Re: automatically generating node support functions - Mailing list pgsql-hackers

From Tom Lane
Subject Re: automatically generating node support functions
Date
Msg-id 2592455.1657140387@sss.pgh.pa.us
Whole thread Raw
In response to Re: automatically generating node support functions  (Peter Eisentraut <peter.eisentraut@enterprisedb.com>)
Responses Re: automatically generating node support functions
Re: automatically generating node support functions
List pgsql-hackers
Peter Eisentraut <peter.eisentraut@enterprisedb.com> writes:
> [ v7-0001-Automatically-generate-node-support-functions.patch ]

I have gone through this and made some proposed changes (attached),
and I think it is almost committable.  There is one nasty problem
we need a solution to, which is that pgindent is not at all on board
with this idea of attaching node attrs to typedefs.  It pushes them
to the next line, like this:

@@ -691,7 +709,8 @@
      (rel)->reloptkind == RELOPT_OTHER_JOINREL || \
      (rel)->reloptkind == RELOPT_OTHER_UPPER_REL)
 
-typedef struct RelOptInfo pg_node_attr(no_copy_equal, no_read)
+typedef struct RelOptInfo
+pg_node_attr(no_copy_equal, no_read)
 {
     NodeTag        type;
 
which is already enough to break the simplistic parsing in
gen_node_support.pl.  Now, we could fix that parsing logic to deal
with this layout, but this also seems to change pgindent's opinion
of whether the subsequent braced material is part of a typedef or a
function.  That results in it injecting a lot of vertical space
that wasn't there before, which is annoying.

I experimented a bit and found that we could do it this way:

 typedef struct RelOptInfo
 {
+    pg_node_attr(no_copy_equal, no_read)
+
     NodeTag        type;

without (AFAICT) confusing pgindent, but I've not tried to adapt
the perl script or the code to that style.

Anyway, besides that, I have some comments that I've implemented
in the attached delta patch.

* After further thought I'm okay with your theory that attaching
special copy/equal rules to specific field types is appropriate.
We might at some point want the pg_node_attr(copy_as_scalar)
approach too, but we can always add that later.  However, I thought
some more comments about it were needed in the *nodes.h files,
so I added those.  (My general feeling about this is that if
anyone needs to look into gen_node_support.pl to understand how
the backend works, we've failed at documentation.)

* As written, the patch created equal() support for all Plan structs,
which is quite a bit of useless code bloat.  I solved this by
separating no_copy and no_equal properties, so that we could mark
Plan as no_equal while still having copy support.

* I did not like the semantics of copy_ignore one bit: it was
relying on the pre-zeroing behavior of makeNode() to be sane at
all, and I don't want that to be a requirement.  (I foresee
wanting to flat-copy node contents and turn COPY_SCALAR_FIELD
into a no-op.)  I replaced it with copy_as(VALUE) to provide
better-defined semantics.

* Likewise, read_write_ignore left the contents of the field after
reading too squishy for me.  I invented read_as(VALUE) parallel
to copy_as() to fix the semantics, and added a check that you
can only use read_write_ignore if the struct is no_read or
you provide read_as().  (This could be factored differently
of course.)

* I threw in a bunch more no_read markers to bring the readfuncs.c
contents into closer alignment with what we have today.  Maybe
there is an argument for accepting that code bloat, but it's a
discussion to have later.  In any case, most of the pathnodes.h
structs HAVE to be marked no_read because there's no sane way
to reconstruct them from outfuncs output.

* I got rid of the code that stripped underscores from outfuncs
struct labels.  That seemed like an entirely unnecessary
behavioral change.

* FWIW, I'm okay with the question about

         # XXX Previously, for subtyping, only the leaf field name is
         # used. Ponder whether we want to keep it that way.

I thought that it might make the output too cluttered, but after
some study of the results from printing plans and planner data
structures, it's not a big addition, and indeed I kind of like it.

* Fixed a bug in write_only_req_outer code.

* Made Plan and Join into abstract nodes.

Anyway, if we can fix the impedance mismatch with pgindent,
I think this is committable.  There is a lot of follow-on
work that could be considered, but I'd like to get the present
changes in place ASAP so that other patches can be rebased
onto something stable.

I've attached a delta patch, and also repeated v7 so as not
to confuse the cfbot.

            regards, tom lane

From c82ee081a7a8cdc77b44f325d1df695b55a60b06 Mon Sep 17 00:00:00 2001
From: Peter Eisentraut <peter@eisentraut.org>
Date: Wed, 6 Jul 2022 12:13:32 +0200
Subject: [PATCH v7] Automatically generate node support functions

Add a script to automatically generate the node support functions
(copy, equal, out, and read, as well as the node tags enum) from the
struct definitions.

For each of the four node support files, it creates two include files,
e.g., copyfuncs.funcs.c and copyfuncs.switch.c, to include in the main
file.  All the scaffolding of the main file stays in place.

TODO: In this patch, I have only ifdef'ed out the code to could be
removed, mainly so that it won't constantly have merge conflicts.
Eventually, that should all be changed to delete the code.  All the
code comments that are worth keeping from those sections have already
been moved to the header files where the structs are defined.

I have tried to mostly make the coverage of the output match what is
currently there.  For example, one could now do out/read coverage of
utility statement nodes, but I have manually excluded those for now.
The reason is mainly that it's easier to diff the before and after,
and adding a bunch of stuff like this might require a separate
analysis and review.

Subtyping (TidScan -> Scan) is supported.

For the hard cases, you can just write a manual function and exclude
generating one.  For the not so hard cases, there is a way of
annotating struct fields to get special behaviors.  For example,
pg_node_attr(equal_ignore) has the field ignored in equal functions.

Discussion: https://www.postgresql.org/message-id/flat/c1097590-a6a4-486a-64b1-e1f9cc0533ce%40enterprisedb.com
---
 src/backend/Makefile                  |  10 +-
 src/backend/nodes/.gitignore          |   4 +
 src/backend/nodes/Makefile            |  59 ++
 src/backend/nodes/copyfuncs.c         |  19 +-
 src/backend/nodes/equalfuncs.c        |  22 +-
 src/backend/nodes/gen_node_support.pl | 770 ++++++++++++++++++++++++++
 src/backend/nodes/outfuncs.c          |  34 +-
 src/backend/nodes/readfuncs.c         |  23 +-
 src/include/Makefile                  |   1 +
 src/include/executor/tuptable.h       |   8 +-
 src/include/nodes/.gitignore          |   2 +
 src/include/nodes/extensible.h        |   2 +-
 src/include/nodes/nodes.h             |  52 ++
 src/include/nodes/parsenodes.h        |  19 +-
 src/include/nodes/pathnodes.h         | 315 +++++++----
 src/include/nodes/plannodes.h         |  92 +--
 src/include/nodes/primnodes.h         |  45 +-
 src/include/nodes/value.h             |  10 +-
 src/include/utils/rel.h               |  11 +-
 src/tools/msvc/Solution.pm            |  46 ++
 20 files changed, 1320 insertions(+), 224 deletions(-)
 create mode 100644 src/backend/nodes/.gitignore
 create mode 100644 src/backend/nodes/gen_node_support.pl
 create mode 100644 src/include/nodes/.gitignore

diff --git a/src/backend/Makefile b/src/backend/Makefile
index 4a02006788..953c80db5a 100644
--- a/src/backend/Makefile
+++ b/src/backend/Makefile
@@ -143,11 +143,15 @@ storage/lmgr/lwlocknames.h: storage/lmgr/generate-lwlocknames.pl storage/lmgr/lw
 submake-catalog-headers:
     $(MAKE) -C catalog distprep generated-header-symlinks

+# run this unconditionally to avoid needing to know its dependencies here:
+submake-nodes-headers:
+    $(MAKE) -C nodes distprep generated-header-symlinks
+
 # run this unconditionally to avoid needing to know its dependencies here:
 submake-utils-headers:
     $(MAKE) -C utils distprep generated-header-symlinks

-.PHONY: submake-catalog-headers submake-utils-headers
+.PHONY: submake-catalog-headers submake-nodes-headers submake-utils-headers

 # Make symlinks for these headers in the include directory. That way
 # we can cut down on the -I options. Also, a symlink is automatically
@@ -162,7 +166,7 @@ submake-utils-headers:

 .PHONY: generated-headers

-generated-headers: $(top_builddir)/src/include/parser/gram.h $(top_builddir)/src/include/storage/lwlocknames.h
submake-catalog-headerssubmake-utils-headers 
+generated-headers: $(top_builddir)/src/include/parser/gram.h $(top_builddir)/src/include/storage/lwlocknames.h
submake-catalog-headerssubmake-nodes-headers submake-utils-headers 

 $(top_builddir)/src/include/parser/gram.h: parser/gram.h
     prereqdir=`cd '$(dir $<)' >/dev/null && pwd` && \
@@ -185,6 +189,7 @@ distprep:
     $(MAKE) -C parser    gram.c gram.h scan.c
     $(MAKE) -C bootstrap    bootparse.c bootscanner.c
     $(MAKE) -C catalog    distprep
+    $(MAKE) -C nodes    distprep
     $(MAKE) -C replication    repl_gram.c repl_scanner.c syncrep_gram.c syncrep_scanner.c
     $(MAKE) -C storage/lmgr    lwlocknames.h lwlocknames.c
     $(MAKE) -C utils    distprep
@@ -297,6 +302,7 @@ distclean: clean

 maintainer-clean: distclean
     $(MAKE) -C catalog $@
+    $(MAKE) -C nodes $@
     $(MAKE) -C utils $@
     rm -f bootstrap/bootparse.c \
           bootstrap/bootscanner.c \
diff --git a/src/backend/nodes/.gitignore b/src/backend/nodes/.gitignore
new file mode 100644
index 0000000000..0c14b5697b
--- /dev/null
+++ b/src/backend/nodes/.gitignore
@@ -0,0 +1,4 @@
+/node-support-stamp
+/nodetags.h
+/*funcs.funcs.c
+/*funcs.switch.c
diff --git a/src/backend/nodes/Makefile b/src/backend/nodes/Makefile
index 5d2b12a993..1a0d5b9314 100644
--- a/src/backend/nodes/Makefile
+++ b/src/backend/nodes/Makefile
@@ -30,3 +30,62 @@ OBJS = \
     value.o

 include $(top_srcdir)/src/backend/common.mk
+
+node_headers = \
+    nodes/nodes.h \
+    nodes/execnodes.h \
+    nodes/plannodes.h \
+    nodes/primnodes.h \
+    nodes/pathnodes.h \
+    nodes/extensible.h \
+    nodes/parsenodes.h \
+    nodes/replnodes.h \
+    nodes/value.h \
+    commands/trigger.h \
+    commands/event_trigger.h \
+    foreign/fdwapi.h \
+    access/amapi.h \
+    access/tableam.h \
+    access/tsmapi.h \
+    utils/rel.h \
+    nodes/supportnodes.h \
+    executor/tuptable.h \
+    nodes/lockoptions.h \
+    access/sdir.h
+
+# see also catalog/Makefile for an explanation of these make rules
+
+all: distprep generated-header-symlinks
+
+distprep: node-support-stamp
+
+.PHONY: generated-header-symlinks
+
+generated-header-symlinks: $(top_builddir)/src/include/nodes/header-stamp
+
+# node-support-stamp records the last time we ran gen_node_support.pl.
+# We don't rely on the timestamps of the individual output files,
+# because the Perl script won't update them if they didn't change (to
+# avoid unnecessary recompiles).
+node-support-stamp: gen_node_support.pl $(addprefix $(top_srcdir)/src/include/,$(node_headers))
+    $(PERL) $^
+    touch $@
+
+# These generated headers must be symlinked into builddir/src/include/,
+# using absolute links for the reasons explained in src/backend/Makefile.
+# We use header-stamp to record that we've done this because the symlinks
+# themselves may appear older than node-support-stamp.
+$(top_builddir)/src/include/nodes/header-stamp: node-support-stamp
+    prereqdir=`cd '$(dir $<)' >/dev/null && pwd` && \
+    cd '$(dir $@)' && for file in nodetags.h; do \
+      rm -f $$file && $(LN_S) "$$prereqdir/$$file" . ; \
+    done
+    touch $@
+
+copyfuncs.o: copyfuncs.c copyfuncs.funcs.c copyfuncs.switch.c | node-support-stamp
+equalfuncs.o: equalfuncs.c equalfuncs.funcs.c equalfuncs.switch.c | node-support-stamp
+outfuncs.o: outfuncs.c outfuncs.funcs.c outfuncs.switch.c | node-support-stamp
+readfuncs.o:  readfuncs.c readfuncs.funcs.c readfuncs.switch.c | node-support-stamp
+
+maintainer-clean: clean
+    rm -f node-support-stamp $(addsuffix funcs.funcs.c,copy equal out read) $(addsuffix funcs.switch.c,copy equal out
read)nodetags.h 
diff --git a/src/backend/nodes/copyfuncs.c b/src/backend/nodes/copyfuncs.c
index 706d283a92..48778aa4ef 100644
--- a/src/backend/nodes/copyfuncs.c
+++ b/src/backend/nodes/copyfuncs.c
@@ -23,11 +23,7 @@
 #include "postgres.h"

 #include "miscadmin.h"
-#include "nodes/extensible.h"
-#include "nodes/pathnodes.h"
-#include "nodes/plannodes.h"
 #include "utils/datum.h"
-#include "utils/rel.h"


 /*
@@ -73,6 +69,9 @@
     (newnode->fldname = from->fldname)


+#include "copyfuncs.funcs.c"
+
+#ifdef OBSOLETE
 /* ****************************************************************
  *                     plannodes.h copy functions
  * ****************************************************************
@@ -1465,6 +1464,7 @@ _copyVar(const Var *from)

     return newnode;
 }
+#endif /*OBSOLETE*/

 /*
  * _copyConst
@@ -1504,6 +1504,7 @@ _copyConst(const Const *from)
     return newnode;
 }

+#ifdef OBSOLETE
 /*
  * _copyParam
  */
@@ -3248,6 +3249,7 @@ _copyParamRef(const ParamRef *from)

     return newnode;
 }
+#endif /*OBSOLETE*/

 static A_Const *
 _copyA_Const(const A_Const *from)
@@ -3288,6 +3290,7 @@ _copyA_Const(const A_Const *from)
     return newnode;
 }

+#ifdef OBSOLETE
 static FuncCall *
 _copyFuncCall(const FuncCall *from)
 {
@@ -5453,6 +5456,7 @@ _copyDropSubscriptionStmt(const DropSubscriptionStmt *from)

     return newnode;
 }
+#endif /*OBSOLETE*/

 /* ****************************************************************
  *                    extensible.h copy functions
@@ -5475,6 +5479,7 @@ _copyExtensibleNode(const ExtensibleNode *from)
     return newnode;
 }

+#ifdef OBSOLETE
 /* ****************************************************************
  *                    value.h copy functions
  * ****************************************************************
@@ -5545,6 +5550,7 @@ _copyForeignKeyCacheInfo(const ForeignKeyCacheInfo *from)

     return newnode;
 }
+#endif /*OBSOLETE*/

 /*
  * copyObjectImpl -- implementation of copyObject(); see nodes/nodes.h
@@ -5565,6 +5571,8 @@ copyObjectImpl(const void *from)

     switch (nodeTag(from))
     {
+#include "copyfuncs.switch.c"
+#ifdef OBSOLETE
             /*
              * PLAN NODES
              */
@@ -6009,6 +6017,7 @@ copyObjectImpl(const void *from)
         case T_BitString:
             retval = _copyBitString(from);
             break;
+#endif /*OBSOLETE*/

             /*
              * LIST NODES
@@ -6026,6 +6035,7 @@ copyObjectImpl(const void *from)
             retval = list_copy(from);
             break;

+#ifdef OBSOLETE
             /*
              * EXTENSIBLE NODES
              */
@@ -6577,6 +6587,7 @@ copyObjectImpl(const void *from)
         case T_ForeignKeyCacheInfo:
             retval = _copyForeignKeyCacheInfo(from);
             break;
+#endif /*OBSOLETE*/

         default:
             elog(ERROR, "unrecognized node type: %d", (int) nodeTag(from));
diff --git a/src/backend/nodes/equalfuncs.c b/src/backend/nodes/equalfuncs.c
index fccc0b4a18..7f09ccd978 100644
--- a/src/backend/nodes/equalfuncs.c
+++ b/src/backend/nodes/equalfuncs.c
@@ -10,9 +10,6 @@
  * because the circular linkages between RelOptInfo and Path nodes can't
  * be handled easily in a simple depth-first traversal.
  *
- * Currently, in fact, equal() doesn't know how to compare Plan trees
- * either.  This might need to be fixed someday.
- *
  * NOTE: it is intentional that parse location fields (in nodes that have
  * one) are not compared.  This is because we want, for example, a variable
  * "x" to be considered equal() to another reference to "x" in the query.
@@ -30,8 +27,6 @@
 #include "postgres.h"

 #include "miscadmin.h"
-#include "nodes/extensible.h"
-#include "nodes/pathnodes.h"
 #include "utils/datum.h"


@@ -97,6 +92,9 @@
     ((void) 0)


+#include "equalfuncs.funcs.c"
+
+#ifdef OBSOLETE
 /*
  *    Stuff from primnodes.h
  */
@@ -258,6 +256,7 @@ _equalVar(const Var *a, const Var *b)

     return true;
 }
+#endif /*OBSOLETE*/

 static bool
 _equalConst(const Const *a, const Const *b)
@@ -280,6 +279,7 @@ _equalConst(const Const *a, const Const *b)
                         a->constbyval, a->constlen);
 }

+#ifdef OBSOLETE
 static bool
 _equalParam(const Param *a, const Param *b)
 {
@@ -1304,6 +1304,7 @@ _equalPlaceHolderInfo(const PlaceHolderInfo *a, const PlaceHolderInfo *b)

     return true;
 }
+#endif /*OBSOLETE*/

 /*
  * Stuff from extensible.h
@@ -1325,6 +1326,7 @@ _equalExtensibleNode(const ExtensibleNode *a, const ExtensibleNode *b)
     return true;
 }

+#ifdef OBSOLETE
 /*
  * Stuff from parsenodes.h
  */
@@ -2815,6 +2817,7 @@ _equalParamRef(const ParamRef *a, const ParamRef *b)

     return true;
 }
+#endif /*OBSOLETE*/

 static bool
 _equalA_Const(const A_Const *a, const A_Const *b)
@@ -2831,6 +2834,7 @@ _equalA_Const(const A_Const *a, const A_Const *b)
     return true;
 }

