,7 @@ getOpfamilies(Archive *fout, int *numOpfamilies)
i_tableoid = PQfnumber(res, "tableoid");
i_oid = PQfnumber(res, "oid");
i_opfname = PQfnumber(res, "opfname");
+ i_opfmethod = PQfnumber(res, "opfmethod");
i_opfnamespace = PQfnumber(res, "opfnamespace");
i_rolname = PQfnumber(res, "rolname");
@@ -5948,6 +5981,7 @@ getOpfamilies(Archive *fout, int *numOpfamilies)
opfinfo[i].dobj.name = pg_strdup(PQgetvalue(res, i, i_opfname));
opfinfo[i].dobj.namespace =
findNamespace(atooid(PQgetvalue(res, i,
i_opfnamespace)));
+ opfinfo[i].opfmethod = atooid(PQgetvalue(res, i, i_opfmethod));
opfinfo[i].rolname = pg_strdup(PQgetvalue(res, i, i_rolname));
/* Decide whether we want to dump it */
diff --git a/src/bin/pg_dump/pg_dump.h b/src/bin/pg_dump/pg_dump.h
index 832aa86..bcd7841 100644
--- a/src/bin/pg_dump/pg_dump.h
+++ b/src/bin/pg_dump/pg_dump.h
@@ -222,6 +222,8 @@ typedef struct _oprInfo
DumpableObject dobj;
char *rolname;
char oprkind;
+ Oid oprleft;
+ Oid oprright;
Oid oprcode;
} OprInfo;
@@ -235,12 +237,14 @@ typedef struct _accessMethodInfo
typedef struct _opclassInfo
{
DumpableObject dobj;
+ Oid opcmethod;
char *rolname;
} OpclassInfo;
typedef struct _opfamilyInfo
{
DumpableObject dobj;
+ Oid opfmethod;
char *rolname;
} OpfamilyInfo;
@@ -248,6 +252,7 @@ typedef struct _collInfo
{
DumpableObject dobj;
char *rolname;
+ int collencoding;
} CollInfo;
typedef struct _convInfo
@@ -675,6 +680,7 @@ extern TableInfo *findTableByOid(Oid oid);
extern TypeInfo *findTypeByOid(Oid oid);
extern FuncInfo *findFuncByOid(Oid oid);
extern OprInfo *findOprByOid(Oid oid);
+extern AccessMethodInfo *findAccessMethodByOid(Oid oid);
extern CollInfo *findCollationByOid(Oid oid);
extern NamespaceInfo *findNamespaceByOid(Oid oid);
extern ExtensionInfo *findExtensionByOid(Oid oid);
diff --git a/src/bin/pg_dump/pg_dump_sort.c b/src/bin/pg_dump/pg_dump_sort.c
index 039b3be..b6d7337 100644
--- a/src/bin/pg_dump/pg_dump_sort.c
+++ b/src/bin/pg_dump/pg_dump_sort.c
@@ -155,6 +155,8 @@ static DumpId postDataBoundId;
static int DOTypeNameCompare(const void *p1, const void *p2);
+static int pgTypeNameCompare(Oid typid1, Oid typid2);
+static int accessMethodNameCompare(Oid am1, Oid am2);
static bool TopoSort(DumpableObject **objs,
int numObjs,
DumpableObject **ordering,
@@ -222,12 +224,39 @@ DOTypeNameCompare(const void *p1, const void *p2)
else if (obj2->namespace)
return 1;
- /* Sort by name */
+ /*
+ * Sort by name. With a few exceptions, names here are single catalog
+ * columns. To get a fuller picture, grep pg_dump.c for "dobj.name = ".
+ * Names here don't match "Name:" in plain format output, which is a
+ * _tocEntry.tag. For example, DumpableObject.name of a constraint is
+ * pg_constraint.conname, but _tocEntry.tag of a constraint is relname
and
+ * conname joined with a space.
+ */
cmpval = strcmp(obj1->name, obj2->name);
if (cmpval != 0)
return cmpval;
- /* To have a stable sort order, break ties for some object types */
+ /*
+ * Sort by type. This helps types that share a type priority without
+ * sharing a unique name constraint, e.g. opclass and opfamily.
+ */
+ cmpval = obj1->objType - obj2->objType;
+ if (cmpval != 0)
+ return cmpval;
+
+ /*
+ * To have a stable sort order, break ties for some object types. Most
+ * catalogs have a natural key, e.g. pg_proc_proname_args_nsp_index.
Where
+ * the above "namespace" and "name" comparisons don't cover all natural
+ * key columns, compare the rest here.
+ *
+ * The natural key usually refers to other catalogs by surrogate keys.
+ * Hence, this translates each of those references to the natural key of
+ * the referenced catalog. That may descend through multiple levels of
+ * catalog references. For example, to sort by pg_proc.proargtypes,
+ * descend to each pg_type and then further to its pg_namespace, for an
+ * overall sort by (nspname, typname).
