v47-0005-Row-pattern-recognition-patch-executor-and-comma.patch

application/octet-stream

Filename: v47-0005-Row-pattern-recognition-patch-executor-and-comma.patch
Type: application/octet-stream
Part: 4
Message: Re: Row pattern recognition
From 9a401a09b56b18bd08ca576ba44b4017be231fc2 Mon Sep 17 00:00:00 2001
From: Tatsuo Ishii <ishii@postgresql.org>
Date: Sat, 2 May 2026 13:40:29 +0900
Subject: [PATCH v47 5/9] Row pattern recognition patch (executor and
 commands).

---
 src/backend/commands/explain.c        |  522 ++++++++
 src/backend/executor/Makefile         |    1 +
 src/backend/executor/README.rpr       | 1578 ++++++++++++++++++++++
 src/backend/executor/execExpr.c       |   92 ++
 src/backend/executor/execExprInterp.c |  267 ++++
 src/backend/executor/execRPR.c        | 1772 +++++++++++++++++++++++++
 src/backend/executor/meson.build      |    1 +
 src/backend/executor/nodeWindowAgg.c  | 1066 ++++++++++++++-
 src/backend/jit/llvm/llvmjit_expr.c   |   78 +-
 src/backend/jit/llvm/llvmjit_types.c  |    2 +
 src/backend/utils/adt/windowfuncs.c   |  119 +-
 src/include/catalog/pg_proc.dat       |   24 +
 src/include/executor/execExpr.h       |   20 +
 src/include/executor/execRPR.h        |   40 +
 src/include/executor/nodeWindowAgg.h  |    3 +
 src/include/nodes/execnodes.h         |  130 ++
 16 files changed, 5703 insertions(+), 12 deletions(-)
 create mode 100644 src/backend/executor/README.rpr
 create mode 100644 src/backend/executor/execRPR.c
 create mode 100644 src/include/executor/execRPR.h

diff --git a/src/backend/commands/explain.c b/src/backend/commands/explain.c
index 112c17b0d64..99de36b57f2 100644
--- a/src/backend/commands/explain.c
+++ b/src/backend/commands/explain.c
@@ -30,6 +30,7 @@
 #include "nodes/extensible.h"
 #include "nodes/makefuncs.h"
 #include "nodes/nodeFuncs.h"
+#include "optimizer/rpr.h"
 #include "parser/analyze.h"
 #include "parser/parsetree.h"
 #include "rewrite/rewriteHandler.h"
@@ -119,6 +120,20 @@ static void show_window_def(WindowAggState *planstate,
 static void show_window_keys(StringInfo buf, PlanState *planstate,
 							 int nkeys, AttrNumber *keycols,
 							 List *ancestors, ExplainState *es);
+static void append_rpr_quantifier(StringInfo buf, RPRPatternElement *elem);
+static char *deparse_rpr_pattern(RPRPattern *pattern);
+static void deparse_rpr_elements(RPRPattern *pattern, int *idx,
+								 StringInfoData *buf, RPRDepth groupDepth,
+								 RPRDepth *prevDepth, bool *needSpace);
+static void deparse_rpr_group(RPRPattern *pattern, int *idx,
+							  StringInfoData *buf, RPRDepth *prevDepth,
+							  bool *needSpace);
+static void deparse_rpr_alt(RPRPattern *pattern, int *idx,
+							StringInfoData *buf, RPRDepth *prevDepth,
+							bool *needSpace, List **altSeps);
+static void deparse_rpr_var(RPRPattern *pattern, int *idx,
+							StringInfoData *buf, RPRDepth *prevDepth,
+							bool *needSpace, List **altSeps);
 static void show_storage_info(char *maxStorageType, int64 maxSpaceUsed,
 							  ExplainState *es);
 static void show_tablesample(TableSampleClause *tsc, PlanState *planstate,
@@ -129,6 +144,7 @@ static void show_incremental_sort_info(IncrementalSortState *incrsortstate,
 static void show_hash_info(HashState *hashstate, ExplainState *es);
 static void show_material_info(MaterialState *mstate, ExplainState *es);
 static void show_windowagg_info(WindowAggState *winstate, ExplainState *es);
+static void show_rpr_nfa_stats(WindowAggState *winstate, ExplainState *es);
 static void show_ctescan_info(CteScanState *ctescanstate, ExplainState *es);
 static void show_table_func_scan_info(TableFuncScanState *tscanstate,
 									  ExplainState *es);
@@ -2898,6 +2914,284 @@ show_sortorder_options(StringInfo buf, Node *sortexpr,
 	}
 }
 
+/*
+ * Append quantifier suffix for a pattern element.