+#ifdef OBSOLETE
 static bool
 _equalFuncCall(const FuncCall *a, const FuncCall *b)
 {
@@ -3468,6 +3472,7 @@ _equalPartitionCmd(const PartitionCmd *a, const PartitionCmd *b)

     return true;
 }
+#endif /*OBSOLETE*/

 /*
  * Stuff from pg_list.h
@@ -3528,6 +3533,7 @@ _equalList(const List *a, const List *b)
     return true;
 }

+#ifdef OBSOLETE
 /*
  * Stuff from value.h
  */
@@ -3571,6 +3577,7 @@ _equalBitString(const BitString *a, const BitString *b)

     return true;
 }
+#endif /*OBSOLETE*/

 /*
  * equal
@@ -3601,6 +3608,8 @@ equal(const void *a, const void *b)

     switch (nodeTag(a))
     {
+#include "equalfuncs.switch.c"
+#ifdef OBSOLETE
             /*
              * PRIMITIVE NODES
              */
@@ -3821,6 +3830,7 @@ equal(const void *a, const void *b)
         case T_PlaceHolderInfo:
             retval = _equalPlaceHolderInfo(a, b);
             break;
+#endif /*OBSOLETE*/

         case T_List:
         case T_IntList:
@@ -3828,6 +3838,7 @@ equal(const void *a, const void *b)
             retval = _equalList(a, b);
             break;

+#ifdef OBSOLETE
         case T_Integer:
             retval = _equalInteger(a, b);
             break;
@@ -4430,6 +4441,7 @@ equal(const void *a, const void *b)
         case T_JsonTableColumn:
             retval = _equalJsonTableColumn(a, b);
             break;
+#endif /*OBSOLETE*/

         default:
             elog(ERROR, "unrecognized node type: %d",
diff --git a/src/backend/nodes/gen_node_support.pl b/src/backend/nodes/gen_node_support.pl
new file mode 100644
index 0000000000..6aaf401a72
--- /dev/null
+++ b/src/backend/nodes/gen_node_support.pl
@@ -0,0 +1,770 @@
+#!/usr/bin/perl
+#----------------------------------------------------------------------
+#
+# Generate node support files:
+# - nodetags.h
+# - copyfuncs
+# - equalfuncs
+# - readfuncs
+# - outfuncs
+#
+# Portions Copyright (c) 1996-2022, PostgreSQL Global Development Group
+# Portions Copyright (c) 1994, Regents of the University of California
+#
+# src/backend/nodes/gen_node_support.pl
+#
+#----------------------------------------------------------------------
+
+use strict;
+use warnings;
+
+use File::Basename;
+
+use FindBin;
+use lib "$FindBin::RealBin/../catalog";
+
+use Catalog;  # for RenameTempFile
+
+
+# Test whether first argument is element of the list in the second
+# argument
+sub elem
+{
+    my $x = shift;
+    return grep { $_ eq $x } @_;
+}
+
+
+# collect node names
+my @node_types = qw(Node);
+# collect info for each node type
+my %node_type_info;
+
+# node types we don't want copy/equal support for
+my @no_copy_equal;
+# node types we don't want read support for
+my @no_read;
+# node types we don't want read/write support for
+my @no_read_write;
+
+# types that are copied by straight assignment
+my @scalar_types = qw(
+    bits32 bool char double int int8 int16 int32 int64 long uint8 uint16 uint32 uint64
+    AclMode AttrNumber Cardinality Cost Index Oid Selectivity Size StrategyNumber SubTransactionId TimeLineID
XLogRecPtr
+);
+
+# collect enum types
+my @enum_types;
+
+my @abstract_types = qw(Node);
+
+# Special cases that either don't have their own struct or the struct
+# is not in a header file.  We just generate node tags for them, but
+# they otherwise don't participate in node support.
+my @extra_tags = qw(
+    IntList OidList XidList
+    AllocSetContext GenerationContext SlabContext
+    TIDBitmap
+    WindowObjectData
+);
+
+# This is a regular node, but we skip parsing it from its header file
+# since we won't use its internal structure here anyway.
+push @node_types, qw(List);
+# See special treatment in outNode() and nodeRead().
+push @no_read_write, qw(List);
+
+# Nodes with custom copy/equal implementations are skipped from
+# .funcs.c but need case statements in .switch.c.
+my @custom_copy_equal;
+
+# Similarly for custom read/write implementations.
+my @custom_read_write;
+
+# EquivalenceClasses are never moved, so just shallow-copy the pointer
+push @scalar_types, qw(EquivalenceClass* EquivalenceMember*);
+
+# This is a struct, so we can copy it by assignment.  Equal support is
+# currently not required.
+push @scalar_types, qw(QualCost);
+
+# XXX various things we are not publishing right now to stay level
+# with the manual system
+push @no_copy_equal, qw(CallContext InlineCodeBlock);
+push @no_read_write, qw(AccessPriv AlterTableCmd CallContext CreateOpClassItem FunctionParameter InferClause
InlineCodeBlockObjectWithArgs OnConflictClause PartitionCmd RoleSpec VacuumRelation); 
+
+
+## read input
+
+foreach my $infile (@ARGV)
+{
+    my $in_struct;
+    my $subline;
+    my $is_node_struct;
+    my $supertype;
+    my $supertype_field;
+
+    my @my_fields;
+    my %my_field_types;
+    my %my_field_attrs;
+
+    open my $ifh, '<', $infile or die "could not open \"$infile\": $!";
+
+    my $file_content = do { local $/; <$ifh> };
+
+    # strip C comments
+    $file_content =~ s{/\*.*?\*/}{}gs;
+
+    foreach my $line (split /\n/, $file_content)
+    {
+        chomp $line;
+        $line =~ s/\s*$//;
+        next if $line eq '';
+        next if $line =~ /^#(define|ifdef|endif)/;
+
+        # we are analyzing a struct definition
+        if ($in_struct)
+        {
+            $subline++;
+
+            # first line should have opening brace
+            if ($subline == 1)
+            {
+                $is_node_struct = 0;
+                $supertype = undef;
+                next if $line eq '{';
+                die;
+            }
+            # second line should have node tag or supertype
+            elsif ($subline == 2)
+            {
+                if ($line =~ /^\s*NodeTag\s+type;/)
+                {
+                    $is_node_struct = 1;
+                    next;
+                }
+                elsif ($line =~ /\s*(\w+)\s+(\w+);/ and elem $1, @node_types)
+                {
+                    $is_node_struct = 1;
+                    $supertype = $1;
+                    $supertype_field = $2;
+                    next;
+                }
+            }
+
+            # end of struct
+            if ($line =~ /^\}\s*$in_struct;$/ || $line =~ /^\};$/)
+            {
+                if ($is_node_struct)
+                {
+                    # This is the end of a node struct definition.
+                    # Save everything we have collected.
+
+                    # node name
+                    push @node_types, $in_struct;
+
+                    # field names, types, attributes
+                    my @f = @my_fields;
+                    my %ft = %my_field_types;
+                    my %fa = %my_field_attrs;
+
+                    # If there is a supertype, add those fields, too.
+                    if ($supertype)
+                    {
+                        my @superfields;
+                        foreach my $sf (@{$node_type_info{$supertype}->{fields}})
+                        {
+                            my $fn = "${supertype_field}.$sf";
+                            push @superfields, $fn;
+                            $ft{$fn} = $node_type_info{$supertype}->{field_types}{$sf};
+                            if ($node_type_info{$supertype}->{field_attrs}{$sf})
+                            {
+                                # Copy any attributes, adjusting array_size field references
+                                my @newa = @{$node_type_info{$supertype}->{field_attrs}{$sf}};
+                                foreach my $a (@newa)
+                                {
+                                    $a =~ s/array_size\((\w+)\)/array_size(${supertype_field}.$1)/;
+                                }
+                                $fa{$fn} = \@newa;
+                            }
+                        }
+                        unshift @f, @superfields;
+                    }
+                    # save in global info structure
+                    $node_type_info{$in_struct}->{fields} = \@f;
+                    $node_type_info{$in_struct}->{field_types} = \%ft;
+                    $node_type_info{$in_struct}->{field_attrs} = \%fa;
+
+                    # Nodes from these files don't need to be
+                    # supported, except the node tags.
+                    if (elem basename($infile),
+                        qw(execnodes.h trigger.h event_trigger.h amapi.h tableam.h
+                            tsmapi.h fdwapi.h tuptable.h replnodes.h supportnodes.h))
+                    {
+                        push @no_copy_equal, $in_struct;
+                        push @no_read_write, $in_struct;
+                    }
+
+                    # Propagate some node attributes from supertypes
+                    if ($supertype)
+                    {
+                        push @no_copy_equal, $in_struct if elem $supertype, @no_copy_equal;
+                        push @no_read, $in_struct if elem $supertype, @no_read;
+                    }
+                }
+
+                # start new cycle
+                $in_struct = undef;
+                @my_fields = ();
+                %my_field_types = ();
+                %my_field_attrs = ();
+            }
+            # normal struct field
+            elsif ($line =~ /^\s*(.+)\s*\b(\w+)(\[\w+\])?\s*(?:pg_node_attr\(([\w(), ]*)\))?;/)
+            {
+                if ($is_node_struct)
+                {
+                    my $type = $1;
+                    my $name = $2;
+                    my $array_size = $3;
+                    my $attrs = $4;
+
+                    # strip "const"
+                    $type =~ s/^const\s*//;
+                    # strip trailing space
+                    $type =~ s/\s*$//;
+                    # strip space between type and "*" (pointer) */
+                    $type =~ s/\s+\*$/*/;
+
+                    die if $type eq '';
+
+                    my @attrs;
+                    if ($attrs)
+                    {
+                        @attrs = split /,\s*/, $attrs;
+                        foreach my $attr (@attrs)
+                        {
+                            if ($attr !~ /^array_size\(\w+\)$/ &&
+                                !elem $attr, qw(copy_ignore equal_ignore equal_ignore_if_zero read_write_ignore
+                                    write_only_relids write_only_nondefault_pathtarget write_only_req_outer))
+                            {
+                                die "$infile:$.: unrecognized attribute \"$attr\"\n";
+                            }
+                        }
+                    }
+
+                    $type = $type . $array_size if $array_size;
+                    push @my_fields, $name;
+                    $my_field_types{$name} = $type;
+                    $my_field_attrs{$name} = \@attrs;
+                }
+            }
+            else
+            {
+                if ($is_node_struct)
+                {
+                    #warn "$infile:$.: could not parse \"$line\"\n";
+                }
+            }
+        }
+        # not in a struct
+        else
+        {
+            # start of a struct?
+            if ($line =~ /^(?:typedef )?struct (\w+)\s*(?:pg_node_attr\(([\w(), ]*)\))?$/ && $1 ne 'Node')
+            {
+                $in_struct = $1;
+                my $node_attrs = $2 || '';
+                $subline = 0;
+
+                foreach my $attr (split /,\s*/, $node_attrs)
+                {
+                    if ($attr eq 'abstract')
+                    {
+                        push @abstract_types, $in_struct;
+                    }
+                    elsif ($attr eq 'custom_copy_equal')
+                    {
+                        push @custom_copy_equal, $in_struct;
+                    }
+                    elsif ($attr eq 'custom_read_write')
+                    {
+                        push @custom_read_write, $in_struct;
+                    }
+                    elsif ($attr eq 'no_copy_equal')
+                    {
+                        push @no_copy_equal, $in_struct;
+                    }
+                    elsif ($attr eq 'no_read')
+                    {
+                        push @no_read, $in_struct;
+                    }
+                    elsif ($attr eq 'special_read_write')
+                    {
+                        # This attribute is called
+                        # "special_read_write" because there is
+                        # special treatment in outNode() and
+                        # nodeRead() for these nodes.  For this
+                        # script, it's the same as "no_read_write",
+                        # but calling the attribute that externally
+                        # would probably be confusing, since
+                        # read/write support does in fact exist.
+                        push @no_read_write, $in_struct;
+                    }
+                    else
+                    {
+                        die "$infile:$.: unrecognized attribute \"$attr\"\n";
+                    }
+                }
+            }
+            # one node type typedef'ed directly from another
+            elsif ($line =~ /^typedef (\w+) (\w+);$/ and elem $1, @node_types)
+            {
+                my $alias_of = $1;
+                my $n = $2;
+
+                # copy everything over
+                push @node_types, $n;
+                my @f = @{$node_type_info{$alias_of}->{fields}};
+                my %ft = %{$node_type_info{$alias_of}->{field_types}};
+                my %fa = %{$node_type_info{$alias_of}->{field_attrs}};
+                $node_type_info{$n}->{fields} = \@f;
+                $node_type_info{$n}->{field_types} = \%ft;
+                $node_type_info{$n}->{field_attrs} = \%fa;
+            }
+            # collect enum names
+            elsif ($line =~ /^typedef enum (\w+)(\s*\/\*.*)?$/)
+            {
+                push @enum_types, $1;
+            }
+        }
+    }
+
+    if ($in_struct)
+    {
+        die "runaway \"$in_struct\" in file \"$infile\"\n";
+    }
+
+    close $ifh;
+} # for each file
+
+
+## write output
+
+my $tmpext  = ".tmp$$";
+
+# nodetags.h
+
+open my $nt, '>', 'nodetags.h' . $tmpext or die $!;
+
+my $i = 1;
+foreach my $n (@node_types,@extra_tags)
+{
+    next if elem $n, @abstract_types;
+    print $nt "\tT_${n} = $i,\n";
+    $i++;
+}
+
+close $nt;
+
+
+# make #include lines necessary to pull in all the struct definitions
+my $node_includes = '';
+foreach my $infile (sort @ARGV)
+{
+    $infile =~ s!.*src/include/!!;
+    $node_includes .= qq{#include "$infile"\n};
+}
+
+
+# copyfuncs.c, equalfuncs.c
+
+open my $cff, '>', 'copyfuncs.funcs.c' . $tmpext or die $!;
+open my $eff, '>', 'equalfuncs.funcs.c' . $tmpext or die $!;
+open my $cfs, '>', 'copyfuncs.switch.c' . $tmpext or die $!;
+open my $efs, '>', 'equalfuncs.switch.c' . $tmpext or die $!;
+
+# add required #include lines to each file set
+print $cff $node_includes;
+print $eff $node_includes;
+
+foreach my $n (@node_types)
+{
+    next if elem $n, @abstract_types;
+    next if elem $n, @no_copy_equal;
+    next if $n eq 'List';
+
+    print $cfs "\t\tcase T_${n}:\n".
+      "\t\t\tretval = _copy${n}(from);\n".
+      "\t\t\tbreak;\n";
+
+    print $efs "\t\tcase T_${n}:\n".
+      "\t\t\tretval = _equal${n}(a, b);\n".
+      "\t\t\tbreak;\n";
+
+    next if elem $n, @custom_copy_equal;
+
+    print $cff "
+static $n *
+_copy${n}(const $n *from)
+{
+\t${n} *newnode = makeNode($n);
+
+";
+
+    print $eff "
+static bool
+_equal${n}(const $n *a, const $n *b)
+{
+";
+
+    # print instructions for each field
+    foreach my $f (@{$node_type_info{$n}->{fields}})
+    {
+        my $t = $node_type_info{$n}->{field_types}{$f};
+        my @a = @{ $node_type_info{$n}->{field_attrs}{$f} };
+        my $copy_ignore = (elem 'copy_ignore', @a);
+        my $equal_ignore = (elem 'equal_ignore', @a);
+
+        # select instructions by field type
+        if ($t eq 'char*')
+        {
+            print $cff "\tCOPY_STRING_FIELD($f);\n" unless $copy_ignore;
+            print $eff "\tCOMPARE_STRING_FIELD($f);\n" unless $equal_ignore;
+        }
+        elsif ($t eq 'Bitmapset*' || $t eq 'Relids')
+        {
+            print $cff "\tCOPY_BITMAPSET_FIELD($f);\n" unless $copy_ignore;
+            print $eff "\tCOMPARE_BITMAPSET_FIELD($f);\n" unless $equal_ignore;
+        }
+        elsif ($t eq 'int' && $f =~ 'location$')
+        {
+            print $cff "\tCOPY_LOCATION_FIELD($f);\n" unless $copy_ignore;
+            print $eff "\tCOMPARE_LOCATION_FIELD($f);\n" unless $equal_ignore;
+        }
+        elsif (elem $t, @scalar_types or elem $t, @enum_types)
+        {
+            print $cff "\tCOPY_SCALAR_FIELD($f);\n" unless $copy_ignore;
+            if (elem 'equal_ignore_if_zero', @a)
+            {
+                print $eff "\tif (a->$f != b->$f && a->$f != 0 && b->$f != 0)\n\t\treturn false;\n";
+            }
+            else
+            {
+                print $eff "\tCOMPARE_SCALAR_FIELD($f);\n" unless $equal_ignore || $t eq 'CoercionForm';
+            }
+        }
+        # scalar type pointer
+        elsif ($t =~ /(\w+)\*/ and elem $1, @scalar_types)
+        {
+            my $tt = $1;
+            my $array_size_field;
+            foreach my $a (@a)
+            {
+                if ($a =~ /^array_size.([\w.]+)/)
+                {
+                    $array_size_field = $1;
+                    last;
+                }
+            }
+            if (!$array_size_field)
+            {
+                die "no array size defined for $n.$f of type $t";
+            }
+            if ($node_type_info{$n}->{field_types}{$array_size_field} eq 'List*')
+            {
+                print $cff "\tCOPY_POINTER_FIELD($f, list_length(from->$array_size_field) * sizeof($tt));\n" unless
$copy_ignore;
+                print $eff "\tCOMPARE_POINTER_FIELD($f, list_length(a->$array_size_field) * sizeof($tt));\n" unless
$equal_ignore;
+            }
+            else
+            {
+                print $cff "\tCOPY_POINTER_FIELD($f, from->$array_size_field * sizeof($tt));\n" unless $copy_ignore;
+                print $eff "\tCOMPARE_POINTER_FIELD($f, a->$array_size_field * sizeof($tt));\n" unless $equal_ignore;
+            }
+        }
+        # node type
+        elsif ($t =~ /(\w+)\*/ and elem $1, @node_types)
+        {
+            print $cff "\tCOPY_NODE_FIELD($f);\n" unless $copy_ignore;
+            print $eff "\tCOMPARE_NODE_FIELD($f);\n" unless $equal_ignore;
+        }
+        # array (inline)
+        elsif ($t =~ /\w+\[/)
+        {
+            print $cff "\tCOPY_ARRAY_FIELD($f);\n" unless $copy_ignore;
+            print $eff "\tCOMPARE_ARRAY_FIELD($f);\n" unless $equal_ignore;
+        }
+        elsif ($t eq 'struct CustomPathMethods*' ||    $t eq 'struct CustomScanMethods*')
+        {
+            # Fields of these types are required to be a pointer to a
+            # static table of callback functions.  So we don't copy
+            # the table itself, just reference the original one.
+            print $cff "\tCOPY_SCALAR_FIELD($f);\n" unless $copy_ignore;
+            print $eff "\tCOMPARE_SCALAR_FIELD($f);\n" unless $equal_ignore;
+        }
+        else
+        {
+            die "could not handle type \"$t\" in struct \"$n\" field \"$f\"";
+        }
+    }
+
+    print $cff "
+\treturn newnode;
+}
+";
+    print $eff "
+\treturn true;
+}
+";
+}
+
+close $cff;
+close $eff;
+close $cfs;
+close $efs;
+
+
+# outfuncs.c, readfuncs.c
+
+open my $off, '>', 'outfuncs.funcs.c' . $tmpext or die $!;
+open my $rff, '>', 'readfuncs.funcs.c' . $tmpext or die $!;
+open my $ofs, '>', 'outfuncs.switch.c' . $tmpext or die $!;
+open my $rfs, '>', 'readfuncs.switch.c' . $tmpext or die $!;
+
+print $off $node_includes;
+print $rff $node_includes;
+
+foreach my $n (@node_types)
+{
+    next if elem $n, @abstract_types;
+    next if elem $n, @no_read_write;
+
+    # XXX For now, skip all "Stmt"s except that ones that were there before.
+    if ($n =~ /Stmt$/)
+    {
+        my @keep = qw(AlterStatsStmt CreateForeignTableStmt CreateStatsStmt CreateStmt DeclareCursorStmt
ImportForeignSchemaStmtIndexStmt NotifyStmt PlannedStmt PLAssignStmt RawStmt ReturnStmt SelectStmt SetOperationStmt); 
+        next unless elem $n, @keep;
+    }
+
+    my $no_read = (elem $n, @no_read);
+
+    # output format starts with upper case node type, underscores stripped
+    my $N = uc $n;
+    $N =~ s/_//g;
+
+    print $ofs "\t\t\tcase T_${n}:\n".
+      "\t\t\t\t_out${n}(str, obj);\n".
+      "\t\t\t\tbreak;\n";
+
+    print $rfs "\telse if (MATCH(\"$N\", " . length($N) . "))\n".
+      "\t\treturn_value = _read${n}();\n" unless $no_read;
+
+    next if elem $n, @custom_read_write;
+
+    print $off "
+static void
+_out${n}(StringInfo str, const $n *node)
+{
+\tWRITE_NODE_TYPE(\"$N\");
+
+";
+
+    print $rff "
+static $n *
+_read${n}(void)
+{
+\tREAD_LOCALS($n);
+
+" unless $no_read;
+
+    # print instructions for each field
+    foreach my $f (@{$node_type_info{$n}->{fields}})
+    {
+        my $t = $node_type_info{$n}->{field_types}{$f};
+        my @a = @{ $node_type_info{$n}->{field_attrs}{$f} };
+        next if (elem 'read_write_ignore', @a);
+
+        # XXX Previously, for subtyping, only the leaf field name is
+        # used. Ponder whether we want to keep it that way.
+
+        # select instructions by field type
+        if ($t eq 'bool')
+        {
+            print $off "\tWRITE_BOOL_FIELD($f);\n";
+            print $rff "\tREAD_BOOL_FIELD($f);\n" unless $no_read;
+        }
+        elsif ($t eq 'int' && $f =~ 'location$')
+        {
+            print $off "\tWRITE_LOCATION_FIELD($f);\n";
+            print $rff "\tREAD_LOCATION_FIELD($f);\n" unless $no_read;
+        }
+        elsif ($t eq 'int' || $t eq 'int32' || $t eq 'AttrNumber' || $t eq 'StrategyNumber')
+        {
+            print $off "\tWRITE_INT_FIELD($f);\n";
+            print $rff "\tREAD_INT_FIELD($f);\n" unless $no_read;
+        }
+        elsif ($t eq 'uint32' || $t eq 'bits32' || $t eq 'AclMode' || $t eq 'BlockNumber' || $t eq 'Index' || $t eq
'SubTransactionId')
+        {
+            print $off "\tWRITE_UINT_FIELD($f);\n";
+            print $rff "\tREAD_UINT_FIELD($f);\n" unless $no_read;
+        }
+        elsif ($t eq 'uint64')
+        {
+            print $off "\tWRITE_UINT64_FIELD($f);\n";
+            print $rff "\tREAD_UINT64_FIELD($f);\n" unless $no_read;
+        }
+        elsif ($t eq 'Oid')
+        {
+            print $off "\tWRITE_OID_FIELD($f);\n";
+            print $rff "\tREAD_OID_FIELD($f);\n" unless $no_read;
+        }
+        elsif ($t eq 'long')
+        {
+            print $off "\tWRITE_LONG_FIELD($f);\n";
+            print $rff "\tREAD_LONG_FIELD($f);\n" unless $no_read;
+        }
+        elsif ($t eq 'char')
+        {
+            print $off "\tWRITE_CHAR_FIELD($f);\n";
+            print $rff "\tREAD_CHAR_FIELD($f);\n" unless $no_read;
+        }
+        elsif ($t eq 'double')
+        {
+            print $off "\tWRITE_FLOAT_FIELD($f, \"%.6f\");\n";
+            print $rff "\tREAD_FLOAT_FIELD($f);\n" unless $no_read;
+        }
+        elsif ($t eq 'Cardinality')
+        {
+            print $off "\tWRITE_FLOAT_FIELD($f, \"%.0f\");\n";
+            print $rff "\tREAD_FLOAT_FIELD($f);\n" unless $no_read;
+        }
+        elsif ($t eq 'Cost')
+        {
+            print $off "\tWRITE_FLOAT_FIELD($f, \"%.2f\");\n";
+            print $rff "\tREAD_FLOAT_FIELD($f);\n" unless $no_read;
+        }
+        elsif ($t eq 'QualCost')
+        {
+            print $off "\tWRITE_FLOAT_FIELD($f.startup, \"%.2f\");\n";
+            print $off "\tWRITE_FLOAT_FIELD($f.per_tuple, \"%.2f\");\n";
+            print $rff "\tREAD_FLOAT_FIELD($f.startup);\n" unless $no_read;
+            print $rff "\tREAD_FLOAT_FIELD($f.per_tuple);\n" unless $no_read;
+        }
+        elsif ($t eq 'Selectivity')
+        {
+            print $off "\tWRITE_FLOAT_FIELD($f, \"%.4f\");\n";
+            print $rff "\tREAD_FLOAT_FIELD($f);\n" unless $no_read;
+        }
+        elsif ($t eq 'char*')
+        {
+            print $off "\tWRITE_STRING_FIELD($f);\n";
+            print $rff "\tREAD_STRING_FIELD($f);\n" unless $no_read;
+        }
+        elsif ($t eq 'Bitmapset*' || $t eq 'Relids')
+        {
+            print $off "\tWRITE_BITMAPSET_FIELD($f);\n";
+            print $rff "\tREAD_BITMAPSET_FIELD($f);\n" unless $no_read;
+        }
+        elsif (elem $t, @enum_types)
+        {
+            print $off "\tWRITE_ENUM_FIELD($f, $t);\n";
+            print $rff "\tREAD_ENUM_FIELD($f, $t);\n" unless $no_read;
+        }
+        # arrays
+        elsif ($t =~ /(\w+)(\*|\[)/ and elem $1, @scalar_types)
+        {
+            my $tt = uc $1;
+            my $array_size_field;
+            foreach my $a (@a)
+            {
+                if ($a =~ /^array_size.([\w.]+)/)
+                {
+                    $array_size_field = $1;
+                    last;
+                }
+            }
+            if (!$array_size_field)
+            {
+                die "no array size defined for $n.$f of type $t";
+            }
+            if ($node_type_info{$n}->{field_types}{$array_size_field} eq 'List*')
+            {
+                print $off "\tWRITE_${tt}_ARRAY($f, list_length(node->$array_size_field));\n";
+                print $rff "\tREAD_${tt}_ARRAY($f, list_length(local_node->$array_size_field));\n" unless $no_read;
+            }
+            else
+            {
+                print $off "\tWRITE_${tt}_ARRAY($f, node->$array_size_field);\n";
+                print $rff "\tREAD_${tt}_ARRAY($f, local_node->$array_size_field);\n" unless $no_read;
+            }
+        }
+        # Special treatments of several Path node fields
+        elsif ($t eq 'RelOptInfo*' && elem 'write_only_relids', @a)
+        {
+            print $off "\tappendStringInfoString(str, \" :parent_relids \");\n".
+              "\toutBitmapset(str, node->$f->relids);\n";
+        }
+        elsif ($t eq 'PathTarget*' && elem 'write_only_nondefault_pathtarget', @a)
+        {
+            (my $f2 = $f) =~ s/pathtarget/parent/;
+            print $off "\tif (node->$f != node->$f2->reltarget)\n".
+              "\t\tWRITE_NODE_FIELD($f);\n";
+        }
+        elsif ($t eq 'ParamPathInfo*' && elem 'write_only_req_outer', @a)
+        {
+            print $off "\tif (node->$f)\n".
+              "\t\toutBitmapset(str, node->$f->ppi_req_outer);\n".
+              "\telse\n".
+              "\t\toutBitmapset(str, NULL);\n";
+        }
+        # node type
+        elsif ($t =~ /(\w+)\*/ and elem $1, @node_types)
+        {
+            print $off "\tWRITE_NODE_FIELD($f);\n";
+            print $rff "\tREAD_NODE_FIELD($f);\n" unless $no_read;
+        }
+        elsif ($t eq 'struct CustomPathMethods*' ||    $t eq 'struct CustomScanMethods*')
+        {
+            print $off q{
+    /* CustomName is a key to lookup CustomScanMethods */
+    appendStringInfoString(str, " :methods ");
+    outToken(str, node->methods->CustomName);
+};
+            print $rff q!
+    {
+        /* Lookup CustomScanMethods by CustomName */
+        char       *custom_name;
+        const CustomScanMethods *methods;
+        token = pg_strtok(&length); /* skip methods: */
+        token = pg_strtok(&length); /* CustomName */
+        custom_name = nullable_string(token, length);
+        methods = GetCustomScanMethods(custom_name, false);
+        local_node->methods = methods;
+    }
+! unless $no_read;
+        }
+        else
+        {
+            die "could not handle type \"$t\" in struct \"$n\" field \"$f\"";
+        }
+    }
+
+    print $off "}
+";
+    print $rff "
+\tREAD_DONE();
+}
+" unless $no_read;
+}
+
+close $off;
+close $rff;
+close $ofs;
+close $rfs;
+
+
+# now rename the temporary files to their final name
+foreach my $file (qw(nodetags.h copyfuncs.funcs.c copyfuncs.switch.c equalfuncs.funcs.c equalfuncs.switch.c
outfuncs.funcs.coutfuncs.switch.c readfuncs.funcs.c readfuncs.switch.c)) 
+{
+    Catalog::RenameTempFile($file, $tmpext);
+}
diff --git a/src/backend/nodes/outfuncs.c b/src/backend/nodes/outfuncs.c
index 4315c53080..37508af94d 100644
--- a/src/backend/nodes/outfuncs.c
+++ b/src/backend/nodes/outfuncs.c
@@ -31,11 +31,10 @@