+ */
if (obj1->objType == DO_FUNC || obj1->objType == DO_AGG)
{
FuncInfo *fobj1 = *(FuncInfo *const *) p1;
@@ -240,22 +269,10 @@ DOTypeNameCompare(const void *p1, const void *p2)
return cmpval;
for (i = 0; i < fobj1->nargs; i++)
{
- TypeInfo *argtype1 =
findTypeByOid(fobj1->argtypes[i]);
- TypeInfo *argtype2 =
findTypeByOid(fobj2->argtypes[i]);
-
- if (argtype1 && argtype2)
- {
- if (argtype1->dobj.namespace &&
argtype2->dobj.namespace)
- {
- cmpval =
strcmp(argtype1->dobj.namespace->dobj.name,
-
argtype2->dobj.namespace->dobj.name);
- if (cmpval != 0)
- return cmpval;
- }
- cmpval = strcmp(argtype1->dobj.name,
argtype2->dobj.name);
- if (cmpval != 0)
- return cmpval;
- }
+ cmpval = pgTypeNameCompare(fobj1->argtypes[i],
+
fobj2->argtypes[i]);
+ if (cmpval != 0)
+ return cmpval;
}
}
else if (obj1->objType == DO_OPERATOR)
@@ -267,6 +284,57 @@ DOTypeNameCompare(const void *p1, const void *p2)
cmpval = (oobj2->oprkind - oobj1->oprkind);
if (cmpval != 0)
return cmpval;
+ /* Within an oprkind, sort by argument type names */
+ cmpval = pgTypeNameCompare(oobj1->oprleft, oobj2->oprleft);
+ if (cmpval != 0)
+ return cmpval;
+ cmpval = pgTypeNameCompare(oobj1->oprright, oobj2->oprright);
+ if (cmpval != 0)
+ return cmpval;
+ }
+ else if (obj1->objType == DO_OPCLASS)
+ {
+ OpclassInfo *opcobj1 = *(OpclassInfo *const *) p1;
+ OpclassInfo *opcobj2 = *(OpclassInfo *const *) p2;
+
+ /* Sort by access method name, per pg_opclass_am_name_nsp_index
*/
+ cmpval = accessMethodNameCompare(opcobj1->opcmethod,
+
opcobj2->opcmethod);
+ if (cmpval != 0)
+ return cmpval;
+ }
+ else if (obj1->objType == DO_OPFAMILY)
+ {
+ OpfamilyInfo *opfobj1 = *(OpfamilyInfo *const *) p1;
+ OpfamilyInfo *opfobj2 = *(OpfamilyInfo *const *) p2;
+
+ /* Sort by access method name, per
pg_opfamily_am_name_nsp_index */
+ cmpval = accessMethodNameCompare(opfobj1->opfmethod,
+
opfobj2->opfmethod);
+ if (cmpval != 0)
+ return cmpval;
+ }
+ else if (obj1->objType == DO_COLLATION)
+ {
+ CollInfo *cobj1 = *(CollInfo *const *) p1;
+ CollInfo *cobj2 = *(CollInfo *const *) p2;
+
+ /*
+ * Sort by encoding, per pg_collation_name_enc_nsp_index.
Technically,
+ * this is not necessary, because wherever this changes dump
order,
+ * restoring the dump fails anyway. CREATE COLLATION can't
create a
+ * tie for this to break, because it imposes restrictions to
make
+ * (nspname, collname) uniquely identify a collation within a
given
+ * DatabaseEncoding. While pg_import_system_collations() can
create a
+ * tie, pg_dump+restore fails after
+ * pg_import_system_collations('my_schema') does so. However,
there's
+ * little to gain by ignoring one natural key column on the
basis of
+ * those limitations elsewhere, so respect the full natural key
like
+ * we do for other object types.
+ */
+ cmpval = cobj1->collencoding - cobj2->collencoding;
+ if (cmpval != 0)
+ return cmpval;
}
else if (obj1->objType == DO_ATTRDEF)
{
@@ -311,11 +379,132 @@ DOTypeNameCompare(const void *p1, const void *p2)
if (cmpval != 0)
return cmpval;
}
+ else if (obj1->objType == DO_CONSTRAINT)
+ {
+ ConstraintInfo *robj1 = *(ConstraintInfo *const *) p1;
+ ConstraintInfo *robj2 = *(ConstraintInfo *const *) p2;
+
+ /*
+ * Sort domain constraints before table constraints, for
consistency
+ * with our decision to sort CREATE DOMAIN before CREATE TABLE.
+ */
+ if (robj1->condomain)
+ {
+ if (robj2->condomain)
+ {
+ /* Sort by domain name (domain namespace was
considered) */
+ cmpval = strcmp(robj1->condomain->dobj.name,
+
robj2->condomain->dobj.name);
+ if (cmpval != 0)
+ return cmpval;
+ }
+ else
+ return PRIO_TYPE - PRIO_TABLE;
+ }
+ else if (robj2->condomain)
+ return PRIO_TABLE - PRIO_TYPE;
+ else
+ {
+ /* Sort by table name (table namespace was considered
already) */
+ cmpval = strcmp(robj1->contable->dobj.name,
+
robj2->contable->dobj.name);
+ if (cmpval != 0)
+ return cmpval;
+ }
+ }
+ else if (obj1->objType == DO_PUBLICATION_REL)
+ {
+ PublicationRelInfo *probj1 = *(PublicationRelInfo *const *) p1;
+ PublicationRelInfo *probj2 = *(PublicationRelInfo *const *) p2;
+
+ /* Sort by publication name, since (namespace, name) match the
rel */
+ cmpval = strcmp(probj1->publication->dobj.name,
+ probj2->publication->dobj.name);
+ if (cmpval != 0)
+ return cmpval;
+ }
- /* Usually shouldn't get here, but if we do, sort by OID */
+ /*
+ * Shouldn't get here except after catalog corruption, but if we do,
sort
+ * by OID. This may make logically-identical databases differ in the
+ * order of objects in dump output. Users will get spurious schema
diffs.