+ */
+static void
+append_rpr_quantifier(StringInfo buf, RPRPatternElement *elem)
+{
+	/* Append quantifier if not {1,1} */
+	if (elem->min == 0 && elem->max == RPR_QUANTITY_INF)
+		appendStringInfoChar(buf, '*');
+	else if (elem->min == 1 && elem->max == RPR_QUANTITY_INF)
+		appendStringInfoChar(buf, '+');
+	else if (elem->min == 0 && elem->max == 1)
+		appendStringInfoChar(buf, '?');
+	else if (elem->max == RPR_QUANTITY_INF)
+		appendStringInfo(buf, "{%d,}", elem->min);
+	else if (elem->min == elem->max && elem->min != 1)
+		appendStringInfo(buf, "{%d}", elem->min);
+	else if (elem->min != 1 || elem->max != 1)
+		appendStringInfo(buf, "{%d,%d}", elem->min, elem->max);
+
+	if (RPRElemIsReluctant(elem))
+	{
+		if (elem->min == 1 && elem->max == 1)
+			appendStringInfo(buf, "{1}");	/* make reluctant ? unambiguous */
+		appendStringInfoChar(buf, '?');
+	}
+
+	/* Append absorption markers: " for judgment point, ' for branch only */
+	if (RPRElemIsAbsorbable(elem))
+	{
+		Assert(elem->max == RPR_QUANTITY_INF);
+		appendStringInfoChar(buf, '"');
+	}
+	else if (RPRElemIsAbsorbableBranch(elem))
+		appendStringInfoChar(buf, '\'');
+}
+
+/*
+ * Deparse a compiled RPRPattern (bytecode) back to pattern string.
+ *
+ * Walks the flat bytecode array using mutual recursion: deparse_rpr_elements
+ * processes sequential elements, and deparse_rpr_group handles BEGIN...END
+ * groups by recursing back into deparse_rpr_elements for the group content.
+ */
+static char *
+deparse_rpr_pattern(RPRPattern *pattern)
+{
+	StringInfoData buf;
+	int			idx = 0;
+	RPRDepth	prevDepth = 0;
+	bool		needSpace = false;
+
+	Assert(pattern != NULL && pattern->numElements >= 2);
+
+	initStringInfo(&buf);
+
+	deparse_rpr_elements(pattern, &idx, &buf, RPR_DEPTH_NONE,
+						 &prevDepth, &needSpace);
+
+	/* Close remaining open parens */
+	while (prevDepth > 0)
+	{
+		appendStringInfoChar(&buf, ')');
+		prevDepth--;
+	}
+
+	return buf.data;
+}
+
+/*
+ * Process pattern elements sequentially until FIN or END at groupDepth.
+ *
+ * When groupDepth >= 0, stops at the matching END element (leaving idx
+ * pointing to it) so the caller (deparse_rpr_group) can consume it.
+ * When groupDepth < 0, processes until FIN (top-level call).
+ */
+static void
+deparse_rpr_elements(RPRPattern *pattern, int *idx, StringInfoData *buf,
+					 RPRDepth groupDepth, RPRDepth *prevDepth,
+					 bool *needSpace)
+{
+	List	   *altSeps = NIL;	/* pending alternation separator indices */
+
+	while (*idx < pattern->numElements)
+	{
+		RPRPatternElement *elem = &pattern->elements[*idx];
+
+		if (RPRElemIsFin(elem))
+			break;
+
+		/* Stop at END matching our group depth; caller handles it */
+		if (RPRElemIsEnd(elem) && elem->depth == groupDepth)
+			break;
+
+		/* Alternation separator */
+		if (list_member_int(altSeps, *idx))
+		{
+			/* Close parens to match separator depth first */
+			while (*prevDepth > elem->depth)
+			{
+				appendStringInfoChar(buf, ')');
+				(*prevDepth)--;
+			}
+			appendStringInfoString(buf, " | ");
+			*needSpace = false;
+			altSeps = list_delete_int(altSeps, *idx);
+		}
+
+		/* Dispatch to element-type handlers */
+		if (RPRElemIsAlt(elem))
+			deparse_rpr_alt(pattern, idx, buf, prevDepth,
+							needSpace, &altSeps);
+		else if (RPRElemIsBegin(elem))
+			deparse_rpr_group(pattern, idx, buf, prevDepth,
+							  needSpace);
+		else if (RPRElemIsVar(elem))
+			deparse_rpr_var(pattern, idx, buf, prevDepth,
+							needSpace, &altSeps);
+	}
+	list_free(altSeps);
+}
+
+/*
+ * Process a BEGIN...END group.