 #include "lib/stringinfo.h"
 #include "miscadmin.h"
-#include "nodes/extensible.h"
-#include "nodes/pathnodes.h"
-#include "nodes/plannodes.h"
+#include "nodes/bitmapset.h"
+#include "nodes/nodes.h"
+#include "nodes/pg_list.h"
 #include "utils/datum.h"
-#include "utils/rel.h"

 static void outChar(StringInfo str, char c);

@@ -306,6 +305,9 @@ outDatum(StringInfo str, Datum value, int typlen, bool typbyval)
 }


+#include "outfuncs.funcs.c"
+
+#ifdef OBSOLETE
 /*
  *    Stuff from plannodes.h
  */
@@ -1155,6 +1157,7 @@ _outVar(StringInfo str, const Var *node)
     WRITE_INT_FIELD(varattnosyn);
     WRITE_LOCATION_FIELD(location);
 }
+#endif /*OBSOLETE*/

 static void
 _outConst(StringInfo str, const Const *node)
@@ -1176,6 +1179,7 @@ _outConst(StringInfo str, const Const *node)
         outDatum(str, node->constvalue, node->constlen, node->constbyval);
 }

+#ifdef OBSOLETE
 static void
 _outParam(StringInfo str, const Param *node)
 {
@@ -1346,6 +1350,7 @@ _outScalarArrayOpExpr(StringInfo str, const ScalarArrayOpExpr *node)
     WRITE_NODE_FIELD(args);
     WRITE_LOCATION_FIELD(location);
 }
+#endif /*OBSOLETE*/

 static void
 _outBoolExpr(StringInfo str, const BoolExpr *node)
@@ -1374,6 +1379,7 @@ _outBoolExpr(StringInfo str, const BoolExpr *node)
     WRITE_LOCATION_FIELD(location);
 }

+#ifdef OBSOLETE
 static void
 _outSubLink(StringInfo str, const SubLink *node)
 {
@@ -2586,6 +2592,7 @@ _outIndexOptInfo(StringInfo str, const IndexOptInfo *node)
     WRITE_BOOL_FIELD(hypothetical);
     /* we don't bother with fields copied from the index AM's API struct */
 }
+#endif /* OBSOLETE */

 static void
 _outForeignKeyOptInfo(StringInfo str, const ForeignKeyOptInfo *node)
@@ -2613,6 +2620,7 @@ _outForeignKeyOptInfo(StringInfo str, const ForeignKeyOptInfo *node)
         appendStringInfo(str, " %d", list_length(node->rinfos[i]));
 }

+#ifdef OBSOLETE
 static void
 _outStatisticExtInfo(StringInfo str, const StatisticExtInfo *node)
 {
@@ -2624,6 +2632,7 @@ _outStatisticExtInfo(StringInfo str, const StatisticExtInfo *node)
     WRITE_CHAR_FIELD(kind);
     WRITE_BITMAPSET_FIELD(keys);
 }
+#endif /* OBSOLETE */

 static void
 _outEquivalenceClass(StringInfo str, const EquivalenceClass *node)
@@ -2652,6 +2661,7 @@ _outEquivalenceClass(StringInfo str, const EquivalenceClass *node)
     WRITE_UINT_FIELD(ec_max_security);
 }

+#ifdef OBSOLETE
 static void
 _outEquivalenceMember(StringInfo str, const EquivalenceMember *node)
 {
@@ -2836,6 +2846,7 @@ _outPlannerParamItem(StringInfo str, const PlannerParamItem *node)
     WRITE_NODE_FIELD(item);
     WRITE_INT_FIELD(paramId);
 }
+#endif /*OBSOLETE*/

 /*****************************************************************************
  *
@@ -2858,6 +2869,7 @@ _outExtensibleNode(StringInfo str, const ExtensibleNode *node)
     methods->nodeOut(str, node);
 }

+#ifdef OBSOLETE
 /*****************************************************************************
  *
  *    Stuff from parsenodes.h.
@@ -3191,6 +3203,7 @@ _outStatsElem(StringInfo str, const StatsElem *node)
     WRITE_STRING_FIELD(name);
     WRITE_NODE_FIELD(expr);
 }
+#endif /*OBSOLETE*/

 static void
 _outQuery(StringInfo str, const Query *node)
@@ -3265,6 +3278,7 @@ _outQuery(StringInfo str, const Query *node)
     WRITE_INT_FIELD(stmt_len);
 }

+#ifdef OBSOLETE
 static void
 _outWithCheckOption(StringInfo str, const WithCheckOption *node)
 {
@@ -3430,6 +3444,7 @@ _outSetOperationStmt(StringInfo str, const SetOperationStmt *node)
     WRITE_NODE_FIELD(colCollations);
     WRITE_NODE_FIELD(groupClauses);
 }
+#endif /*OBSOLETE*/

 static void
 _outRangeTblEntry(StringInfo str, const RangeTblEntry *node)
@@ -3510,6 +3525,7 @@ _outRangeTblEntry(StringInfo str, const RangeTblEntry *node)
     WRITE_NODE_FIELD(securityQuals);
 }

+#ifdef OBSOLETE
 static void
 _outRangeTblFunction(StringInfo str, const RangeTblFunction *node)
 {
@@ -3533,6 +3549,7 @@ _outTableSampleClause(StringInfo str, const TableSampleClause *node)
     WRITE_NODE_FIELD(args);
     WRITE_NODE_FIELD(repeatable);
 }
+#endif /*OBSOLETE*/

 static void
 _outA_Expr(StringInfo str, const A_Expr *node)
@@ -3651,6 +3668,7 @@ _outBitString(StringInfo str, const BitString *node)
     appendStringInfoString(str, node->bsval);
 }

+#ifdef OBSOLETE
 static void
 _outColumnRef(StringInfo str, const ColumnRef *node)
 {
@@ -3682,6 +3700,7 @@ _outRawStmt(StringInfo str, const RawStmt *node)
     WRITE_LOCATION_FIELD(stmt_location);
     WRITE_INT_FIELD(stmt_len);
 }
+#endif /*OBSOLETE*/

 static void
 _outA_Const(StringInfo str, const A_Const *node)
@@ -3698,6 +3717,7 @@ _outA_Const(StringInfo str, const A_Const *node)
     WRITE_LOCATION_FIELD(location);
 }

+#ifdef OBSOLETE
 static void
 _outA_Star(StringInfo str, const A_Star *node)
 {
@@ -3842,6 +3862,7 @@ _outRangeTableFuncCol(StringInfo str, const RangeTableFuncCol *node)
     WRITE_NODE_FIELD(coldefexpr);
     WRITE_LOCATION_FIELD(location);
 }
+#endif /*OBSOLETE*/

 static void
 _outConstraint(StringInfo str, const Constraint *node)
@@ -3964,6 +3985,7 @@ _outConstraint(StringInfo str, const Constraint *node)
     }
 }

+#ifdef OBSOLETE
 static void
 _outForeignKeyCacheInfo(StringInfo str, const ForeignKeyCacheInfo *node)
 {
@@ -4024,6 +4046,7 @@ _outPartitionRangeDatum(StringInfo str, const PartitionRangeDatum *node)
     WRITE_NODE_FIELD(value);
     WRITE_LOCATION_FIELD(location);
 }
+#endif /*OBSOLETE*/