+ * Expect flaky failures of 002_pg_upgrade.pl test 'dump outputs from
+ * original and restored regression databases match' if the regression
+ * database contains objects allowing that test to reach here. That's a
+ * consequence of the test using "pg_restore -j", which doesn't fully
+ * constrain OID assignment order.
+ */
+ Assert(false);
return oidcmp(obj1->catId.oid, obj2->catId.oid);
}
+/* Compare two OID-identified pg_type values by nspname, then by typname. */
+static int
+pgTypeNameCompare(Oid typid1, Oid typid2)
+{
+ TypeInfo *typobj1;
+ TypeInfo *typobj2;
+ int cmpval;
+
+ if (typid1 == typid2)
+ return 0;
+
+ typobj1 = findTypeByOid(typid1);
+ typobj2 = findTypeByOid(typid2);
+
+ if (!typobj1 || !typobj2)
+ {
+ /*
+ * getTypes() didn't find some OID. Assume catalog corruption,
e.g.
+ * an oprright value without the corresponding OID in a pg_type
row.
+ * Report as "equal", so the caller uses the next available
basis for
+ * comparison, e.g. the next function argument.
+ *
+ * Unary operators have InvalidOid in oprleft (if oprkind='r')
or in
+ * oprright (if oprkind='l'). Caller already sorted by oprkind,
+ * calling us only for like-kind operators. Hence, "typid1 ==
typid2"
+ * took care of InvalidOid. (v14 removed postfix operator
support.
+ * Hence, when dumping from v14+, only oprleft can be
InvalidOid.)
+ */
+ Assert(false);
+ return 0;
+ }
+
+ if (!typobj1->dobj.namespace || !typobj2->dobj.namespace)
+ Assert(false); /* catalog corruption */
+ else
+ {
+ cmpval = strcmp(typobj1->dobj.namespace->dobj.name,
+
typobj2->dobj.namespace->dobj.name);
+ if (cmpval != 0)
+ return cmpval;
+ }
+ return strcmp(typobj1->dobj.name, typobj2->dobj.name);
+}
+
+/* Compare two OID-identified pg_am values by amname. */
+static int
+accessMethodNameCompare(Oid am1, Oid am2)
+{
+ AccessMethodInfo *amobj1;
+ AccessMethodInfo *amobj2;
+
+ if (am1 == am2)
+ return 0;
+
+ amobj1 = findAccessMethodByOid(am1);
+ amobj2 = findAccessMethodByOid(am2);
+
+ if (!amobj1 || !amobj2)
+ {
+ /* catalog corruption: handle like pgTypeNameCompare() does */
+ Assert(false);
+ return 0;
+ }
+
+ return strcmp(amobj1->dobj.name, amobj2->dobj.name);
+}
+
/*
* Sort the given objects into a safe dump order using dependency
diff --git a/src/test/regress/expected/publication.out
b/src/test/regress/expected/publication.out
index b7ce080..35595fb 100644
--- a/src/test/regress/expected/publication.out
+++ b/src/test/regress/expected/publication.out
@@ -344,3 +344,11 @@ NOTICE: drop cascades to table pub_test.testpub_nopk
RESET SESSION AUTHORIZATION;
DROP ROLE regress_publication_user, regress_publication_user2;
DROP ROLE regress_publication_user_dummy;
+-- stage objects for pg_dump tests
+CREATE SCHEMA pubme CREATE TABLE t0 (c int, d int) CREATE TABLE t1 (c int);
+SET client_min_messages = 'ERROR';
+CREATE PUBLICATION dump_pub_1ct FOR TABLE ONLY pubme.t0;
+CREATE PUBLICATION dump_pub_2ct FOR TABLE ONLY pubme.t0, pubme.t1;
+CREATE PUBLICATION dump_pub_all FOR TABLE ONLY pubme.t0, pubme.t1
+ WITH (publish_via_partition_root = true);
+RESET client_min_messages;
diff --git a/src/test/regress/sql/publication.sql
b/src/test/regress/sql/publication.sql
index 7d5c937..021d4e2 100644
--- a/src/test/regress/sql/publication.sql
+++ b/src/test/regress/sql/publication.sql
@@ -218,3 +218,12 @@ DROP SCHEMA pub_test CASCADE;
RESET SESSION AUTHORIZATION;
DROP ROLE regress_publication_user, regress_publication_user2;
DROP ROLE regress_publication_user_dummy;
+
+-- stage objects for pg_dump tests
+CREATE SCHEMA pubme CREATE TABLE t0 (c int, d int) CREATE TABLE t1 (c int);
+SET client_min_messages = 'ERROR';
+CREATE PUBLICATION dump_pub_1ct FOR TABLE ONLY pubme.t0;
+CREATE PUBLICATION dump_pub_2ct FOR TABLE ONLY pubme.t0, pubme.t1;
+CREATE PUBLICATION dump_pub_all FOR TABLE ONLY pubme.t0, pubme.t1
+ WITH (publish_via_partition_root = true);
+RESET client_min_messages;
commit 5aadf5d (HEAD, zzy_test-commit-REL_13_STABLE)
Author: Noah Misch <n...@leadboat.com>
AuthorDate: Thu Jul 24 17:32:07 2025 -0700
Commit: Noah Misch <n...@leadboat.com>
CommitDate: Thu Jul 24 17:41:00 2025 -0700
Sort dump objects independent of OIDs, for the 7 holdout object types.
pg_dump sorts objects by their logical names, e.g. (nspname, relname,
tgname), before dependency-driven reordering. That removes one source
of logically-identical databases differing in their schema-only dumps.