+ *
+ * Consumes BEGIN, recurses into deparse_rpr_elements for group content,
+ * then consumes END and outputs the group quantifier.
+ *
+ * When the group wraps a single ALT with no siblings, the group-level
+ * parenthesis is suppressed since the ALT-to-children depth transition
+ * already provides it (avoids double parens like "((a | b))+").
+ */
+static void
+deparse_rpr_group(RPRPattern *pattern, int *idx, StringInfoData *buf,
+				  RPRDepth *prevDepth, bool *needSpace)
+{
+	RPRPatternElement *begin = &pattern->elements[*idx];
+	RPRDepth	childDepth = begin->depth + 1;
+	bool		singleAlt = false;
+	RPRPatternElement *end;
+
+	/*
+	 * Check if this group wraps a single ALT with no siblings. Scan from
+	 * after ALT to END: if no element at childDepth exists, the ALT is the
+	 * sole child.
+	 */
+	if (*idx + 1 < pattern->numElements &&
+		RPRElemIsAlt(&pattern->elements[*idx + 1]))
+	{
+		int			j;
+
+		singleAlt = true;
+		for (j = *idx + 2; j < pattern->numElements; j++)
+		{
+			RPRPatternElement *e = &pattern->elements[j];
+
+			if (RPRElemIsEnd(e) && e->depth == begin->depth)
+				break;
+			if (e->depth <= childDepth)
+			{
+				singleAlt = false;
+				break;
+			}
+		}
+	}
+
+	/* Open group paren (unless single ALT provides it) */
+	if (!singleAlt)
+	{
+		if (*needSpace)
+			appendStringInfoChar(buf, ' ');
+		appendStringInfoChar(buf, '(');
+		*needSpace = false;
+	}
+	*prevDepth = childDepth;
+	(*idx)++;					/* consume BEGIN */
+
+	/* Process group children; stops at matching END */
+	deparse_rpr_elements(pattern, idx, buf, begin->depth,
+						 prevDepth, needSpace);
+
+	/* Consume END and output quantifier */
+	Assert(*idx < pattern->numElements);
+	end = &pattern->elements[*idx];
+	Assert(RPRElemIsEnd(end) && end->depth == begin->depth);
+
+	while (*prevDepth > end->depth + 1)
+	{
+		appendStringInfoChar(buf, ')');
+		(*prevDepth)--;
+	}
+	if (!singleAlt)
+		appendStringInfoChar(buf, ')');
+	append_rpr_quantifier(buf, end);
+	*prevDepth = end->depth;
+	*needSpace = true;
+	(*idx)++;					/* consume END */
+}
+
+/*
+ * Process an ALT element: adjust depth parens and register separator positions.
+ */
+static void
+deparse_rpr_alt(RPRPattern *pattern, int *idx, StringInfoData *buf,
+				RPRDepth *prevDepth, bool *needSpace, List **altSeps)
+{
+	RPRPatternElement *elem = &pattern->elements[*idx];
+
+	/* Close parens for depth decrease */
+	while (*prevDepth > elem->depth)
+	{
+		appendStringInfoChar(buf, ')');
+		(*prevDepth)--;
+		*needSpace = true;
+	}
+
+	/* Open parens up to ALT's depth */
+	while (*prevDepth < elem->depth)
+	{
+		if (*needSpace)
+			appendStringInfoChar(buf, ' ');
+		appendStringInfoChar(buf, '(');
+		(*prevDepth)++;
+		*needSpace = false;
+	}
+
+	/* Register next alternation separator position */
+	if (elem->next != RPR_ELEMIDX_INVALID)
+	{
+		RPRPatternElement *firstElem = &pattern->elements[elem->next];
+
+		if (firstElem->jump != RPR_ELEMIDX_INVALID)
+			*altSeps = lappend_int(*altSeps, firstElem->jump);
+	}
+	if (elem->jump != RPR_ELEMIDX_INVALID)
+		*altSeps = lappend_int(*altSeps, elem->jump);
+	(*idx)++;
+}
+
+/*
+ * Process a VAR element: adjust depth parens and output variable name.