 /*
  * outNode -
@@ -4055,6 +4078,8 @@ outNode(StringInfo str, const void *obj)
         appendStringInfoChar(str, '{');
         switch (nodeTag(obj))
         {
+#include "outfuncs.switch.c"
+#ifdef OBSOLETE
             case T_PlannedStmt:
                 _outPlannedStmt(str, obj);
                 break;
@@ -4766,6 +4791,7 @@ outNode(StringInfo str, const void *obj)
             case T_JsonTableSibling:
                 _outJsonTableSibling(str, obj);
                 break;
+#endif /*OBSOLETE*/

             default:

diff --git a/src/backend/nodes/readfuncs.c b/src/backend/nodes/readfuncs.c
index 6a05b69415..f427aa05ec 100644
--- a/src/backend/nodes/readfuncs.c
+++ b/src/backend/nodes/readfuncs.c
@@ -33,9 +33,7 @@
 #include <math.h>

 #include "miscadmin.h"
-#include "nodes/extensible.h"
-#include "nodes/parsenodes.h"
-#include "nodes/plannodes.h"
+#include "nodes/bitmapset.h"
 #include "nodes/readfuncs.h"


@@ -238,6 +236,8 @@ readBitmapset(void)
     return _readBitmapset();
 }

+#include "readfuncs.funcs.c"
+
 /*
  * _readQuery
  */
@@ -291,6 +291,7 @@ _readQuery(void)
     READ_DONE();
 }

+#ifdef OBSOLETE
 /*
  * _readNotifyStmt
  */
@@ -629,6 +630,7 @@ _readVar(void)

     READ_DONE();
 }
+#endif /*OBSOLETE*/

 /*
  * _readConst
@@ -655,6 +657,7 @@ _readConst(void)
     READ_DONE();
 }

+#ifdef OBSOLETE
 /*
  * _readParam
  */
@@ -880,6 +883,7 @@ _readScalarArrayOpExpr(void)

     READ_DONE();
 }
+#endif /*OBSOLETE*/

 /*
  * _readBoolExpr
@@ -907,6 +911,7 @@ _readBoolExpr(void)
     READ_DONE();
 }

+#ifdef OBSOLETE
 /*
  * _readSubLink
  */
@@ -1649,6 +1654,7 @@ _readAppendRelInfo(void)
 /*
  *    Stuff from parsenodes.h.
  */
+#endif /*OBSOLETE*/

 /*
  * _readRangeTblEntry
@@ -1744,6 +1750,7 @@ _readRangeTblEntry(void)
     READ_DONE();
 }

+#ifdef OBSOLETE
 /*
  * _readRangeTblFunction
  */
@@ -2872,6 +2879,7 @@ _readAlternativeSubPlan(void)

     READ_DONE();
 }
+#endif /*OBSOLETE*/

 /*
  * _readExtensibleNode
@@ -2903,6 +2911,7 @@ _readExtensibleNode(void)
     READ_DONE();
 }

+#ifdef OBSOLETE
 /*
  * _readPartitionBoundSpec
  */
@@ -2937,6 +2946,7 @@ _readPartitionRangeDatum(void)

     READ_DONE();
 }
+#endif /*OBSOLETE*/

 /*
  * parseNodeString
@@ -2961,7 +2971,11 @@ parseNodeString(void)
 #define MATCH(tokname, namelen) \
     (length == namelen && memcmp(token, tokname, namelen) == 0)

-    if (MATCH("QUERY", 5))
+    if (false)
+        ;
+#include "readfuncs.switch.c"
+#ifdef OBSOLETE
+    else if (MATCH("QUERY", 5))
         return_value = _readQuery();
     else if (MATCH("WITHCHECKOPTION", 15))
         return_value = _readWithCheckOption();
@@ -3235,6 +3249,7 @@ parseNodeString(void)
         return_value = _readJsonTableParent();
     else if (MATCH("JSONTABLESIBLING", 16))
         return_value = _readJsonTableSibling();
+#endif /*OBSOLETE*/
     else
     {
         elog(ERROR, "badly formatted node string \"%.32s\"...", token);
diff --git a/src/include/Makefile b/src/include/Makefile
index 5f257a958c..17cfd268b8 100644
--- a/src/include/Makefile
+++ b/src/include/Makefile
@@ -81,6 +81,7 @@ clean:
     rm -f parser/gram.h storage/lwlocknames.h utils/probes.h
     rm -f catalog/schemapg.h catalog/system_fk_info.h
     rm -f catalog/pg_*_d.h catalog/header-stamp
+    rm -f nodes/nodetags.h nodes/header-stamp

 distclean maintainer-clean: clean
     rm -f pg_config.h pg_config_ext.h pg_config_os.h stamp-h stamp-ext-h
diff --git a/src/include/executor/tuptable.h b/src/include/executor/tuptable.h
index 6306bb6fc6..37c11522ee 100644
--- a/src/include/executor/tuptable.h
+++ b/src/include/executor/tuptable.h
@@ -235,14 +235,14 @@ extern PGDLLIMPORT const TupleTableSlotOps TTSOpsBufferHeapTuple;
  * Tuple table slot implementations.
  */

-typedef struct VirtualTupleTableSlot
+typedef struct VirtualTupleTableSlot pg_node_attr(abstract)
 {
     TupleTableSlot base;

     char       *data;            /* data for materialized slots */
 } VirtualTupleTableSlot;

-typedef struct HeapTupleTableSlot
+typedef struct HeapTupleTableSlot pg_node_attr(abstract)
 {
     TupleTableSlot base;

@@ -254,7 +254,7 @@ typedef struct HeapTupleTableSlot
 } HeapTupleTableSlot;

 /* heap tuple residing in a buffer */
-typedef struct BufferHeapTupleTableSlot
+typedef struct BufferHeapTupleTableSlot pg_node_attr(abstract)
 {
     HeapTupleTableSlot base;

@@ -267,7 +267,7 @@ typedef struct BufferHeapTupleTableSlot
     Buffer        buffer;            /* tuple's buffer, or InvalidBuffer */
 } BufferHeapTupleTableSlot;

-typedef struct MinimalTupleTableSlot
+typedef struct MinimalTupleTableSlot pg_node_attr(abstract)
 {
     TupleTableSlot base;

diff --git a/src/include/nodes/.gitignore b/src/include/nodes/.gitignore
new file mode 100644
index 0000000000..99fb1d3787
--- /dev/null
+++ b/src/include/nodes/.gitignore
@@ -0,0 +1,2 @@
+/nodetags.h
+/header-stamp
diff --git a/src/include/nodes/extensible.h b/src/include/nodes/extensible.h
index 6244c8d961..fab5bf690b 100644
--- a/src/include/nodes/extensible.h
+++ b/src/include/nodes/extensible.h
@@ -29,7 +29,7 @@
  * specific type of node.  extnodename can be looked up to find the
  * ExtensibleNodeMethods for this node type.
  */
-typedef struct ExtensibleNode
+typedef struct ExtensibleNode pg_node_attr(custom_copy_equal, custom_read_write)
 {
     NodeTag        type;
     const char *extnodename;    /* identifier of ExtensibleNodeMethods */
diff --git a/src/include/nodes/nodes.h b/src/include/nodes/nodes.h
index 7ce1fc4deb..d77aad6473 100644
--- a/src/include/nodes/nodes.h
+++ b/src/include/nodes/nodes.h
@@ -27,6 +27,8 @@ typedef enum NodeTag
 {
     T_Invalid = 0,

+#include "nodes/nodetags.h"
+#ifdef OBSOLETE
     /*
      * TAGS FOR EXECUTOR NODES (execnodes.h)
      */
@@ -563,8 +565,58 @@ typedef enum NodeTag
     T_SupportRequestRows,        /* in nodes/supportnodes.h */
     T_SupportRequestIndexCondition, /* in nodes/supportnodes.h */
     T_SupportRequestWFuncMonotonic    /* in nodes/supportnodes.h */
+#endif /*OBSOLETE*/
 } NodeTag;

+/*
+ * pg_node_attr() - Used in node definitions to set extra information for
+ * gen_node_support.pl
+ *
+ * Attributes can be attached to a node as a whole (the attribute
+ * specification must be on the same line as "struct") or to a specific field
+ * (must be at the end of the line).  The argument is a comma-separated list
+ * of attributes.  Unrecognized attributes cause an error.
+ *
+ * Valid node attributes:
+ *
+ * - abstract: Abstract types are types that cannot be instantiated but that
+ *   can be supertypes of other types.  We track their fields, so that
+ *   subtypes can use them, but we don't emit a node tag, so you can't
+ *   instantiate them.
+ *
+ * - custom_copy_equal: Has custom implementations in copyfuncs.c and
+ *   equalfuncs.c.
+ *
+ * - custom_read_write: Has custom implementations in outfuncs.c and
+ *   readfuncs.c.
+ *
+ * - no_copy_equal: Does not support copyObject() and equal() at all.
+ *
+ * - no_read: Does not support nodeRead() at all.
+ *
+ * - special_read_write: Has special treatment in outNode() and nodeRead().
+ *
+ * Valid node field attributes:
+ *
+ * - array_size(OTHERFIELD): This field is a dynamically allocated array with
+ *   size indicated by the mentioned other field.  The other field is either a
+ *   scalar or a list, in which case the length of the list is used.
+ *
+ * - copy_ignore: Ignore the field for copy.
+ *
+ * - equal_ignore: Ignore the field for equality.
+ *
+ * - equal_ignore_if_zero: Ignore the field for equality if it is zero.
+ *   (Otherwise, compare normally.)
+ *
+ * - read_write_ignore: Ignore the field for read/write.
+ *
+ * - write_only_relids, write_only_nondefault_pathtarget, write_only_req_outer:
+ *   Special handling for Path struct; see there.
+ *
+ */
+#define pg_node_attr(...)
+
 /*
  * The first field of a node of any type is guaranteed to be the NodeTag.
  * Hence the type of any node can be gotten by casting it to Node. Declaring
diff --git a/src/include/nodes/parsenodes.h b/src/include/nodes/parsenodes.h
index f93d866548..fb026c6b1f 100644
--- a/src/include/nodes/parsenodes.h
+++ b/src/include/nodes/parsenodes.h
@@ -115,7 +115,7 @@ typedef uint32 AclMode;            /* a bitmask of privilege bits */
  *      Planning converts a Query tree into a Plan tree headed by a PlannedStmt
  *      node --- the Query structure is not used by the executor.
  */
-typedef struct Query
+typedef struct Query pg_node_attr(custom_read_write)
 {
     NodeTag        type;

@@ -123,8 +123,11 @@ typedef struct Query

     QuerySource querySource;    /* where did I come from? */

-    /* query identifier (can be set by plugins) */
-    uint64        queryId;
+    /*
+     * query identifier (can be set by plugins); ignored for equal, might not
+     * be set
+     */
+    uint64        queryId pg_node_attr(equal_ignore);

     bool        canSetTag;        /* do I set the command result tag? */

@@ -286,7 +289,7 @@ typedef enum A_Expr_Kind
     AEXPR_NOT_BETWEEN_SYM        /* name must be "NOT BETWEEN SYMMETRIC" */
 } A_Expr_Kind;

-typedef struct A_Expr
+typedef struct A_Expr pg_node_attr(custom_read_write, no_read)
 {
     NodeTag        type;
     A_Expr_Kind kind;            /* see above */
@@ -299,7 +302,7 @@ typedef struct A_Expr
 /*
  * A_Const - a literal constant
  */
-typedef struct A_Const
+typedef struct A_Const pg_node_attr(custom_copy_equal, custom_read_write, no_read)
 {
     NodeTag        type;

@@ -398,7 +401,7 @@ typedef struct FuncCall
  * This can appear within ColumnRef.fields, A_Indirection.indirection, and
  * ResTarget.indirection lists.
  */
-typedef struct A_Star
+typedef struct A_Star pg_node_attr(no_read)
 {
     NodeTag        type;
 } A_Star;
@@ -1010,7 +1013,7 @@ typedef enum RTEKind
                                  * present during parsing or rewriting */
 } RTEKind;

-typedef struct RangeTblEntry
+typedef struct RangeTblEntry pg_node_attr(custom_read_write)
 {
     NodeTag        type;

@@ -2606,7 +2609,7 @@ typedef enum ConstrType            /* types of constraints */
 #define FKCONSTR_MATCH_PARTIAL        'p'
 #define FKCONSTR_MATCH_SIMPLE        's'

-typedef struct Constraint
+typedef struct Constraint pg_node_attr(custom_read_write, no_read)
 {
     NodeTag        type;
     ConstrType    contype;        /* see above */
diff --git a/src/include/nodes/pathnodes.h b/src/include/nodes/pathnodes.h
index b88cfb8dc0..4212610d5e 100644
--- a/src/include/nodes/pathnodes.h
+++ b/src/include/nodes/pathnodes.h
@@ -3,6 +3,8 @@
  * pathnodes.h
  *      Definitions for planner's internal data structures, especially Paths.
  *
+ * We don't support copying RelOptInfo, IndexOptInfo, or Path nodes.
+ * There are some subsidiary structs that are useful to copy, though.
  *
  * Portions Copyright (c) 1996-2022, PostgreSQL Global Development Group
  * Portions Copyright (c) 1994, Regents of the University of California
@@ -85,17 +87,20 @@ typedef enum UpperRelationKind
  * PlannerGlobal holds state for an entire planner invocation; this state
  * is shared across all levels of sub-Queries that exist in the command being
  * planned.
+ *
+ * Not all fields are printed.  (In some cases, there is no print support for
+ * the field type.)
  *----------
  */
-typedef struct PlannerGlobal
+typedef struct PlannerGlobal pg_node_attr(no_copy_equal)
 {
     NodeTag        type;

-    ParamListInfo boundParams;    /* Param values provided to planner() */
+    ParamListInfo boundParams pg_node_attr(read_write_ignore);    /* Param values provided to planner() */

     List       *subplans;        /* Plans for SubPlan nodes */

-    List       *subroots;        /* PlannerInfos for SubPlan nodes */
+    List       *subroots pg_node_attr(read_write_ignore);        /* PlannerInfos for SubPlan nodes */

     Bitmapset  *rewindPlanIDs;    /* indices of subplans that require REWIND */

@@ -129,7 +134,7 @@ typedef struct PlannerGlobal

     char        maxParallelHazard;    /* worst PROPARALLEL hazard level */

-    PartitionDirectory partition_directory; /* partition descriptors */
+    PartitionDirectory partition_directory pg_node_attr(read_write_ignore); /* partition descriptors */
 } PlannerGlobal;

 /* macro for fetching the Plan associated with a SubPlan node */
@@ -148,6 +153,9 @@ typedef struct PlannerGlobal
  *
  * For reasons explained in optimizer/optimizer.h, we define the typedef
  * either here or in that header, whichever is read first.
+ *
+ * Not all fields are printed.  (In some cases, there is no print support for
+ * the field type.)
  *----------
  */
 #ifndef HAVE_PLANNERINFO_TYPEDEF
@@ -155,7 +163,7 @@ typedef struct PlannerInfo PlannerInfo;
 #define HAVE_PLANNERINFO_TYPEDEF 1
 #endif

-struct PlannerInfo
+struct PlannerInfo pg_node_attr(no_copy_equal)
 {
     NodeTag        type;

@@ -165,7 +173,7 @@ struct PlannerInfo

     Index        query_level;    /* 1 at the outermost Query */

-    PlannerInfo *parent_root;    /* NULL at outermost Query */
+    PlannerInfo *parent_root pg_node_attr(read_write_ignore);    /* NULL at outermost Query */

     /*
      * plan_params contains the expressions that this query level needs to
@@ -183,15 +191,15 @@ struct PlannerInfo
      * does not correspond to a base relation, such as a join RTE or an
      * unreferenced view RTE; or if the RelOptInfo hasn't been made yet.
      */
-    struct RelOptInfo **simple_rel_array;    /* All 1-rel RelOptInfos */
-    int            simple_rel_array_size;    /* allocated size of array */
+    struct RelOptInfo **simple_rel_array pg_node_attr(read_write_ignore);    /* All 1-rel RelOptInfos */
+    int            simple_rel_array_size pg_node_attr(read_write_ignore);    /* allocated size of array */

     /*
      * simple_rte_array is the same length as simple_rel_array and holds
      * pointers to the associated rangetable entries.  Using this is a shade
      * faster than using rt_fetch(), mostly due to fewer indirections.
      */
-    RangeTblEntry **simple_rte_array;    /* rangetable as an array */
+    RangeTblEntry **simple_rte_array pg_node_attr(read_write_ignore);    /* rangetable as an array */

     /*
      * append_rel_array is the same length as the above arrays, and holds
@@ -199,7 +207,7 @@ struct PlannerInfo
      * child_relid, or NULL if the rel is not an appendrel child.  The array
      * itself is not allocated if append_rel_list is empty.
      */
-    struct AppendRelInfo **append_rel_array;
+    struct AppendRelInfo **append_rel_array pg_node_attr(read_write_ignore);

     /*
      * all_baserels is a Relids set of all base relids (but not "other"
@@ -227,7 +235,7 @@ struct PlannerInfo
      * GEQO.
      */
     List       *join_rel_list;
-    struct HTAB *join_rel_hash;
+    struct HTAB *join_rel_hash pg_node_attr(read_write_ignore);

     /*
      * When doing a dynamic-programming-style join search, join_rel_level[k]
@@ -236,7 +244,7 @@ struct PlannerInfo
      * automatically added to the join_rel_level[join_cur_level] list.
      * join_rel_level is NULL if not in use.
      */
-    List      **join_rel_level; /* lists of join-relation RelOptInfos */
+    List      **join_rel_level pg_node_attr(read_write_ignore); /* lists of join-relation RelOptInfos */
     int            join_cur_level; /* index of list being extended */

     List       *init_plans;        /* init SubPlans for query */
@@ -299,16 +307,16 @@ struct PlannerInfo
     List       *distinct_pathkeys;    /* distinctClause pathkeys, if any */
     List       *sort_pathkeys;    /* sortClause pathkeys, if any */

-    List       *part_schemes;    /* Canonicalised partition schemes used in the
+    List       *part_schemes pg_node_attr(read_write_ignore);    /* Canonicalised partition schemes used in the
                                  * query. */

-    List       *initial_rels;    /* RelOptInfos we are now trying to join */
+    List       *initial_rels pg_node_attr(read_write_ignore);    /* RelOptInfos we are now trying to join */

     /* Use fetch_upper_rel() to get any particular upper rel */
-    List       *upper_rels[UPPERREL_FINAL + 1]; /* upper-rel RelOptInfos */
+    List       *upper_rels[UPPERREL_FINAL + 1] pg_node_attr(read_write_ignore); /* upper-rel RelOptInfos */