In other words, it helps with schema diffing. The logical name sort
ignored essential sort keys for constraints, operators, PUBLICATION
... FOR TABLE, PUBLICATION ... FOR TABLES IN SCHEMA, operator classes,
and operator families. pg_dump's sort then depended on object OID,
yielding spurious schema diffs. After this change, OIDs affect dump
order only in the event of catalog corruption. While pg_dump also
wrongly ignored pg_collation.collencoding, CREATE COLLATION restrictions
have been keeping that imperceptible in practical use.
Use techniques like we use for object types already having full sort key
coverage. Where the pertinent queries weren't fetching the ignored sort
keys, this adds columns to those queries and stores those keys in memory
for the long term.
The ignorance of sort keys became more problematic when commit
172259afb563d35001410dc6daad78b250924038 added a schema diff test
sensitive to it. However, dump order stability isn't a new goal, and
this might avoid other dump comparison failures. Hence, back-patch to
v13 (all supported versions).
Reviewed-by: Robert Haas <robertmh...@gmail.com>
Discussion: https://postgr.es/m/20250707192654.9e.nmi...@google.com
Backpatch-through: 13
Conflicts:
src/bin/pg_dump/pg_dump.c
---
src/bin/pg_dump/common.c | 19 +++
src/bin/pg_dump/pg_dump.c | 62 ++++++--
src/bin/pg_dump/pg_dump.h | 6 +
src/bin/pg_dump/pg_dump_sort.c | 227 +++++++++++++++++++++++++++---
src/test/regress/expected/publication.out | 8 ++
src/test/regress/sql/publication.sql | 9 ++
6 files changed, 298 insertions(+), 33 deletions(-)
diff --git a/src/bin/pg_dump/common.c b/src/bin/pg_dump/common.c
index c68b86b..e4a5812 100644
--- a/src/bin/pg_dump/common.c
+++ b/src/bin/pg_dump/common.c
@@ -17,6 +17,7 @@
#include <ctype.h>
+#include "catalog/pg_am_d.h"
#include "catalog/pg_class_d.h"
#include "fe_utils/string_utils.h"
#include "pg_backup_archiver.h"
@@ -891,6 +892,24 @@ findOprByOid(Oid oid)
}
/*
+ * findAccessMethodByOid
+ * finds the DumpableObject for the access method with the given oid
+ * returns NULL if not found
+ */
+AccessMethodInfo *
+findAccessMethodByOid(Oid oid)
+{
+ CatalogId catId;
+ DumpableObject *dobj;
+
+ catId.tableoid = AccessMethodRelationId;
+ catId.oid = oid;
+ dobj = findObjectByCatalogId(catId);
+ Assert(dobj == NULL || dobj->objType == DO_ACCESS_METHOD);
+ return (AccessMethodInfo *) dobj;
+}
+
+/*
* findCollationByOid
* finds the entry (in collinfo) of the collation with the given oid
* returns NULL if not found
diff --git a/src/bin/pg_dump/pg_dump.c b/src/bin/pg_dump/pg_dump.c
index 4b13669..6745520 100644
--- a/src/bin/pg_dump/pg_dump.c
+++ b/src/bin/pg_dump/pg_dump.c
@@ -1767,6 +1767,13 @@ selectDumpableProcLang(ProcLangInfo *plang, Archive
*fout)
static void
selectDumpableAccessMethod(AccessMethodInfo *method, Archive *fout)
{
+ /* see getAccessMethods() comment about v9.6. */
+ if (fout->remoteVersion < 90600)
+ {
+ method->dobj.dump = DUMP_COMPONENT_NONE;
+ return;
+ }
+
if (checkExtensionMembership(&method->dobj, fout))
return; /* extension membership
overrides all else */
@@ -5325,6 +5332,8 @@ getOperators(Archive *fout, int *numOprs)
int i_oprnamespace;
int i_rolname;
int i_oprkind;
+ int i_oprleft;
+ int i_oprright;
int i_oprcode;
/*
@@ -5336,6 +5345,8 @@ getOperators(Archive *fout, int *numOprs)
"oprnamespace, "
"(%s oprowner) AS rolname, "
"oprkind, "
+ "oprleft, "
+ "oprright, "
"oprcode::oid AS oprcode "
"FROM pg_operator",
username_subquery);
@@ -5353,6 +5364,8 @@ getOperators(Archive *fout, int *numOprs)
i_oprnamespace = PQfnumber(res, "oprnamespace");
i_rolname = PQfnumber(res, "rolname");
i_oprkind = PQfnumber(res, "oprkind");
+ i_oprleft = PQfnumber(res, "oprleft");
+ i_oprright = PQfnumber(res, "oprright");
i_oprcode = PQfnumber(res, "oprcode");
for (i = 0; i < ntups; i++)
@@ -5367,6 +5380,8 @@ getOperators(Archive *fout, int *numOprs)
atooid(PQgetvalue(res, i,
i_oprnamespace)));
oprinfo[i].rolname = pg_strdup(PQgetvalue(res, i, i_rolname));
oprinfo[i].oprkind = (PQgetvalue(res, i, i_oprkind))[0];
+ oprinfo[i].oprleft = atooid(PQgetvalue(res, i, i_oprleft));
+ oprinfo[i].oprright = atooid(PQgetvalue(res, i, i_oprright));
oprinfo[i].oprcode = atooid(PQgetvalue(res, i, i_oprcode));
/* Decide whether we want to dump it */
@@ -5407,6 +5422,7 @@ getCollations(Archive *fout, int *numCollations)
int i_collname;
int i_collnamespace;
int i_rolname;
+ int i_collencoding;
/* Collations didn't exist pre-9.1 */
if (fout->remoteVersion < 90100)
@@ -5424,7 +5440,8 @@ getCollations(Archive *fout, int *numCollations)
appendPQExpBuffer(query, "SELECT tableoid, oid, collname, "
"collnamespace, "
- "(%s collowner) AS rolname "
+ "(%s collowner) AS rolname, "
+ "collencoding "
"FROM pg_collation",
username_subquery);
@@ -5440,6 +5457,7 @@ getCollations(Archive *fout, int *numCollations)
i_collname = PQfnumber(res, "collname");
i_collnamespace = PQfnumber(res, "collnamespace");
i_rolname = PQfnumber(res, "rolname");
+ i_collencoding = PQfnumber(res, "collencoding");
for (i = 0; i < ntups; i++)
{
@@ -5452,6 +5470,7 @@ getCollations(Archive *fout, int *numCollations)
findNamespace(fout,
atooid(PQgetvalue(res, i,
i_collnamespace)));
collinfo[i].rolname = pg_strdup(PQgetvalue(res, i, i_rolname));
+ collinfo[i].collencoding = atoi(PQgetvalue(res, i,
i_collencoding));
/* Decide whether we want to dump it */
selectDumpableObject(&(collinfo[i].dobj), fout);
@@ -5561,19 +5580,28 @@ getAccessMethods(Archive *fout, int *numAccessMethods)
int i_amhandler;
int i_amtype;
- /* Before 9.6, there are no user-defined access methods */
- if (fout->remoteVersion < 90600)
- {
- *numAccessMethods = 0;
- return NULL;
- }
-
query = createPQExpBuffer();
- /* Select all access methods from pg_am table */
- appendPQExpBufferStr(query, "SELECT tableoid, oid, amname, amtype, "
- "amhandler::pg_catalog.regproc
AS amhandler "
- "FROM pg_am");
+ /*
+ * Select all access methods from pg_am table. v9.6 introduced CREATE
+ * ACCESS METHOD, so earlier versions usually have only built-in access
+ * methods. v9.6 also changed the access method API, replacing dozens
of
+ * pg_am columns with amhandler. Even if a user created an access
method
+ * by "INSERT INTO pg_am", we have no way to translate pre-v9.6 pg_am
+ * columns to a v9.6+ CREATE ACCESS METHOD. Hence, before v9.6, read
+ * pg_am just to facilitate findAccessMethodByOid() providing the
+ * OID-to-name mapping.
+ */
+ appendPQExpBufferStr(query, "SELECT tableoid, oid, amname, ");
+ if (fout->remoteVersion >= 90600)
+ appendPQExpBufferStr(query,
+ "amtype, "
+
"amhandler::pg_catalog.regproc AS amhandler ");
+ else
+ appendPQExpBufferStr(query,
+
"'i'::pg_catalog.\"char\" AS amtype, "
+
"'-'::pg_catalog.regproc AS amhandler ");
+ appendPQExpBufferStr(query, "FROM pg_am");
res = ExecuteSqlQuery(fout, query->data, PGRES_TUPLES_OK);
@@ -5631,6 +5659,7 @@ getOpclasses(Archive *fout, int *numOpclasses)
OpclassInfo *opcinfo;
int i_tableoid;
int i_oid;
+ int i_opcmethod;
int i_opcname;
int i_opcnamespace;
int i_rolname;
@@ -5640,7 +5669,7 @@ getOpclasses(Archive *fout, int *numOpclasses)
* system-defined opclasses at dump-out time.
*/
- appendPQExpBuffer(query, "SELECT tableoid, oid, opcname, "
+ appendPQExpBuffer(query, "SELECT tableoid, oid, opcmethod, opcname, "
"opcnamespace, "
"(%s opcowner) AS rolname "
"FROM pg_opclass",
@@ -5655,6 +5684,7 @@ getOpclasses(Archive *fout, int *numOpclasses)
i_tableoid = PQfnumber(res, "tableoid");
i_oid = PQfnumber(res, "oid");
+ i_opcmethod = PQfnumber(res, "opcmethod");
i_opcname = PQfnumber(res, "opcname");
i_opcnamespace = PQfnumber(res, "opcnamespace");
i_rolname = PQfnumber(res, "rolname");
@@ -5669,6 +5699,7 @@ getOpclasses(Archive *fout, int *numOpclasses)
opcinfo[i].dobj.namespace =
findNamespace(fout,
atooid(PQgetvalue(res, i,
i_opcnamespace)));
+ opcinfo[i].opcmethod = atooid(PQgetvalue(res, i, i_opcmethod));
opcinfo[i].rolname = pg_strdup(PQgetvalue(res, i, i_rolname));
/* Decide whether we want to dump it */
@@ -5706,6 +5737,7 @@ getOpfamilies(Archive *fout, int *numOpfamilies)
OpfamilyInfo *opfinfo;
int i_tableoid;
int i_oid;
+ int i_opfmethod;
int i_opfname;
int i_opfnamespace;
int i_rolname;
@@ -5724,7 +5756,7 @@ getOpfamilies(Archive *fout, int *numOpfamilies)
* system-defined opfamilies at dump-out time.