     /* Result tlists chosen by grouping_planner for upper-stage processing */
-    struct PathTarget *upper_targets[UPPERREL_FINAL + 1];
+    struct PathTarget *upper_targets[UPPERREL_FINAL + 1] pg_node_attr(read_write_ignore);

     /*
      * The fully-processed targetlist is kept here.  It differs from
@@ -333,12 +341,12 @@ struct PlannerInfo
      * Fields filled during create_plan() for use in setrefs.c
      */
     /* for GroupingFunc fixup */
-    AttrNumber *grouping_map;
+    AttrNumber *grouping_map pg_node_attr(array_size(update_colnos), read_write_ignore);
     /* List of MinMaxAggInfos */
     List       *minmax_aggs;

     /* context holding PlannerInfo */
-    MemoryContext planner_cxt;
+    MemoryContext planner_cxt pg_node_attr(read_write_ignore);

     Cardinality total_table_pages;    /* # of pages in all non-dummy tables of
                                      * query */
@@ -360,11 +368,11 @@ struct PlannerInfo
     /*
      * Information about aggregates. Filled by preprocess_aggrefs().
      */
-    List       *agginfos;        /* AggInfo structs */
-    List       *aggtransinfos;    /* AggTransInfo structs */
-    int            numOrderedAggs; /* number w/ DISTINCT/ORDER BY/WITHIN GROUP */
-    bool        hasNonPartialAggs;    /* does any agg not support partial mode? */
-    bool        hasNonSerialAggs;    /* is any partial agg non-serializable? */
+    List       *agginfos pg_node_attr(read_write_ignore);        /* AggInfo structs */
+    List       *aggtransinfos pg_node_attr(read_write_ignore);    /* AggTransInfo structs */
+    int            numOrderedAggs pg_node_attr(read_write_ignore); /* number w/ DISTINCT/ORDER BY/WITHIN GROUP */
+    bool        hasNonPartialAggs pg_node_attr(read_write_ignore);    /* does any agg not support partial mode? */
+    bool        hasNonSerialAggs pg_node_attr(read_write_ignore);    /* is any partial agg non-serializable? */

     /* These fields are used only when hasRecursion is true: */
     int            wt_param_id;    /* PARAM_EXEC ID for the work table */
@@ -378,11 +386,11 @@ struct PlannerInfo
      * These fields are workspace for setrefs.c.  Each is an array
      * corresponding to glob->subplans.
      */
-    bool       *isAltSubplan;
-    bool       *isUsedSubplan;
+    bool       *isAltSubplan pg_node_attr(read_write_ignore);
+    bool       *isUsedSubplan pg_node_attr(read_write_ignore);

     /* optional private data for join_search_hook, e.g., GEQO */
-    void       *join_search_private;
+    void       *join_search_private pg_node_attr(read_write_ignore);

     /* Does this query modify any partition key columns? */
     bool        partColsUpdated;
@@ -639,6 +647,9 @@ typedef struct PartitionSchemeData *PartitionScheme;
  * Furthermore, FULL JOINs add extra nullable_partexprs expressions
  * corresponding to COALESCE expressions of the left and right join columns,
  * to simplify matching join clauses to those lists.
+ *
+ * Not all fields are printed.  (In some cases, there is no print support for
+ * the field type.)
  *----------
  */

@@ -680,7 +691,7 @@ typedef enum RelOptKind
      (rel)->reloptkind == RELOPT_OTHER_JOINREL || \
      (rel)->reloptkind == RELOPT_OTHER_UPPER_REL)

-typedef struct RelOptInfo
+typedef struct RelOptInfo pg_node_attr(no_copy_equal)
 {
     NodeTag        type;

@@ -747,9 +758,9 @@ typedef struct RelOptInfo
     /* largest attrno of rel */
     AttrNumber    max_attr;
     /* array indexed [min_attr .. max_attr] */
-    Relids       *attr_needed;
+    Relids       *attr_needed pg_node_attr(read_write_ignore);
     /* array indexed [min_attr .. max_attr] */
-    int32       *attr_widths;
+    int32       *attr_widths pg_node_attr(read_write_ignore);
     /* LATERAL Vars and PHVs referenced by rel */
     List       *lateral_vars;
     /* rels that reference me laterally */
@@ -784,16 +795,18 @@ typedef struct RelOptInfo
     /* join is only valid for current user */
     bool        useridiscurrent;
     /* use "struct FdwRoutine" to avoid including fdwapi.h here */
-    struct FdwRoutine *fdwroutine;
-    void       *fdw_private;
+    struct FdwRoutine *fdwroutine pg_node_attr(read_write_ignore);
+    void       *fdw_private pg_node_attr(read_write_ignore);

     /*
      * cache space for remembering if we have proven this relation unique
+     *
+     * can't print unique_for_rels/non_unique_for_rels; BMSes aren't Nodes
      */
     /* known unique for these other relid set(s) */
-    List       *unique_for_rels;
+    List       *unique_for_rels pg_node_attr(read_write_ignore);
     /* known not unique for these set(s) */
-    List       *non_unique_for_rels;
+    List       *non_unique_for_rels pg_node_attr(read_write_ignore);

     /*
      * used by various scans and joins:
@@ -821,24 +834,24 @@ typedef struct RelOptInfo
      * used for partitioned relations:
      */
     /* Partitioning scheme */
-    PartitionScheme part_scheme;
+    PartitionScheme part_scheme pg_node_attr(read_write_ignore);

     /*
      * Number of partitions; -1 if not yet set; in case of a join relation 0
      * means it's considered unpartitioned
      */
-    int            nparts;
+    int            nparts pg_node_attr(read_write_ignore);
     /* Partition bounds */
-    struct PartitionBoundInfoData *boundinfo;
+    struct PartitionBoundInfoData *boundinfo pg_node_attr(read_write_ignore);
     /* True if partition bounds were created by partition_bounds_merge() */
     bool        partbounds_merged;
     /* Partition constraint, if not the root */
-    List       *partition_qual;
+    List       *partition_qual pg_node_attr(read_write_ignore);

     /*
      * Array of RelOptInfos of partitions, stored in the same order as bounds
      */
-    struct RelOptInfo **part_rels;
+    struct RelOptInfo **part_rels pg_node_attr(read_write_ignore);

     /*
      * Bitmap with members acting as indexes into the part_rels[] array to
@@ -848,9 +861,9 @@ typedef struct RelOptInfo
     /* Relids set of all partition relids */
     Relids        all_partrels;
     /* Non-nullable partition key expressions */
-    List      **partexprs;
+    List      **partexprs pg_node_attr(read_write_ignore);
     /* Nullable partition key expressions */
-    List      **nullable_partexprs;
+    List      **nullable_partexprs pg_node_attr(read_write_ignore);
 } RelOptInfo;

 /*
@@ -909,7 +922,7 @@ typedef struct IndexOptInfo IndexOptInfo;
 #define HAVE_INDEXOPTINFO_TYPEDEF 1
 #endif

-struct IndexOptInfo
+struct IndexOptInfo pg_node_attr(no_copy_equal)
 {
     NodeTag        type;

@@ -917,8 +930,8 @@ struct IndexOptInfo
     Oid            indexoid;
     /* tablespace of index (not table) */
     Oid            reltablespace;
-    /* back-link to index's table */
-    RelOptInfo *rel;
+    /* back-link to index's table; don't print, else infinite recursion */
+    RelOptInfo *rel pg_node_attr(read_write_ignore);

     /*
      * index-size statistics (from pg_class and elsewhere)
@@ -938,31 +951,39 @@ struct IndexOptInfo
     /* number of key columns in index */
     int            nkeycolumns;

+    /*
+     * array fields aren't really worth the trouble to print
+     */
+
     /*
      * column numbers of index's attributes both key and included columns, or
      * 0
      */
-    int           *indexkeys;
+    int           *indexkeys pg_node_attr(read_write_ignore);
     /* OIDs of collations of index columns */
-    Oid           *indexcollations;
+    Oid           *indexcollations pg_node_attr(read_write_ignore);
     /* OIDs of operator families for columns */
-    Oid           *opfamily;
+    Oid           *opfamily pg_node_attr(read_write_ignore);
     /* OIDs of opclass declared input data types */
-    Oid           *opcintype;
+    Oid           *opcintype pg_node_attr(read_write_ignore);
     /* OIDs of btree opfamilies, if orderable */
-    Oid           *sortopfamily;
+    Oid           *sortopfamily pg_node_attr(read_write_ignore);
     /* is sort order descending? */
-    bool       *reverse_sort;
+    bool       *reverse_sort pg_node_attr(read_write_ignore);
     /* do NULLs come first in the sort order? */
-    bool       *nulls_first;
+    bool       *nulls_first pg_node_attr(read_write_ignore);
     /* opclass-specific options for columns */
-    bytea      **opclassoptions;
+    bytea      **opclassoptions pg_node_attr(read_write_ignore);
     /* which index cols can be returned in an index-only scan? */
-    bool       *canreturn;
+    bool       *canreturn pg_node_attr(read_write_ignore);
     /* OID of the access method (in pg_am) */
     Oid            relam;
-    /* expressions for non-simple index columns */
-    List       *indexprs;
+
+    /*
+     * expressions for non-simple index columns; redundant to print since we
+     * print indextlist
+     */
+    List       *indexprs pg_node_attr(read_write_ignore);
     /* predicate if a partial index, else NIL */
     List       *indpred;

@@ -989,17 +1010,17 @@ struct IndexOptInfo
      * Remaining fields are copied from the index AM's API struct
      * (IndexAmRoutine)
      */
-    bool        amcanorderbyop;
-    bool        amoptionalkey;
-    bool        amsearcharray;
-    bool        amsearchnulls;
+    bool        amcanorderbyop pg_node_attr(read_write_ignore);
+    bool        amoptionalkey pg_node_attr(read_write_ignore);
+    bool        amsearcharray pg_node_attr(read_write_ignore);
+    bool        amsearchnulls pg_node_attr(read_write_ignore);
     /* does AM have amgettuple interface? */
-    bool        amhasgettuple;
+    bool        amhasgettuple pg_node_attr(read_write_ignore);
     /* does AM have amgetbitmap interface? */
-    bool        amhasgetbitmap;
-    bool        amcanparallel;
+    bool        amhasgetbitmap pg_node_attr(read_write_ignore);
+    bool        amcanparallel pg_node_attr(read_write_ignore);
     /* does AM have ammarkpos interface? */
-    bool        amcanmarkpos;
+    bool        amcanmarkpos pg_node_attr(read_write_ignore);
     /* Rather than include amapi.h here, we declare amcostestimate like this */
     void        (*amcostestimate) ();    /* AM's cost estimator */
 };
@@ -1012,7 +1033,7 @@ struct IndexOptInfo
  * INDEX_MAX_KEYS columns in a foreign key constraint.  Each array has
  * nkeys valid entries.
  */
-typedef struct ForeignKeyOptInfo
+typedef struct ForeignKeyOptInfo pg_node_attr(custom_read_write, no_copy_equal, no_read)
 {
     NodeTag        type;

@@ -1027,11 +1048,11 @@ typedef struct ForeignKeyOptInfo
     /* number of columns in the foreign key */
     int            nkeys;
     /* cols in referencing table */
-    AttrNumber    conkey[INDEX_MAX_KEYS];
+    AttrNumber    conkey[INDEX_MAX_KEYS] pg_node_attr(array_size(nkeys));
     /* cols in referenced table */
-    AttrNumber    confkey[INDEX_MAX_KEYS];
+    AttrNumber    confkey[INDEX_MAX_KEYS] pg_node_attr(array_size(nkeys));
     /* PK = FK operator OIDs */
-    Oid            conpfeqop[INDEX_MAX_KEYS];
+    Oid            conpfeqop[INDEX_MAX_KEYS] pg_node_attr(array_size(nkeys));

     /*
      * Derived info about whether FK's equality conditions match the query:
@@ -1060,7 +1081,7 @@ typedef struct ForeignKeyOptInfo
  * Each pg_statistic_ext row is represented by one or more nodes of this
  * type, or even zero if ANALYZE has not computed them.
  */
-typedef struct StatisticExtInfo
+typedef struct StatisticExtInfo pg_node_attr(no_copy_equal)
 {
     NodeTag        type;

@@ -1068,10 +1089,13 @@ typedef struct StatisticExtInfo
     Oid            statOid;

     /* includes child relations */
-    bool        inherit;
+    bool        inherit pg_node_attr(read_write_ignore);

-    /* back-link to statistic's table */
-    RelOptInfo *rel;
+    /*
+     * back-link to statistic's table; don't print, infinite recursion on plan
+     * tree dump
+     */
+    RelOptInfo *rel pg_node_attr(read_write_ignore);

     /* statistics kind of this entry */
     char        kind;
@@ -1123,7 +1147,7 @@ typedef struct StatisticExtInfo
  * NB: if ec_merged isn't NULL, this class has been merged into another, and
  * should be ignored in favor of using the pointed-to class.
  */
-typedef struct EquivalenceClass
+typedef struct EquivalenceClass pg_node_attr(custom_read_write, no_copy_equal, no_read)
 {
     NodeTag        type;

@@ -1173,7 +1197,7 @@ typedef struct EquivalenceClass
  * anyarray_ops would never work without this.  Use em_datatype when
  * looking up a specific btree operator to work with this expression.
  */
-typedef struct EquivalenceMember
+typedef struct EquivalenceMember pg_node_attr(no_copy_equal)
 {
     NodeTag        type;

@@ -1257,7 +1281,7 @@ typedef enum VolatileFunctionStatus
  * deal with sort/group refnos when needed with less expense than including
  * TargetEntry nodes in the exprs list.
  */
-typedef struct PathTarget
+typedef struct PathTarget pg_node_attr(no_copy_equal, no_read)
 {
     NodeTag        type;

@@ -1265,7 +1289,7 @@ typedef struct PathTarget
     List       *exprs;

     /* corresponding sort/group refnos, or 0 */
-    Index       *sortgrouprefs;
+    Index       *sortgrouprefs pg_node_attr(array_size(exprs));

     /* cost of evaluating the expressions */
     QualCost    cost;
@@ -1296,7 +1320,7 @@ typedef struct PathTarget
  * on how the join is formed.  The relevant clauses will appear in each
  * parameterized join path's joinrestrictinfo list, instead.
  */
-typedef struct ParamPathInfo
+typedef struct ParamPathInfo pg_node_attr(no_copy_equal)
 {
     NodeTag        type;

@@ -1334,22 +1358,41 @@ typedef struct ParamPathInfo
  *
  * "pathkeys" is a List of PathKey nodes (see above), describing the sort
  * ordering of the path's output rows.
+ *
+ * We do not support copying Path trees, mainly because the circular linkages
+ * between RelOptInfo and Path nodes can't be handled easily in a simple
+ * depth-first traversal.  We also don't have read support at the moment.
  */
-typedef struct Path
+typedef struct Path pg_node_attr(no_copy_equal, no_read)
 {
     NodeTag        type;

     /* tag identifying scan/join method */
     NodeTag        pathtype;

-    /* the relation this path can build */
-    RelOptInfo *parent;
+    /*
+     * the relation this path can build
+     *
+     * We do NOT print the parent, else we'd be in infinite recursion.  We can
+     * print the parent's relids for identification purposes, though.
+     */
+    RelOptInfo *parent pg_node_attr(write_only_relids);

-    /* list of Vars/Exprs, cost, width */
-    PathTarget *pathtarget;
+    /*
+     * list of Vars/Exprs, cost, width
+     *
+     * We print the pathtarget only if it's not the default one for the rel.
+     */
+    PathTarget *pathtarget pg_node_attr(write_only_nondefault_pathtarget);

-    /* parameterization info, or NULL if none */
-    ParamPathInfo *param_info;
+    /*
+     * parameterization info, or NULL if none
+     *
+     * We do not print the whole of param_info, since it's printed via
+     * RelOptInfo; it's sufficient and less cluttering to print just the
+     * required outer relids.
+     */
+    ParamPathInfo *param_info pg_node_attr(write_only_req_outer);

     /* engage parallel-aware logic? */
     bool        parallel_aware;
@@ -1455,7 +1498,7 @@ typedef struct IndexPath
  * column, i.e. the indexcol values must form a nondecreasing sequence.
  * (The order of multiple clauses for the same index column is unspecified.)
  */
-typedef struct IndexClause
+typedef struct IndexClause pg_node_attr(no_copy_equal)
 {
     NodeTag        type;
     struct RestrictInfo *rinfo; /* original restriction or join clause */
@@ -1750,7 +1793,7 @@ typedef struct GatherMergePath
  * All join-type paths share these fields.
  */

-typedef struct JoinPath
+typedef struct JoinPath pg_node_attr(abstract)
 {
     Path        path;

@@ -1952,14 +1995,14 @@ typedef struct AggPath
  * Various annotations used for grouping sets in the planner.
  */

-typedef struct GroupingSetData
+typedef struct GroupingSetData pg_node_attr(no_copy_equal)
 {
     NodeTag        type;
     List       *set;            /* grouping set as list of sortgrouprefs */
     Cardinality numGroups;        /* est. number of result groups */
 } GroupingSetData;

-typedef struct RollupData
+typedef struct RollupData pg_node_attr(no_copy_equal)
 {
     NodeTag        type;
     List       *groupClause;    /* applicable subset of parse->groupClause */
@@ -2224,6 +2267,12 @@ typedef struct LimitPath
  * apply only one.  We mark clauses of this kind by setting parent_ec to
  * point to the generating EquivalenceClass.  Multiple clauses with the same
  * parent_ec in the same join are redundant.
+ *
+ * Most fields are ignored for equality, since they may not be set yet, and
+ * should be derivable from the clause anyway.
+ *
+ * parent_ec, left_ec, right_ec are not printed, lest it lead to infinite
+ * recursion in plan tree dump.
  */

 typedef struct RestrictInfo
@@ -2240,22 +2289,22 @@ typedef struct RestrictInfo
     bool        outerjoin_delayed;

     /* see comment above */
-    bool        can_join;
+    bool        can_join pg_node_attr(equal_ignore);

     /* see comment above */
-    bool        pseudoconstant;
+    bool        pseudoconstant pg_node_attr(equal_ignore);

     /* true if known to contain no leaked Vars */
-    bool        leakproof;
+    bool        leakproof pg_node_attr(equal_ignore);

     /* to indicate if clause contains any volatile functions. */
-    VolatileFunctionStatus has_volatile;
+    VolatileFunctionStatus has_volatile pg_node_attr(equal_ignore);

     /* see comment above */
     Index        security_level;