*/
- appendPQExpBuffer(query, "SELECT tableoid, oid, opfname, "
+ appendPQExpBuffer(query, "SELECT tableoid, oid, opfmethod, opfname, "
"opfnamespace, "
"(%s opfowner) AS rolname "
"FROM pg_opfamily",
@@ -5740,6 +5772,7 @@ getOpfamilies(Archive *fout, int *numOpfamilies)
i_tableoid = PQfnumber(res, "tableoid");
i_oid = PQfnumber(res, "oid");
i_opfname = PQfnumber(res, "opfname");
+ i_opfmethod = PQfnumber(res, "opfmethod");
i_opfnamespace = PQfnumber(res, "opfnamespace");
i_rolname = PQfnumber(res, "rolname");
@@ -5753,6 +5786,7 @@ getOpfamilies(Archive *fout, int *numOpfamilies)
opfinfo[i].dobj.namespace =
findNamespace(fout,
atooid(PQgetvalue(res, i,
i_opfnamespace)));
+ opfinfo[i].opfmethod = atooid(PQgetvalue(res, i, i_opfmethod));
opfinfo[i].rolname = pg_strdup(PQgetvalue(res, i, i_rolname));
/* Decide whether we want to dump it */
diff --git a/src/bin/pg_dump/pg_dump.h b/src/bin/pg_dump/pg_dump.h
index 552e9e1..7fcc544 100644
--- a/src/bin/pg_dump/pg_dump.h
+++ b/src/bin/pg_dump/pg_dump.h
@@ -219,6 +219,8 @@ typedef struct _oprInfo
DumpableObject dobj;
char *rolname;
char oprkind;
+ Oid oprleft;
+ Oid oprright;
Oid oprcode;
} OprInfo;
@@ -232,12 +234,14 @@ typedef struct _accessMethodInfo
typedef struct _opclassInfo
{
DumpableObject dobj;
+ Oid opcmethod;
char *rolname;
} OpclassInfo;
typedef struct _opfamilyInfo
{
DumpableObject dobj;
+ Oid opfmethod;
char *rolname;
} OpfamilyInfo;
@@ -245,6 +249,7 @@ typedef struct _collInfo
{
DumpableObject dobj;
char *rolname;
+ int collencoding;
} CollInfo;
typedef struct _convInfo
@@ -662,6 +667,7 @@ extern TableInfo *findTableByOid(Oid oid);
extern TypeInfo *findTypeByOid(Oid oid);
extern FuncInfo *findFuncByOid(Oid oid);
extern OprInfo *findOprByOid(Oid oid);
+extern AccessMethodInfo *findAccessMethodByOid(Oid oid);
extern CollInfo *findCollationByOid(Oid oid);
extern NamespaceInfo *findNamespaceByOid(Oid oid);
extern ExtensionInfo *findExtensionByOid(Oid oid);
diff --git a/src/bin/pg_dump/pg_dump_sort.c b/src/bin/pg_dump/pg_dump_sort.c
index 51e6be9..892eda7 100644
--- a/src/bin/pg_dump/pg_dump_sort.c
+++ b/src/bin/pg_dump/pg_dump_sort.c
@@ -104,6 +104,8 @@ static DumpId postDataBoundId;
static int DOTypeNameCompare(const void *p1, const void *p2);
+static int pgTypeNameCompare(Oid typid1, Oid typid2);
+static int accessMethodNameCompare(Oid am1, Oid am2);
static bool TopoSort(DumpableObject **objs,
int numObjs,
DumpableObject **ordering,
@@ -171,12 +173,39 @@ DOTypeNameCompare(const void *p1, const void *p2)
else if (obj2->namespace)
return 1;
- /* Sort by name */
+ /*
+ * Sort by name. With a few exceptions, names here are single catalog
+ * columns. To get a fuller picture, grep pg_dump.c for "dobj.name = ".
+ * Names here don't match "Name:" in plain format output, which is a
+ * _tocEntry.tag. For example, DumpableObject.name of a constraint is
+ * pg_constraint.conname, but _tocEntry.tag of a constraint is relname
and
+ * conname joined with a space.
+ */
cmpval = strcmp(obj1->name, obj2->name);
if (cmpval != 0)
return cmpval;
- /* To have a stable sort order, break ties for some object types */
+ /*
+ * Sort by type. This helps types that share a type priority without
+ * sharing a unique name constraint, e.g. opclass and opfamily.
+ */
+ cmpval = obj1->objType - obj2->objType;
+ if (cmpval != 0)
+ return cmpval;
+
+ /*
+ * To have a stable sort order, break ties for some object types. Most
+ * catalogs have a natural key, e.g. pg_proc_proname_args_nsp_index.
Where
+ * the above "namespace" and "name" comparisons don't cover all natural
+ * key columns, compare the rest here.
+ *
+ * The natural key usually refers to other catalogs by surrogate keys.
+ * Hence, this translates each of those references to the natural key of
+ * the referenced catalog. That may descend through multiple levels of
+ * catalog references. For example, to sort by pg_proc.proargtypes,
+ * descend to each pg_type and then further to its pg_namespace, for an
+ * overall sort by (nspname, typname).
+ */
if (obj1->objType == DO_FUNC || obj1->objType == DO_AGG)
{
FuncInfo *fobj1 = *(FuncInfo *const *) p1;
@@ -189,22 +218,10 @@ DOTypeNameCompare(const void *p1, const void *p2)
return cmpval;
for (i = 0; i < fobj1->nargs; i++)
{
- TypeInfo *argtype1 =
findTypeByOid(fobj1->argtypes[i]);
- TypeInfo *argtype2 =
findTypeByOid(fobj2->argtypes[i]);
-
- if (argtype1 && argtype2)
- {
- if (argtype1->dobj.namespace &&
argtype2->dobj.namespace)
- {
- cmpval =
strcmp(argtype1->dobj.namespace->dobj.name,
-
argtype2->dobj.namespace->dobj.name);
- if (cmpval != 0)
- return cmpval;
- }
- cmpval = strcmp(argtype1->dobj.name,
argtype2->dobj.name);
- if (cmpval != 0)
- return cmpval;
- }
+ cmpval = pgTypeNameCompare(fobj1->argtypes[i],
+
fobj2->argtypes[i]);
+ if (cmpval != 0)
+ return cmpval;
}
}
else if (obj1->objType == DO_OPERATOR)
@@ -216,6 +233,57 @@ DOTypeNameCompare(const void *p1, const void *p2)
cmpval = (oobj2->oprkind - oobj1->oprkind);
if (cmpval != 0)
return cmpval;
+ /* Within an oprkind, sort by argument type names */
+ cmpval = pgTypeNameCompare(oobj1->oprleft, oobj2->oprleft);
+ if (cmpval != 0)
+ return cmpval;
+ cmpval = pgTypeNameCompare(oobj1->oprright, oobj2->oprright);
+ if (cmpval != 0)
+ return cmpval;
+ }
+ else if (obj1->objType == DO_OPCLASS)
+ {
+ OpclassInfo *opcobj1 = *(OpclassInfo *const *) p1;
+ OpclassInfo *opcobj2 = *(OpclassInfo *const *) p2;
+
+ /* Sort by access method name, per pg_opclass_am_name_nsp_index
*/
+ cmpval = accessMethodNameCompare(opcobj1->opcmethod,
+
opcobj2->opcmethod);
+ if (cmpval != 0)
+ return cmpval;
+ }
+ else if (obj1->objType == DO_OPFAMILY)
+ {
+ OpfamilyInfo *opfobj1 = *(OpfamilyInfo *const *) p1;
+ OpfamilyInfo *opfobj2 = *(OpfamilyInfo *const *) p2;
+
+ /* Sort by access method name, per
pg_opfamily_am_name_nsp_index */
+ cmpval = accessMethodNameCompare(opfobj1->opfmethod,
+
opfobj2->opfmethod);
+ if (cmpval != 0)
+ return cmpval;
+ }
+ else if (obj1->objType == DO_COLLATION)
+ {
+ CollInfo *cobj1 = *(CollInfo *const *) p1;
+ CollInfo *cobj2 = *(CollInfo *const *) p2;
+
+ /*
+ * Sort by encoding, per pg_collation_name_enc_nsp_index.
Technically,
+ * this is not necessary, because wherever this changes dump
order,
+ * restoring the dump fails anyway. CREATE COLLATION can't
create a
+ * tie for this to break, because it imposes restrictions to
make
+ * (nspname, collname) uniquely identify a collation within a
given
+ * DatabaseEncoding. While pg_import_system_collations() can
create a
+ * tie, pg_dump+restore fails after
+ * pg_import_system_collations('my_schema') does so. However,
there's
+ * little to gain by ignoring one natural key column on the
basis of
+ * those limitations elsewhere, so respect the full natural key
like
+ * we do for other object types.
+ */
+ cmpval = cobj1->collencoding - cobj2->collencoding;
+ if (cmpval != 0)
+ return cmpval;
}
else if (obj1->objType == DO_ATTRDEF)
{
@@ -260,11 +328,132 @@ DOTypeNameCompare(const void *p1, const void *p2)
if (cmpval != 0)
return cmpval;
}
+ else if (obj1->objType == DO_CONSTRAINT)
+ {
+ ConstraintInfo *robj1 = *(ConstraintInfo *const *) p1;
+ ConstraintInfo *robj2 = *(ConstraintInfo *const *) p2;
+
+ /*
+ * Sort domain constraints before table constraints, for
consistency
+ * with our decision to sort CREATE DOMAIN before CREATE TABLE.
+ */
+ if (robj1->condomain)
+ {
+ if (robj2->condomain)
+ {
+ /* Sort by domain name (domain namespace was
considered) */
+ cmpval = strcmp(robj1->condomain->dobj.name,
+
robj2->condomain->dobj.name);
+ if (cmpval != 0)
+ return cmpval;
+ }
+ else
+ return dbObjectTypePriority[DO_TYPE] -
dbObjectTypePriority[DO_TABLE];
+ }
+ else if (robj2->condomain)
+ return dbObjectTypePriority[DO_TABLE] -
dbObjectTypePriority[DO_TYPE];
+ else
+ {
+ /* Sort by table name (table namespace was considered
already) */
+ cmpval = strcmp(robj1->contable->dobj.name,
+
robj2->contable->dobj.name);
+ if (cmpval != 0)
+ return cmpval;
+ }
+ }
+ else if (obj1->objType == DO_PUBLICATION_REL)
+ {
+ PublicationRelInfo *probj1 = *(PublicationRelInfo *const *) p1;
+ PublicationRelInfo *probj2 = *(PublicationRelInfo *const *) p2;
+
+ /* Sort by publication name, since (namespace, name) match the
rel */
+ cmpval = strcmp(probj1->publication->dobj.name,
+ probj2->publication->dobj.name);
+ if (cmpval != 0)
+ return cmpval;
+ }
- /* Usually shouldn't get here, but if we do, sort by OID */
+ /*
+ * Shouldn't get here except after catalog corruption, but if we do,
sort
+ * by OID. This may make logically-identical databases differ in the
+ * order of objects in dump output. Users will get spurious schema
diffs.