     /* The set of relids (varnos) actually referenced in the clause: */
-    Relids        clause_relids;
+    Relids        clause_relids pg_node_attr(equal_ignore);

     /* The set of relids required to evaluate the clause: */
     Relids        required_relids;
@@ -2270,84 +2319,89 @@ typedef struct RestrictInfo
      * Relids in the left/right side of the clause.  These fields are set for
      * any binary opclause.
      */
-    Relids        left_relids;
-    Relids        right_relids;
+    Relids        left_relids pg_node_attr(equal_ignore);
+    Relids        right_relids pg_node_attr(equal_ignore);

     /*
      * Modified clause with RestrictInfos.  This field is NULL unless clause
      * is an OR clause.
      */
-    Expr       *orclause;
+    Expr       *orclause pg_node_attr(equal_ignore);

     /*
      * Generating EquivalenceClass.  This field is NULL unless clause is
      * potentially redundant.
      */
-    EquivalenceClass *parent_ec;
+    EquivalenceClass *parent_ec pg_node_attr(equal_ignore, read_write_ignore);

     /*
      * cache space for cost and selectivity
      */

     /* eval cost of clause; -1 if not yet set */
-    QualCost    eval_cost;
+    QualCost    eval_cost pg_node_attr(equal_ignore);

     /*
      * selectivity for "normal" (JOIN_INNER) semantics; -1 if not yet set; >1
      * means a redundant clause
      */
-    Selectivity norm_selec;
+    Selectivity norm_selec pg_node_attr(equal_ignore);
     /* selectivity for outer join semantics; -1 if not yet set */
-    Selectivity outer_selec;
+    Selectivity outer_selec pg_node_attr(equal_ignore);

     /*
      * opfamilies containing clause operator; valid if clause is
      * mergejoinable, else NIL
      */
-    List       *mergeopfamilies;
+    List       *mergeopfamilies pg_node_attr(equal_ignore);

     /*
      * cache space for mergeclause processing; NULL if not yet set
      */

     /* EquivalenceClass containing lefthand */
-    EquivalenceClass *left_ec;
+    EquivalenceClass *left_ec pg_node_attr(equal_ignore, read_write_ignore);
     /* EquivalenceClass containing righthand */
-    EquivalenceClass *right_ec;
+    EquivalenceClass *right_ec pg_node_attr(equal_ignore, read_write_ignore);
     /* EquivalenceMember for lefthand */
-    EquivalenceMember *left_em;
+    EquivalenceMember *left_em pg_node_attr(equal_ignore);
     /* EquivalenceMember for righthand */
-    EquivalenceMember *right_em;
-    /* list of MergeScanSelCache structs */
-    List       *scansel_cache;
+    EquivalenceMember *right_em pg_node_attr(equal_ignore);
+
+    /*
+     * List of MergeScanSelCache structs.  Those aren't Nodes, so hard to
+     * copy.  Ignoring it will have the effect that copying will just reset
+     * the cache.
+     */
+    List       *scansel_cache pg_node_attr(copy_ignore, equal_ignore);

     /*
      * transient workspace for use while considering a specific join path; T =
      * outer var on left, F = on right
      */
-    bool        outer_is_left;
+    bool        outer_is_left pg_node_attr(equal_ignore);

     /*
      * copy of clause operator; valid if clause is hashjoinable, else
      * InvalidOid
      */
-    Oid            hashjoinoperator;
+    Oid            hashjoinoperator pg_node_attr(equal_ignore);

     /*
      * cache space for hashclause processing; -1 if not yet set
      */
     /* avg bucketsize of left side */
-    Selectivity left_bucketsize;
+    Selectivity left_bucketsize pg_node_attr(equal_ignore);
     /* avg bucketsize of right side */
-    Selectivity right_bucketsize;
+    Selectivity right_bucketsize pg_node_attr(equal_ignore);
     /* left side's most common val's freq */
-    Selectivity left_mcvfreq;
+    Selectivity left_mcvfreq pg_node_attr(equal_ignore);
     /* right side's most common val's freq */
-    Selectivity right_mcvfreq;
+    Selectivity right_mcvfreq pg_node_attr(equal_ignore);

     /* hash equality operators used for memoize nodes, else InvalidOid */
-    Oid            left_hasheqoperator;
-    Oid            right_hasheqoperator;
+    Oid            left_hasheqoperator pg_node_attr(equal_ignore);
+    Oid            right_hasheqoperator pg_node_attr(equal_ignore);
 } RestrictInfo;

 /*
@@ -2397,6 +2451,17 @@ typedef struct MergeScanSelCache
  * Although the planner treats this as an expression node type, it is not
  * recognized by the parser or executor, so we declare it here rather than
  * in primnodes.h.
+ *
+ * We intentionally do not compare phexpr.  Two PlaceHolderVars with the
+ * same ID and levelsup should be considered equal even if the contained
+ * expressions have managed to mutate to different states.  This will
+ * happen during final plan construction when there are nested PHVs, since
+ * the inner PHV will get replaced by a Param in some copies of the outer
+ * PHV.  Another way in which it can happen is that initplan sublinks
+ * could get replaced by differently-numbered Params when sublink folding
+ * is done.  (The end result of such a situation would be some
+ * unreferenced initplans, which is annoying but not really a problem.) On
+ * the same reasoning, there is no need to examine phrels.
  */

 typedef struct PlaceHolderVar
@@ -2404,10 +2469,10 @@ typedef struct PlaceHolderVar
     Expr        xpr;

     /* the represented expression */
-    Expr       *phexpr;
+    Expr       *phexpr pg_node_attr(equal_ignore);

     /* base relids syntactically within expr src */
-    Relids        phrels;
+    Relids        phrels pg_node_attr(equal_ignore);

     /* ID for PHV (unique within planner run) */
     Index        phid;
@@ -2572,7 +2637,7 @@ typedef struct AppendRelInfo
      * child column is dropped or doesn't exist in the parent.
      */
     int            num_child_cols; /* length of array */
-    AttrNumber *parent_colnos;
+    AttrNumber *parent_colnos pg_node_attr(array_size(num_child_cols));

     /*
      * We store the parent table's OID here for inheritance, or InvalidOid for
@@ -2600,7 +2665,7 @@ typedef struct AppendRelInfo
  * We add such a reference to root->processed_tlist when creating the entry,
  * and it propagates into the plan tree from there.
  */
-typedef struct RowIdentityVarInfo
+typedef struct RowIdentityVarInfo pg_node_attr(no_copy_equal)
 {
     NodeTag        type;

@@ -2643,7 +2708,10 @@ typedef struct PlaceHolderInfo
     /* ID for PH (unique within planner run) */
     Index        phid;

-    /* copy of PlaceHolderVar tree */
+    /*
+     * copy of PlaceHolderVar tree (should be redundant for comparison, could
+     * be ignored)
+     */
     PlaceHolderVar *ph_var;

     /* lowest level we can evaluate value at */
@@ -2664,7 +2732,7 @@ typedef struct PlaceHolderInfo
  * function.  MinMaxAggPath contains a list of these, and if we accept that
  * path, the list is stored into root->minmax_aggs for use during setrefs.c.
  */
-typedef struct MinMaxAggInfo
+typedef struct MinMaxAggInfo pg_node_attr(no_copy_equal)
 {
     NodeTag        type;

@@ -2677,8 +2745,11 @@ typedef struct MinMaxAggInfo
     /* expression we are aggregating on */
     Expr       *target;

-    /* modified "root" for planning the subquery */
-    PlannerInfo *subroot;
+    /*
+     * modified "root" for planning the subquery; not printed, too large, not
+     * interesting enough
+     */
+    PlannerInfo *subroot pg_node_attr(read_write_ignore);

     /* access path for subquery */
     Path       *path;
@@ -2737,7 +2808,7 @@ typedef struct MinMaxAggInfo
  * Instead, we just record the assignment of the slot number by appending to
  * root->glob->paramExecTypes.
  */
-typedef struct PlannerParamItem
+typedef struct PlannerParamItem pg_node_attr(no_copy_equal)
 {
     NodeTag        type;

diff --git a/src/include/nodes/plannodes.h b/src/include/nodes/plannodes.h
index d5c0ebe859..19b5ce2ec6 100644
--- a/src/include/nodes/plannodes.h
+++ b/src/include/nodes/plannodes.h
@@ -286,16 +286,16 @@ typedef struct MergeAppend
     int            numCols;

     /* their indexes in the target list */
-    AttrNumber *sortColIdx;
+    AttrNumber *sortColIdx pg_node_attr(array_size(numCols));

     /* OIDs of operators to sort them by */
-    Oid           *sortOperators;
+    Oid           *sortOperators pg_node_attr(array_size(numCols));

     /* OIDs of collations */
-    Oid           *collations;
+    Oid           *collations pg_node_attr(array_size(numCols));

     /* NULLS FIRST/LAST directions */
-    bool       *nullsFirst;
+    bool       *nullsFirst pg_node_attr(array_size(numCols));

     /* Info for run-time subplan pruning; NULL if we're not doing that */
     struct PartitionPruneInfo *part_prune_info;
@@ -322,11 +322,11 @@ typedef struct RecursiveUnion
     int            numCols;

     /* their indexes in the target list */
-    AttrNumber *dupColIdx;
+    AttrNumber *dupColIdx pg_node_attr(array_size(numCols));

     /* equality operators to compare with */
-    Oid           *dupOperators;
-    Oid           *dupCollations;
+    Oid           *dupOperators pg_node_attr(array_size(numCols));
+    Oid           *dupCollations pg_node_attr(array_size(numCols));

     /* estimated number of groups in input */
     long        numGroups;
@@ -812,16 +812,16 @@ typedef struct MergeJoin
     /* these are arrays, but have the same length as the mergeclauses list: */

     /* per-clause OIDs of btree opfamilies */
-    Oid           *mergeFamilies;
+    Oid           *mergeFamilies pg_node_attr(array_size(mergeclauses));

     /* per-clause OIDs of collations */
-    Oid           *mergeCollations;
+    Oid           *mergeCollations pg_node_attr(array_size(mergeclauses));

     /* per-clause ordering (ASC or DESC) */
-    int           *mergeStrategies;
+    int           *mergeStrategies pg_node_attr(array_size(mergeclauses));

     /* per-clause nulls ordering */
-    bool       *mergeNullsFirst;
+    bool       *mergeNullsFirst pg_node_attr(array_size(mergeclauses));
 } MergeJoin;

 /* ----------------
@@ -863,10 +863,10 @@ typedef struct Memoize
     int            numKeys;

     /* hash operators for each key */
-    Oid           *hashOperators;
+    Oid           *hashOperators pg_node_attr(array_size(numKeys));

     /* collations for each key */
-    Oid           *collations;
+    Oid           *collations pg_node_attr(array_size(numKeys));

     /* cache keys in the form of exprs containing parameters */
     List       *param_exprs;
@@ -905,16 +905,16 @@ typedef struct Sort
     int            numCols;

     /* their indexes in the target list */
-    AttrNumber *sortColIdx;
+    AttrNumber *sortColIdx pg_node_attr(array_size(numCols));

     /* OIDs of operators to sort them by */
-    Oid           *sortOperators;
+    Oid           *sortOperators pg_node_attr(array_size(numCols));

     /* OIDs of collations */
-    Oid           *collations;
+    Oid           *collations pg_node_attr(array_size(numCols));

     /* NULLS FIRST/LAST directions */
-    bool       *nullsFirst;
+    bool       *nullsFirst pg_node_attr(array_size(numCols));
 } Sort;

 /* ----------------
@@ -941,11 +941,11 @@ typedef struct Group
     int            numCols;

     /* their indexes in the target list */
-    AttrNumber *grpColIdx;
+    AttrNumber *grpColIdx pg_node_attr(array_size(numCols));

     /* equality operators to compare with */
-    Oid           *grpOperators;
-    Oid           *grpCollations;
+    Oid           *grpOperators pg_node_attr(array_size(numCols));
+    Oid           *grpCollations pg_node_attr(array_size(numCols));
 } Group;

 /* ---------------
@@ -976,11 +976,11 @@ typedef struct Agg
     int            numCols;

     /* their indexes in the target list */
-    AttrNumber *grpColIdx;
+    AttrNumber *grpColIdx pg_node_attr(array_size(numCols));

     /* equality operators to compare with */
-    Oid           *grpOperators;
-    Oid           *grpCollations;
+    Oid           *grpOperators pg_node_attr(array_size(numCols));
+    Oid           *grpCollations pg_node_attr(array_size(numCols));

     /* estimated number of groups in input */
     long        numGroups;
@@ -1015,25 +1015,25 @@ typedef struct WindowAgg
     int            partNumCols;

     /* their indexes in the target list */
-    AttrNumber *partColIdx;
+    AttrNumber *partColIdx pg_node_attr(array_size(partNumCols));

     /* equality operators for partition columns */
-    Oid           *partOperators;
+    Oid           *partOperators pg_node_attr(array_size(partNumCols));

     /* collations for partition columns */
-    Oid           *partCollations;
+    Oid           *partCollations pg_node_attr(array_size(partNumCols));

     /* number of columns in ordering clause */
     int            ordNumCols;

     /* their indexes in the target list */
-    AttrNumber *ordColIdx;
+    AttrNumber *ordColIdx pg_node_attr(array_size(ordNumCols));

     /* equality operators for ordering columns */
-    Oid           *ordOperators;
+    Oid           *ordOperators pg_node_attr(array_size(ordNumCols));

     /* collations for ordering columns */
-    Oid           *ordCollations;
+    Oid           *ordCollations pg_node_attr(array_size(ordNumCols));

     /* frame_clause options, see WindowDef */
     int            frameOptions;
@@ -1086,13 +1086,13 @@ typedef struct Unique
     int            numCols;

     /* their indexes in the target list */
-    AttrNumber *uniqColIdx;
+    AttrNumber *uniqColIdx pg_node_attr(array_size(numCols));

     /* equality operators to compare with */
-    Oid           *uniqOperators;
+    Oid           *uniqOperators pg_node_attr(array_size(numCols));

     /* collations for equality comparisons */
-    Oid           *uniqCollations;
+    Oid           *uniqCollations pg_node_attr(array_size(numCols));
 } Unique;

 /* ------------
@@ -1137,16 +1137,16 @@ typedef struct GatherMerge
     int            numCols;

     /* their indexes in the target list */
-    AttrNumber *sortColIdx;
+    AttrNumber *sortColIdx pg_node_attr(array_size(numCols));

     /* OIDs of operators to sort them by */
-    Oid           *sortOperators;
+    Oid           *sortOperators pg_node_attr(array_size(numCols));

     /* OIDs of collations */
-    Oid           *collations;
+    Oid           *collations pg_node_attr(array_size(numCols));

     /* NULLS FIRST/LAST directions */
-    bool       *nullsFirst;
+    bool       *nullsFirst pg_node_attr(array_size(numCols));

     /*
      * param id's of initplans which are referred at gather merge or one of
@@ -1197,11 +1197,11 @@ typedef struct SetOp
     int            numCols;

     /* their indexes in the target list */
-    AttrNumber *dupColIdx;
+    AttrNumber *dupColIdx pg_node_attr(array_size(numCols));

     /* equality operators to compare with */
-    Oid           *dupOperators;
-    Oid           *dupCollations;
+    Oid           *dupOperators pg_node_attr(array_size(numCols));
+    Oid           *dupCollations pg_node_attr(array_size(numCols));

     /* where is the flag column, if any */
     AttrNumber    flagColIdx;
@@ -1253,13 +1253,13 @@ typedef struct Limit
     int            uniqNumCols;

     /* their indexes in the target list */
-    AttrNumber *uniqColIdx;
+    AttrNumber *uniqColIdx pg_node_attr(array_size(uniqNumCols));

     /* equality operators to compare with */
-    Oid           *uniqOperators;
+    Oid           *uniqOperators pg_node_attr(array_size(uniqNumCols));

     /* collations for equality comparisons */
-    Oid           *uniqCollations;
+    Oid           *uniqCollations pg_node_attr(array_size(uniqNumCols));
 } Limit;


@@ -1425,13 +1425,13 @@ typedef struct PartitionedRelPruneInfo
     int            nparts;

     /* subplan index by partition index, or -1 */
-    int           *subplan_map;
+    int           *subplan_map pg_node_attr(array_size(nparts));

     /* subpart index by partition index, or -1 */
-    int           *subpart_map;
+    int           *subpart_map pg_node_attr(array_size(nparts));

     /* relation OID by partition index, or 0 */
-    Oid           *relid_map;
+    Oid           *relid_map pg_node_attr(array_size(nparts));

     /*
      * initial_pruning_steps shows how to prune during executor startup (i.e.,
@@ -1452,7 +1452,7 @@ typedef struct PartitionedRelPruneInfo
  *
  * step_id is the global identifier of the step within its pruning context.
  */
-typedef struct PartitionPruneStep
+typedef struct PartitionPruneStep pg_node_attr(abstract)
 {
     NodeTag        type;
     int            step_id;
diff --git a/src/include/nodes/primnodes.h b/src/include/nodes/primnodes.h
index 732c00c098..9e6b4bdb1d 100644
--- a/src/include/nodes/primnodes.h
+++ b/src/include/nodes/primnodes.h
@@ -64,8 +64,11 @@ typedef struct RangeVar
 {
     NodeTag        type;

-    /* the catalog (database) name, or NULL */
-    char       *catalogname;
+    /*
+     * the catalog (database) name, or NULL; ignored for read/write, since it
+     * is presently not semantically meaningful
+     */
+    char       *catalogname pg_node_attr(read_write_ignore);

     /* the schema name, or NULL */
     char       *schemaname;
@@ -155,7 +158,7 @@ typedef struct IntoClause
  * contains NodeTag, this is a formality, but it is an easy form of
  * documentation.  See also the ExprState node types in execnodes.h.
  */
-typedef struct Expr
+typedef struct Expr pg_node_attr(abstract)
 {
     NodeTag        type;
 } Expr;
@@ -233,10 +236,15 @@ typedef struct Var
      */
     Index        varlevelsup;

+    /*
+     * varnosyn/varattnosyn are ignored for equality, because Vars with
+     * different syntactic identifiers are semantically the same as long as
+     * their varno/varattno match.
+     */
     /* syntactic relation index (0 if unknown) */
-    Index        varnosyn;
+    Index        varnosyn pg_node_attr(equal_ignore);
     /* syntactic attribute number */
-    AttrNumber    varattnosyn;
+    AttrNumber    varattnosyn pg_node_attr(equal_ignore);