+ * Expect flaky failures of 002_pg_upgrade.pl test 'dump outputs from
+ * original and restored regression databases match' if the regression
+ * database contains objects allowing that test to reach here. That's a
+ * consequence of the test using "pg_restore -j", which doesn't fully
+ * constrain OID assignment order.
+ */
+ Assert(false);
return oidcmp(obj1->catId.oid, obj2->catId.oid);
}
+/* Compare two OID-identified pg_type values by nspname, then by typname. */
+static int
+pgTypeNameCompare(Oid typid1, Oid typid2)
+{
+ TypeInfo *typobj1;
+ TypeInfo *typobj2;
+ int cmpval;
+
+ if (typid1 == typid2)
+ return 0;
+
+ typobj1 = findTypeByOid(typid1);
+ typobj2 = findTypeByOid(typid2);
+
+ if (!typobj1 || !typobj2)
+ {
+ /*
+ * getTypes() didn't find some OID. Assume catalog corruption,
e.g.
+ * an oprright value without the corresponding OID in a pg_type
row.
+ * Report as "equal", so the caller uses the next available
basis for
+ * comparison, e.g. the next function argument.
+ *
+ * Unary operators have InvalidOid in oprleft (if oprkind='r')
or in
+ * oprright (if oprkind='l'). Caller already sorted by oprkind,
+ * calling us only for like-kind operators. Hence, "typid1 ==
typid2"
+ * took care of InvalidOid. (v14 removed postfix operator
support.
+ * Hence, when dumping from v14+, only oprleft can be
InvalidOid.)
+ */
+ Assert(false);
+ return 0;
+ }
+
+ if (!typobj1->dobj.namespace || !typobj2->dobj.namespace)
+ Assert(false); /* catalog corruption */
+ else
+ {
+ cmpval = strcmp(typobj1->dobj.namespace->dobj.name,
+
typobj2->dobj.namespace->dobj.name);
+ if (cmpval != 0)
+ return cmpval;
+ }
+ return strcmp(typobj1->dobj.name, typobj2->dobj.name);
+}
+
+/* Compare two OID-identified pg_am values by amname. */
+static int
+accessMethodNameCompare(Oid am1, Oid am2)
+{
+ AccessMethodInfo *amobj1;
+ AccessMethodInfo *amobj2;
+
+ if (am1 == am2)
+ return 0;
+
+ amobj1 = findAccessMethodByOid(am1);
+ amobj2 = findAccessMethodByOid(am2);
+
+ if (!amobj1 || !amobj2)
+ {
+ /* catalog corruption: handle like pgTypeNameCompare() does */
+ Assert(false);
+ return 0;
+ }
+
+ return strcmp(amobj1->dobj.name, amobj2->dobj.name);
+}
+
/*
* Sort the given objects into a safe dump order using dependency
diff --git a/src/test/regress/expected/publication.out
b/src/test/regress/expected/publication.out
index b7ce080..35595fb 100644
--- a/src/test/regress/expected/publication.out
+++ b/src/test/regress/expected/publication.out
@@ -344,3 +344,11 @@ NOTICE: drop cascades to table pub_test.testpub_nopk
RESET SESSION AUTHORIZATION;
DROP ROLE regress_publication_user, regress_publication_user2;
DROP ROLE regress_publication_user_dummy;
+-- stage objects for pg_dump tests
+CREATE SCHEMA pubme CREATE TABLE t0 (c int, d int) CREATE TABLE t1 (c int);
+SET client_min_messages = 'ERROR';
+CREATE PUBLICATION dump_pub_1ct FOR TABLE ONLY pubme.t0;
+CREATE PUBLICATION dump_pub_2ct FOR TABLE ONLY pubme.t0, pubme.t1;
+CREATE PUBLICATION dump_pub_all FOR TABLE ONLY pubme.t0, pubme.t1
+ WITH (publish_via_partition_root = true);
+RESET client_min_messages;
diff --git a/src/test/regress/sql/publication.sql
b/src/test/regress/sql/publication.sql
index 7d5c937..021d4e2 100644
--- a/src/test/regress/sql/publication.sql
+++ b/src/test/regress/sql/publication.sql
@@ -218,3 +218,12 @@ DROP SCHEMA pub_test CASCADE;
RESET SESSION AUTHORIZATION;
DROP ROLE regress_publication_user, regress_publication_user2;
DROP ROLE regress_publication_user_dummy;
+
+-- stage objects for pg_dump tests
+CREATE SCHEMA pubme CREATE TABLE t0 (c int, d int) CREATE TABLE t1 (c int);
+SET client_min_messages = 'ERROR';
+CREATE PUBLICATION dump_pub_1ct FOR TABLE ONLY pubme.t0;
+CREATE PUBLICATION dump_pub_2ct FOR TABLE ONLY pubme.t0, pubme.t1;
+CREATE PUBLICATION dump_pub_all FOR TABLE ONLY pubme.t0, pubme.t1
+ WITH (publish_via_partition_root = true);
+RESET client_min_messages;