     /* token location, or -1 if unknown */
     int            location;
@@ -250,7 +258,7 @@ typedef struct Var
  * references).  This ensures that the Const node is self-contained and makes
  * it more likely that equal() will see logically identical values as equal.
  */
-typedef struct Const
+typedef struct Const pg_node_attr(custom_copy_equal, custom_read_write)
 {
     Expr        xpr;
     Oid            consttype;        /* pg_type OID of the constant's datatype */
@@ -374,8 +382,11 @@ typedef struct Aggref
     /* OID of collation that function should use */
     Oid            inputcollid;

-    /* type Oid of aggregate's transition value */
-    Oid            aggtranstype;
+    /*
+     * type Oid of aggregate's transition value; ignored for equal since it
+     * might not be set yet
+     */
+    Oid            aggtranstype pg_node_attr(equal_ignore);

     /* type Oids of direct and aggregated args */
     List       *aggargtypes;
@@ -455,10 +466,10 @@ typedef struct GroupingFunc
     List       *args;

     /* ressortgrouprefs of arguments */
-    List       *refs;
+    List       *refs pg_node_attr(equal_ignore);

     /* actual column positions set by planner */
-    List       *cols;
+    List       *cols pg_node_attr(equal_ignore);

     /* same as Aggref.agglevelsup */
     Index        agglevelsup;
@@ -634,7 +645,7 @@ typedef struct OpExpr
     Oid            opno;

     /* PG_PROC OID of underlying function */
-    Oid            opfuncid;
+    Oid            opfuncid pg_node_attr(equal_ignore_if_zero);

     /* PG_TYPE OID of result value */
     Oid            opresulttype;
@@ -698,6 +709,10 @@ typedef OpExpr NullIfExpr;
  * corresponding function and won't be used during execution.  For
  * non-hashtable based NOT INs, negfuncid will be set to InvalidOid.  See
  * convert_saop_to_hashed_saop().
+ *
+ * Similar to OpExpr, opfuncid, hashfuncid, and negfuncid are not necessarily
+ * filled in right away, so will be ignored for equality if they are not set
+ * yet.
  */
 typedef struct ScalarArrayOpExpr
 {
@@ -707,13 +722,13 @@ typedef struct ScalarArrayOpExpr
     Oid            opno;

     /* PG_PROC OID of comparison function */
-    Oid            opfuncid;
+    Oid            opfuncid pg_node_attr(equal_ignore_if_zero);

     /* PG_PROC OID of hash func or InvalidOid */
-    Oid            hashfuncid;
+    Oid            hashfuncid pg_node_attr(equal_ignore_if_zero);

     /* PG_PROC OID of negator of opfuncid function or InvalidOid.  See above */
-    Oid            negfuncid;
+    Oid            negfuncid pg_node_attr(equal_ignore_if_zero);

     /* true for ANY, false for ALL */
     bool        useOr;
@@ -740,7 +755,7 @@ typedef enum BoolExprType
     AND_EXPR, OR_EXPR, NOT_EXPR
 } BoolExprType;

-typedef struct BoolExpr
+typedef struct BoolExpr pg_node_attr(custom_read_write)
 {
     Expr        xpr;
     BoolExprType boolop;
diff --git a/src/include/nodes/value.h b/src/include/nodes/value.h
index eaf937051c..6193f51536 100644
--- a/src/include/nodes/value.h
+++ b/src/include/nodes/value.h
@@ -25,7 +25,7 @@
  * (There used to be a Value node, which encompassed all these different node types.  Hence the name of this file.)
  */

-typedef struct Integer
+typedef struct Integer pg_node_attr(special_read_write)
 {
     NodeTag        type;
     int            ival;
@@ -42,25 +42,25 @@ typedef struct Integer
  * Note that an integer-looking string will get lexed as T_Float if the value
  * is too large to fit in an 'int'.
  */
-typedef struct Float
+typedef struct Float pg_node_attr(special_read_write)
 {
     NodeTag        type;
     char       *fval;
 } Float;

-typedef struct Boolean
+typedef struct Boolean pg_node_attr(special_read_write)
 {
     NodeTag        type;
     bool        boolval;
 } Boolean;

-typedef struct String
+typedef struct String pg_node_attr(special_read_write)
 {
     NodeTag        type;
     char       *sval;
 } String;

-typedef struct BitString
+typedef struct BitString pg_node_attr(special_read_write)
 {
     NodeTag        type;
     char       *bsval;
diff --git a/src/include/utils/rel.h b/src/include/utils/rel.h
index 1896a9a06d..3e6da86688 100644
--- a/src/include/utils/rel.h
+++ b/src/include/utils/rel.h
@@ -265,7 +265,7 @@ typedef struct RelationData
  * Currently, we mostly cache fields of interest to the planner, but the set
  * of fields has already grown the constraint OID for other uses.
  */
-typedef struct ForeignKeyCacheInfo
+typedef struct ForeignKeyCacheInfo pg_node_attr(no_read)
 {
     NodeTag        type;
     Oid            conoid;            /* oid of the constraint itself */
@@ -273,9 +273,12 @@ typedef struct ForeignKeyCacheInfo
     Oid            confrelid;        /* relation referenced by the foreign key */
     int            nkeys;            /* number of columns in the foreign key */
     /* these arrays each have nkeys valid entries: */
-    AttrNumber    conkey[INDEX_MAX_KEYS]; /* cols in referencing table */
-    AttrNumber    confkey[INDEX_MAX_KEYS];    /* cols in referenced table */
-    Oid            conpfeqop[INDEX_MAX_KEYS];    /* PK = FK operator OIDs */
+    /* cols in referencing table */
+    AttrNumber    conkey[INDEX_MAX_KEYS] pg_node_attr(array_size(nkeys));
+    /* cols in referenced table */
+    AttrNumber    confkey[INDEX_MAX_KEYS] pg_node_attr(array_size(nkeys));
+    /* PK = FK operator OIDs */
+    Oid            conpfeqop[INDEX_MAX_KEYS] pg_node_attr(array_size(nkeys));
 } ForeignKeyCacheInfo;


diff --git a/src/tools/msvc/Solution.pm b/src/tools/msvc/Solution.pm
index d30e8fcb11..286b5810c9 100644
--- a/src/tools/msvc/Solution.pm
+++ b/src/tools/msvc/Solution.pm
@@ -841,6 +841,52 @@ EOF
         close($chs);
     }

+    if (IsNewer('src/backend/nodes/node-support-stamp',
+        'src/backend/nodes/gen_node_support.pl'))
+    {
+        # XXX duplicates src/backend/nodes/Makefile
+
+        my @node_headers = qw(
+            nodes/nodes.h
+            nodes/execnodes.h
+            nodes/plannodes.h
+            nodes/primnodes.h
+            nodes/pathnodes.h
+            nodes/extensible.h
+            nodes/parsenodes.h
+            nodes/replnodes.h
+            nodes/value.h
+            commands/trigger.h
+            commands/event_trigger.h
+            foreign/fdwapi.h
+            access/amapi.h
+            access/tableam.h
+            access/tsmapi.h
+            utils/rel.h
+            nodes/supportnodes.h
+            executor/tuptable.h
+            nodes/lockoptions.h
+            access/sdir.h
+        );
+
+        chdir('src/backend/nodes');
+
+        my @node_files = map { "../../../src/include/$_" } @node_headers;
+
+        system("perl gen_node_support.pl @node_files");
+        open(my $f, '>', 'node-support-stamp') || confess "Could not touch node-support-stamp";
+        close($f);
+        chdir('../../..');
+    }
+
+    if (IsNewer(
+            'src/include/nodes/nodetags.h',
+            'src/backend/nodes/nodetags.h'))
+    {
+        copyFile('src/backend/nodes/nodetags.h',
+            'src/include/nodes/nodetags.h');
+    }
+
     open(my $o, '>', "doc/src/sgml/version.sgml")
       || croak "Could not write to version.sgml\n";
     print $o <<EOF;

base-commit: b55f62abb2c2e07dfae99e19a2b3d7ca9e58dc1a
--
2.36.1

diff --git a/src/backend/nodes/gen_node_support.pl b/src/backend/nodes/gen_node_support.pl
index 6aaf401a72..c77a130054 100644
--- a/src/backend/nodes/gen_node_support.pl
+++ b/src/backend/nodes/gen_node_support.pl
@@ -40,8 +40,10 @@ my @node_types = qw(Node);
 # collect info for each node type
 my %node_type_info;

-# node types we don't want copy/equal support for
-my @no_copy_equal;
+# node types we don't want copy support for
+my @no_copy;
+# node types we don't want equal support for
+my @no_equal;
 # node types we don't want read support for
 my @no_read;
 # node types we don't want read/write support for
@@ -90,8 +92,21 @@ push @scalar_types, qw(QualCost);

 # XXX various things we are not publishing right now to stay level
 # with the manual system
-push @no_copy_equal, qw(CallContext InlineCodeBlock);
+push @no_copy, qw(CallContext InlineCodeBlock);
+push @no_equal, qw(CallContext InlineCodeBlock);
 push @no_read_write, qw(AccessPriv AlterTableCmd CallContext CreateOpClassItem FunctionParameter InferClause
InlineCodeBlockObjectWithArgs OnConflictClause PartitionCmd RoleSpec VacuumRelation); 
+push @no_read, qw(A_ArrayExpr A_Indices A_Indirection AlterStatsStmt
+CollateClause ColumnDef ColumnRef CreateForeignTableStmt CreateStatsStmt
+CreateStmt FuncCall ImportForeignSchemaStmt IndexElem IndexStmt
+JsonAggConstructor JsonArgument JsonArrayAgg JsonArrayConstructor
+JsonArrayQueryConstructor JsonCommon JsonFuncExpr JsonKeyValue
+JsonObjectAgg JsonObjectConstructor JsonOutput JsonParseExpr JsonScalarExpr
+JsonSerializeExpr JsonTable JsonTableColumn JsonTablePlan LockingClause
+MultiAssignRef PLAssignStmt ParamRef PartitionElem PartitionSpec
+PlaceHolderVar PublicationObjSpec PublicationTable RangeFunction
+RangeSubselect RangeTableFunc RangeTableFuncCol RangeTableSample RawStmt
+ResTarget ReturnStmt SelectStmt SortBy StatsElem TableLikeClause
+TriggerTransition TypeCast TypeName WindowDef WithClause XmlSerialize);


 ## read input
@@ -201,14 +216,16 @@ foreach my $infile (@ARGV)
                         qw(execnodes.h trigger.h event_trigger.h amapi.h tableam.h
                             tsmapi.h fdwapi.h tuptable.h replnodes.h supportnodes.h))
                     {
-                        push @no_copy_equal, $in_struct;
+                        push @no_copy, $in_struct;
+                        push @no_equal, $in_struct;
                         push @no_read_write, $in_struct;
                     }

                     # Propagate some node attributes from supertypes
                     if ($supertype)
                     {
-                        push @no_copy_equal, $in_struct if elem $supertype, @no_copy_equal;
+                        push @no_copy, $in_struct if elem $supertype, @no_copy;
+                        push @no_equal, $in_struct if elem $supertype, @no_equal;
                         push @no_read, $in_struct if elem $supertype, @no_read;
                     }
                 }
@@ -245,7 +262,9 @@ foreach my $infile (@ARGV)
                         foreach my $attr (@attrs)
                         {
                             if ($attr !~ /^array_size\(\w+\)$/ &&
-                                !elem $attr, qw(copy_ignore equal_ignore equal_ignore_if_zero read_write_ignore
+                                $attr !~ /^copy_as\(\w+\)$/ &&
+                                $attr !~ /^read_as\(\w+\)$/ &&
+                                !elem $attr, qw(equal_ignore equal_ignore_if_zero read_write_ignore
                                     write_only_relids write_only_nondefault_pathtarget write_only_req_outer))
                             {
                                 die "$infile:$.: unrecognized attribute \"$attr\"\n";
@@ -291,9 +310,18 @@ foreach my $infile (@ARGV)
                     {
                         push @custom_read_write, $in_struct;
                     }
+                    elsif ($attr eq 'no_copy')
+                    {
+                        push @no_copy, $in_struct;
+                    }
+                    elsif ($attr eq 'no_equal')
+                    {
+                        push @no_equal, $in_struct;
+                    }
                     elsif ($attr eq 'no_copy_equal')
                     {
-                        push @no_copy_equal, $in_struct;
+                        push @no_copy, $in_struct;
+                        push @no_equal, $in_struct;
                     }
                     elsif ($attr eq 'no_read')
                     {
@@ -391,16 +419,18 @@ print $eff $node_includes;
 foreach my $n (@node_types)
 {
     next if elem $n, @abstract_types;
-    next if elem $n, @no_copy_equal;
+    my $struct_no_copy = (elem $n, @no_copy);
+    my $struct_no_equal = (elem $n, @no_equal);
+    next if $struct_no_copy && $struct_no_equal;
     next if $n eq 'List';

     print $cfs "\t\tcase T_${n}:\n".
       "\t\t\tretval = _copy${n}(from);\n".
-      "\t\t\tbreak;\n";
+      "\t\t\tbreak;\n" unless $struct_no_copy;

     print $efs "\t\tcase T_${n}:\n".
       "\t\t\tretval = _equal${n}(a, b);\n".
-      "\t\t\tbreak;\n";
+      "\t\t\tbreak;\n" unless $struct_no_equal;

     next if elem $n, @custom_copy_equal;

@@ -410,21 +440,47 @@ _copy${n}(const $n *from)
 {
 \t${n} *newnode = makeNode($n);

-";
+" unless $struct_no_copy;

     print $eff "
 static bool
 _equal${n}(const $n *a, const $n *b)
 {
-";
+" unless $struct_no_equal;

     # print instructions for each field
     foreach my $f (@{$node_type_info{$n}->{fields}})
     {
         my $t = $node_type_info{$n}->{field_types}{$f};
         my @a = @{ $node_type_info{$n}->{field_attrs}{$f} };
-        my $copy_ignore = (elem 'copy_ignore', @a);
-        my $equal_ignore = (elem 'equal_ignore', @a);
+        my $copy_ignore = $struct_no_copy;
+        my $equal_ignore = $struct_no_equal;
+
+        # extract per-field attributes
+        my $array_size_field;
+        my $copy_as_field;
+        foreach my $a (@a)
+        {
+            if ($a =~ /^array_size.([\w.]+)/)
+            {
+                $array_size_field = $1;
+            }
+            elsif ($a =~ /^copy_as.([\w.]+)/)
+            {
+                $copy_as_field = $1;
+            }
+            elsif ($a eq 'equal_ignore')
+            {
+                $equal_ignore = 1;
+            }
+        }
+
+        # override type-specific copy method if copy_as is specified
+        if ($copy_as_field)
+        {
+            print $cff "\tnewnode->$f = $copy_as_field;\n" unless $copy_ignore;
+            $copy_ignore = 1;
+        }

         # select instructions by field type
         if ($t eq 'char*')
@@ -458,15 +514,6 @@ _equal${n}(const $n *a, const $n *b)
         elsif ($t =~ /(\w+)\*/ and elem $1, @scalar_types)
         {
             my $tt = $1;
-            my $array_size_field;
-            foreach my $a (@a)
-            {
-                if ($a =~ /^array_size.([\w.]+)/)
-                {
-                    $array_size_field = $1;
-                    last;
-                }
-            }
             if (!$array_size_field)
             {
                 die "no array size defined for $n.$f of type $t";
@@ -511,11 +558,11 @@ _equal${n}(const $n *a, const $n *b)
     print $cff "
 \treturn newnode;
 }
-";
+" unless $struct_no_copy;
     print $eff "
 \treturn true;
 }
-";
+" unless $struct_no_equal;
 }

 close $cff;
@@ -546,18 +593,17 @@ foreach my $n (@node_types)
         next unless elem $n, @keep;
     }

-    my $no_read = (elem $n, @no_read);
+    my $struct_no_read = (elem $n, @no_read);

-    # output format starts with upper case node type, underscores stripped
+    # output format starts with upper case node type name
     my $N = uc $n;
-    $N =~ s/_//g;

     print $ofs "\t\t\tcase T_${n}:\n".
       "\t\t\t\t_out${n}(str, obj);\n".
       "\t\t\t\tbreak;\n";

     print $rfs "\telse if (MATCH(\"$N\", " . length($N) . "))\n".
-      "\t\treturn_value = _read${n}();\n" unless $no_read;
+      "\t\treturn_value = _read${n}();\n" unless $struct_no_read;

     next if elem $n, @custom_read_write;

@@ -575,18 +621,47 @@ _read${n}(void)
 {
 \tREAD_LOCALS($n);

-" unless $no_read;
+" unless $struct_no_read;

     # print instructions for each field
     foreach my $f (@{$node_type_info{$n}->{fields}})
     {
         my $t = $node_type_info{$n}->{field_types}{$f};
         my @a = @{ $node_type_info{$n}->{field_attrs}{$f} };
-        next if (elem 'read_write_ignore', @a);
+        my $no_read = $struct_no_read;
+
+        # extract per-field attributes
+        my $read_write_ignore = 0;
+        my $read_as_field;
+        foreach my $a (@a)
+        {
+            if ($a =~ /^read_as.([\w.]+)/)
+            {
+                $read_as_field = $1;
+            }
+            elsif ($a eq 'read_write_ignore')
+            {
+                $read_write_ignore = 1;
+            }
+        }

         # XXX Previously, for subtyping, only the leaf field name is
         # used. Ponder whether we want to keep it that way.

+        # override type-specific read method if read_as is specified
+        if ($read_as_field)
+        {
+            print $rff "\tlocal_node->$f = $read_as_field;\n" unless $no_read;
+            $no_read = 1;
+        }
+
+        # check this after handling read_as
+        if ($read_write_ignore)
+        {
+            next if $no_read;
+            die "$n.$f must not be marked read_write_ignore\n";
+        }
+
         # select instructions by field type
         if ($t eq 'bool')
         {
@@ -712,7 +787,8 @@ _read${n}(void)
         }
         elsif ($t eq 'ParamPathInfo*' && elem 'write_only_req_outer', @a)
         {
-            print $off "\tif (node->$f)\n".
+            print $off "\tappendStringInfoString(str, \" :required_outer \");\n".
+              "\tif (node->$f)\n".
               "\t\toutBitmapset(str, node->$f->ppi_req_outer);\n".
               "\telse\n".
               "\t\toutBitmapset(str, NULL);\n";
@@ -754,7 +830,7 @@ _read${n}(void)
     print $rff "
 \tREAD_DONE();
 }
-" unless $no_read;
+" unless $struct_no_read;
 }

 close $off;
diff --git a/src/include/nodes/nodes.h b/src/include/nodes/nodes.h
index d77aad6473..9166903606 100644
--- a/src/include/nodes/nodes.h
+++ b/src/include/nodes/nodes.h
@@ -590,26 +590,39 @@ typedef enum NodeTag
  * - custom_read_write: Has custom implementations in outfuncs.c and
  *   readfuncs.c.
  *
- * - no_copy_equal: Does not support copyObject() and equal() at all.
+ * - no_copy: Does not support copyObject() at all.
+ *
+ * - no_equal: Does not support equal() at all.
+ *
+ * - no_copy_equal: Shorthand for both no_copy and no_equal.
  *
  * - no_read: Does not support nodeRead() at all.
  *
  * - special_read_write: Has special treatment in outNode() and nodeRead().
  *
+ * Node types can be supertypes of other types whether or not they are marked
+ * abstract: if a node struct appears as the first field of another struct
+ * type, then it is the supertype of that type.  The no_copy, no_equal,
+ * no_copy_equal, and no_read node attributes are automatically inherited
+ * from the supertype.
+ *
  * Valid node field attributes:
  *
  * - array_size(OTHERFIELD): This field is a dynamically allocated array with
  *   size indicated by the mentioned other field.  The other field is either a
  *   scalar or a list, in which case the length of the list is used.
  *
- * - copy_ignore: Ignore the field for copy.
+ * - copy_as(VALUE): In copyObject(), replace the field's value with VALUE.
  *
  * - equal_ignore: Ignore the field for equality.
  *
  * - equal_ignore_if_zero: Ignore the field for equality if it is zero.
  *   (Otherwise, compare normally.)
  *
- * - read_write_ignore: Ignore the field for read/write.
+ * - read_as(VALUE): In nodeRead(), replace the field's value with VALUE.
+ *
+ * - read_write_ignore: Ignore the field for read/write.  This is only allowed
+ *   if the node type is marked no_read or read_as() is also specified.
  *
  * - write_only_relids, write_only_nondefault_pathtarget, write_only_req_outer:
  *   Special handling for Path struct; see there.
diff --git a/src/include/nodes/parsenodes.h b/src/include/nodes/parsenodes.h
index fb026c6b1f..d870d3b09c 100644
--- a/src/include/nodes/parsenodes.h
+++ b/src/include/nodes/parsenodes.h
@@ -127,7 +127,7 @@ typedef struct Query pg_node_attr(custom_read_write)
      * query identifier (can be set by plugins); ignored for equal, might not
      * be set
      */
-    uint64        queryId pg_node_attr(equal_ignore);
+    uint64        queryId pg_node_attr(equal_ignore, read_as(0));

     bool        canSetTag;        /* do I set the command result tag? */

diff --git a/src/include/nodes/pathnodes.h b/src/include/nodes/pathnodes.h
index 4212610d5e..77fb3c0b8a 100644
--- a/src/include/nodes/pathnodes.h
+++ b/src/include/nodes/pathnodes.h
@@ -92,7 +92,7 @@ typedef enum UpperRelationKind
  * the field type.)
  *----------
  */
-typedef struct PlannerGlobal pg_node_attr(no_copy_equal)
+typedef struct PlannerGlobal pg_node_attr(no_copy_equal, no_read)
 {
     NodeTag        type;

@@ -163,7 +163,7 @@ typedef struct PlannerInfo PlannerInfo;
 #define HAVE_PLANNERINFO_TYPEDEF 1
 #endif

-struct PlannerInfo pg_node_attr(no_copy_equal)
+struct PlannerInfo pg_node_attr(no_copy_equal, no_read)
 {
     NodeTag        type;

@@ -376,7 +376,7 @@ struct PlannerInfo pg_node_attr(no_copy_equal)

     /* These fields are used only when hasRecursion is true: */
     int            wt_param_id;    /* PARAM_EXEC ID for the work table */
-    struct Path *non_recursive_path;    /* a path for non-recursive term */
+    struct Path *non_recursive_path pg_node_attr(read_write_ignore);    /* a path for non-recursive term */

     /* These fields are workspace for createplan.c */
     Relids        curOuterRels;    /* outer rels above current node */
@@ -691,7 +691,7 @@ typedef enum RelOptKind
      (rel)->reloptkind == RELOPT_OTHER_JOINREL || \
      (rel)->reloptkind == RELOPT_OTHER_UPPER_REL)

-typedef struct RelOptInfo pg_node_attr(no_copy_equal)
+typedef struct RelOptInfo pg_node_attr(no_copy_equal, no_read)
 {
     NodeTag        type;

@@ -922,7 +922,7 @@ typedef struct IndexOptInfo IndexOptInfo;
 #define HAVE_INDEXOPTINFO_TYPEDEF 1
 #endif

-struct IndexOptInfo pg_node_attr(no_copy_equal)
+struct IndexOptInfo pg_node_attr(no_copy_equal, no_read)
 {
     NodeTag        type;

@@ -1081,7 +1081,7 @@ typedef struct ForeignKeyOptInfo pg_node_attr(custom_read_write, no_copy_equal,
  * Each pg_statistic_ext row is represented by one or more nodes of this
  * type, or even zero if ANALYZE has not computed them.
  */
-typedef struct StatisticExtInfo pg_node_attr(no_copy_equal)
+typedef struct StatisticExtInfo pg_node_attr(no_copy_equal, no_read)
 {
     NodeTag        type;

@@ -1146,6 +1146,10 @@ typedef struct StatisticExtInfo pg_node_attr(no_copy_equal)
  *
  * NB: if ec_merged isn't NULL, this class has been merged into another, and
  * should be ignored in favor of using the pointed-to class.
+ *
+ * NB: EquivalenceClasses are never copied after creation.  Therefore,
+ * copyObject() copies pointers to them as pointers, and equal() compares
+ * pointers to EquivalenceClasses via pointer equality.
  */
 typedef struct EquivalenceClass pg_node_attr(custom_read_write, no_copy_equal, no_read)
 {
@@ -1197,7 +1201,7 @@ typedef struct EquivalenceClass pg_node_attr(custom_read_write, no_copy_equal, n
  * anyarray_ops would never work without this.  Use em_datatype when
  * looking up a specific btree operator to work with this expression.
  */
-typedef struct EquivalenceMember pg_node_attr(no_copy_equal)
+typedef struct EquivalenceMember pg_node_attr(no_copy_equal, no_read)
 {
     NodeTag        type;

@@ -1226,7 +1230,7 @@ typedef struct EquivalenceMember pg_node_attr(no_copy_equal)
  * BTGreaterStrategyNumber (for DESC).  We assume that all ordering-capable
  * index types will use btree-compatible strategy numbers.
  */
-typedef struct PathKey
+typedef struct PathKey pg_node_attr(no_read)
 {
     NodeTag        type;

@@ -1239,7 +1243,7 @@ typedef struct PathKey
 /*
  * Combines information about pathkeys and the associated clauses.
  */
-typedef struct PathKeyInfo
+typedef struct PathKeyInfo pg_node_attr(no_read)
 {
     NodeTag        type;
     List       *pathkeys;
@@ -1320,7 +1324,7 @@ typedef struct PathTarget pg_node_attr(no_copy_equal, no_read)
  * on how the join is formed.  The relevant clauses will appear in each
  * parameterized join path's joinrestrictinfo list, instead.
  */
-typedef struct ParamPathInfo pg_node_attr(no_copy_equal)
+typedef struct ParamPathInfo pg_node_attr(no_copy_equal, no_read)
 {
     NodeTag        type;

@@ -1498,7 +1502,7 @@ typedef struct IndexPath
  * column, i.e. the indexcol values must form a nondecreasing sequence.
  * (The order of multiple clauses for the same index column is unspecified.)
  */
-typedef struct IndexClause pg_node_attr(no_copy_equal)
+typedef struct IndexClause pg_node_attr(no_copy_equal, no_read)
 {
     NodeTag        type;
     struct RestrictInfo *rinfo; /* original restriction or join clause */
@@ -1995,14 +1999,14 @@ typedef struct AggPath
  * Various annotations used for grouping sets in the planner.
  */

-typedef struct GroupingSetData pg_node_attr(no_copy_equal)
+typedef struct GroupingSetData pg_node_attr(no_copy_equal, no_read)
 {
     NodeTag        type;
     List       *set;            /* grouping set as list of sortgrouprefs */
     Cardinality numGroups;        /* est. number of result groups */
 } GroupingSetData;

-typedef struct RollupData pg_node_attr(no_copy_equal)
+typedef struct RollupData pg_node_attr(no_copy_equal, no_read)
 {
     NodeTag        type;
     List       *groupClause;    /* applicable subset of parse->groupClause */
@@ -2275,7 +2279,7 @@ typedef struct LimitPath
  * recursion in plan tree dump.
  */

-typedef struct RestrictInfo
+typedef struct RestrictInfo pg_node_attr(no_read)
 {
     NodeTag        type;

@@ -2370,10 +2374,10 @@ typedef struct RestrictInfo

     /*
      * List of MergeScanSelCache structs.  Those aren't Nodes, so hard to
-     * copy.  Ignoring it will have the effect that copying will just reset
-     * the cache.
+     * copy; instead replace with NIL.  That has the effect that copying will
+     * just reset the cache.  Likewise, can't compare or print them.
      */
-    List       *scansel_cache pg_node_attr(copy_ignore, equal_ignore);
+    List       *scansel_cache pg_node_attr(copy_as(NIL), equal_ignore, read_write_ignore);

     /*
      * transient workspace for use while considering a specific join path; T =
@@ -2543,7 +2547,7 @@ typedef struct SpecialJoinInfo SpecialJoinInfo;
 #define HAVE_SPECIALJOININFO_TYPEDEF 1
 #endif

-struct SpecialJoinInfo
+struct SpecialJoinInfo pg_node_attr(no_read)
 {
     NodeTag        type;
     Relids        min_lefthand;    /* base relids in minimum LHS for join */
@@ -2665,7 +2669,7 @@ typedef struct AppendRelInfo
  * We add such a reference to root->processed_tlist when creating the entry,
  * and it propagates into the plan tree from there.
  */
-typedef struct RowIdentityVarInfo pg_node_attr(no_copy_equal)
+typedef struct RowIdentityVarInfo pg_node_attr(no_copy_equal, no_read)
 {
     NodeTag        type;

@@ -2701,7 +2705,7 @@ typedef struct RowIdentityVarInfo pg_node_attr(no_copy_equal)
  * don't result in unnecessary constraints on join order.
  */

-typedef struct PlaceHolderInfo
+typedef struct PlaceHolderInfo pg_node_attr(no_read)
 {
     NodeTag        type;

@@ -2732,7 +2736,7 @@ typedef struct PlaceHolderInfo
  * function.  MinMaxAggPath contains a list of these, and if we accept that
  * path, the list is stored into root->minmax_aggs for use during setrefs.c.
  */
-typedef struct MinMaxAggInfo pg_node_attr(no_copy_equal)
+typedef struct MinMaxAggInfo pg_node_attr(no_copy_equal, no_read)
 {
     NodeTag        type;

@@ -2808,7 +2812,7 @@ typedef struct MinMaxAggInfo pg_node_attr(no_copy_equal)
  * Instead, we just record the assignment of the slot number by appending to
  * root->glob->paramExecTypes.
  */
-typedef struct PlannerParamItem pg_node_attr(no_copy_equal)
+typedef struct PlannerParamItem pg_node_attr(no_copy_equal, no_read)
 {
     NodeTag        type;

diff --git a/src/include/nodes/plannodes.h b/src/include/nodes/plannodes.h
index 19b5ce2ec6..ab7fd8054b 100644
--- a/src/include/nodes/plannodes.h
+++ b/src/include/nodes/plannodes.h
@@ -38,9 +38,12 @@
  * nodes; in such cases, commandType == CMD_UTILITY, the statement itself
  * is in the utilityStmt field, and the rest of the struct is mostly dummy.
  * (We do use canSetTag, stmt_location, stmt_len, and possibly queryId.)
+ *
+ * PlannedStmt, as well as all varieties of Plan, do not support equal(),
+ * not because it's not sensible but because we currently have no need.
  * ----------------
  */
-typedef struct PlannedStmt
+typedef struct PlannedStmt pg_node_attr(no_equal)
 {
     NodeTag        type;

@@ -108,7 +111,7 @@ typedef struct PlannedStmt
  * abstract superclass for all Plan-type nodes.
  * ----------------
  */
-typedef struct Plan
+typedef struct Plan pg_node_attr(abstract, no_equal)
 {
     NodeTag        type;

@@ -725,6 +728,12 @@ typedef struct CustomScan
     List       *custom_private; /* private data for custom code */
     List       *custom_scan_tlist;    /* optional tlist describing scan tuple */
     Bitmapset  *custom_relids;    /* RTIs generated by this scan */
+
+    /*
+     * NOTE: The method field of CustomScan is required to be a pointer to a
+     * static table of callback functions.  So we don't copy the table itself,
+     * just reference the original one.
+     */
     const struct CustomScanMethods *methods;
 } CustomScan;

@@ -756,7 +765,7 @@ typedef struct CustomScan
  * the joinquals, ignoring plan.qual, due to where the executor tests it.)
  * ----------------
  */
-typedef struct Join
+typedef struct Join pg_node_attr(abstract)
 {
     Plan        plan;
     JoinType    jointype;
@@ -781,7 +790,7 @@ typedef struct NestLoop
     List       *nestParams;        /* list of NestLoopParam nodes */
 } NestLoop;

-typedef struct NestLoopParam
+typedef struct NestLoopParam pg_node_attr(no_equal)
 {
     NodeTag        type;
     int            paramno;        /* number of the PARAM_EXEC Param to set */
@@ -1343,7 +1352,7 @@ typedef enum RowMarkType
  * Note this means that all tables in an inheritance hierarchy share the
  * same resjunk column names.
  */
-typedef struct PlanRowMark
+typedef struct PlanRowMark pg_node_attr(no_equal)
 {
     NodeTag        type;
     Index        rti;            /* range table index of markable relation */
@@ -1387,7 +1396,7 @@ typedef struct PlanRowMark
  *                        by any of the PartitionedRelPruneInfo nodes in
  *                        "prune_infos".  These subplans must not be pruned.
  */
-typedef struct PartitionPruneInfo
+typedef struct PartitionPruneInfo pg_node_attr(no_equal)
 {
     NodeTag        type;
     List       *prune_infos;
@@ -1411,7 +1420,7 @@ typedef struct PartitionPruneInfo
  * node, but partition indexes are valid only within a particular hierarchy.
  * relid_map[p] contains the partition's OID, or 0 if the partition was pruned.
  */
-typedef struct PartitionedRelPruneInfo
+typedef struct PartitionedRelPruneInfo pg_node_attr(no_equal)
 {
     NodeTag        type;

@@ -1485,7 +1494,7 @@ typedef struct PartitionPruneStep pg_node_attr(abstract)
  * have an expression be present in 'exprs' for a given partition key and
  * the corresponding bit set in 'nullkeys'.
  */
-typedef struct PartitionPruneStepOp
+typedef struct PartitionPruneStepOp pg_node_attr(no_equal)
 {
     PartitionPruneStep step;

@@ -1507,7 +1516,7 @@ typedef enum PartitionPruneCombineOp
     PARTPRUNE_COMBINE_INTERSECT
 } PartitionPruneCombineOp;

-typedef struct PartitionPruneStepCombine
+typedef struct PartitionPruneStepCombine pg_node_attr(no_equal)
 {
     PartitionPruneStep step;

@@ -1525,7 +1534,7 @@ typedef struct PartitionPruneStepCombine
  * to be used with the syscache invalidation mechanism, so it identifies a
  * system catalog entry by cache ID and hash value.
  */
-typedef struct PlanInvalItem
+typedef struct PlanInvalItem pg_node_attr(no_equal)
 {
     NodeTag        type;
     int            cacheId;        /* a syscache ID, see utils/syscache.h */
diff --git a/src/include/nodes/primnodes.h b/src/include/nodes/primnodes.h
index 9e6b4bdb1d..6f3dcc74c3 100644
--- a/src/include/nodes/primnodes.h
+++ b/src/include/nodes/primnodes.h
@@ -68,7 +68,7 @@ typedef struct RangeVar
      * the catalog (database) name, or NULL; ignored for read/write, since it
      * is presently not semantically meaningful
      */
-    char       *catalogname pg_node_attr(read_write_ignore);
+    char       *catalogname pg_node_attr(read_write_ignore, read_as(NULL));

     /* the schema name, or NULL */
     char       *schemaname;
@@ -636,6 +636,7 @@ typedef struct NamedArgExpr
  * Note that opfuncid is not necessarily filled in immediately on creation
  * of the node.  The planner makes sure it is valid before passing the node
  * tree to the executor, but during parsing/planning opfuncid can be 0.
+ * Therefore, equal() will accept a zero value as being equal to other values.
  */
 typedef struct OpExpr
 {
diff --git a/src/include/utils/rel.h b/src/include/utils/rel.h
index 3e6da86688..7b60450b26 100644
--- a/src/include/utils/rel.h
+++ b/src/include/utils/rel.h
@@ -265,7 +265,7 @@ typedef struct RelationData
  * Currently, we mostly cache fields of interest to the planner, but the set
  * of fields has already grown the constraint OID for other uses.
  */
-typedef struct ForeignKeyCacheInfo pg_node_attr(no_read)
+typedef struct ForeignKeyCacheInfo pg_node_attr(no_equal, no_read)
 {
     NodeTag        type;
     Oid            conoid;            /* oid of the constraint itself */
diff --git a/src/test/modules/test_oat_hooks/test_oat_hooks.c b/src/test/modules/test_oat_hooks/test_oat_hooks.c
index 7ef272cc7a..b648ee67ff 100644
--- a/src/test/modules/test_oat_hooks/test_oat_hooks.c
+++ b/src/test/modules/test_oat_hooks/test_oat_hooks.c
@@ -468,9 +468,6 @@ nodetag_to_string(NodeTag tag)
         case T_TupleTableSlot:
             return "TupleTableSlot";
             break;
-        case T_Plan:
-            return "Plan";
-            break;
         case T_Result:
             return "Result";
             break;
@@ -549,9 +546,6 @@ nodetag_to_string(NodeTag tag)
         case T_CustomScan:
             return "CustomScan";
             break;
-        case T_Join:
-            return "Join";
-            break;
         case T_NestLoop:
             return "NestLoop";
             break;

pgsql-hackers by date:

Previous
From: Thomas Munro
Date:
Subject: Re: Bump MIN_WINNT to 0x0600 (Vista) as minimal runtime in 16~
Next
From: Thomas Munro
Date:
Subject: Re: EINTR in ftruncate()