+ */
+static void
+deparse_rpr_var(RPRPattern *pattern, int *idx, StringInfoData *buf,
+				RPRDepth *prevDepth, bool *needSpace, List **altSeps)
+{
+	RPRPatternElement *elem = &pattern->elements[*idx];
+
+	/* Open parens for depth increase */
+	while (*prevDepth < elem->depth)
+	{
+		if (*needSpace)
+			appendStringInfoChar(buf, ' ');
+		appendStringInfoChar(buf, '(');
+		(*prevDepth)++;
+		*needSpace = false;
+	}
+
+	/* Close parens for depth decrease */
+	while (*prevDepth > elem->depth)
+	{
+		appendStringInfoChar(buf, ')');
+		(*prevDepth)--;
+	}
+
+	if (*needSpace)
+		appendStringInfoChar(buf, ' ');
+
+	Assert(elem->varId < pattern->numVars);
+	appendStringInfoString(buf, quote_identifier(pattern->varNames[elem->varId]));
+	append_rpr_quantifier(buf, elem);
+	*needSpace = true;
+
+	if (elem->jump != RPR_ELEMIDX_INVALID)
+		*altSeps = lappend_int(*altSeps, elem->jump);
+	(*idx)++;
+}
+
 /*
  * Show the window definition for a WindowAgg node.
  */
@@ -2956,6 +3250,79 @@ show_window_def(WindowAggState *planstate, List *ancestors, ExplainState *es)
 	appendStringInfoChar(&wbuf, ')');
 	ExplainPropertyText("Window", wbuf.data, es);
 	pfree(wbuf.data);
+
+	/* Show Row Pattern Recognition pattern if present */
+	if (wagg->rpPattern != NULL)
+	{
+		char	   *patternStr = deparse_rpr_pattern(wagg->rpPattern);
+
+		if (patternStr != NULL)
+		{
+			ExplainPropertyText("Pattern", patternStr, es);
+			pfree(patternStr);
+		}
+
+		/*
+		 * Show navigation offsets for tuplestore trim.  For EXPLAIN ANALYZE,
+		 * use the executor-resolved values (which may differ from the plan
+		 * when NEEDS_EVAL was resolved to FIXED or RETAIN_ALL at init).
+		 */
+		{
+			RPRNavOffsetKind maxKind = wagg->navMaxOffsetKind;
+			int64		maxOffset = wagg->navMaxOffset;
+			RPRNavOffsetKind firstKind = wagg->navFirstOffsetKind;
+			int64		firstOffset = wagg->navFirstOffset;
+
+			if (es->analyze)
+			{
+				maxKind = planstate->navMaxOffsetKind;
+				maxOffset = planstate->navMaxOffset;
+				firstKind = planstate->navFirstOffsetKind;
+				firstOffset = planstate->navFirstOffset;
+			}
+
+			switch (maxKind)
+			{
+				case RPR_NAV_OFFSET_NEEDS_EVAL:
+					ExplainPropertyText("Nav Mark Lookback", "runtime", es);
+					break;
+				case RPR_NAV_OFFSET_RETAIN_ALL:
+					ExplainPropertyText("Nav Mark Lookback", "retain all", es);
+					break;
+				case RPR_NAV_OFFSET_FIXED:
+					ExplainPropertyInteger("Nav Mark Lookback", NULL,
+										   maxOffset, es);
+					break;
+				default:
+					elog(ERROR, "unrecognized RPR nav offset kind: %d",
+						 maxKind);
+					break;
+			}
+
+			if (wagg->hasFirstNav)
+			{
+				switch (firstKind)
+				{
+					case RPR_NAV_OFFSET_NEEDS_EVAL:
+						ExplainPropertyText("Nav Mark Lookahead", "runtime",
+											es);
+						break;
+					case RPR_NAV_OFFSET_RETAIN_ALL:
+						ExplainPropertyText("Nav Mark Lookahead", "retain all",
+											es);
+						break;
+					case RPR_NAV_OFFSET_FIXED:
+						ExplainPropertyInteger("Nav Mark Lookahead", NULL,
+											   firstOffset, es);
+						break;
+					default:
+						elog(ERROR, "unrecognized RPR nav offset kind: %d",
+							 firstKind);
+						break;
+				}
+			}
+		}
+	}
 }
 
 /*
@@ -3508,6 +3875,7 @@ show_windowagg_info(WindowAggState *winstate, ExplainState *es)
 {
 	char	   *maxStorageType;
 	int64		maxSpaceUsed;
+	WindowAgg  *wagg = (WindowAgg *) winstate->ss.ps.plan;
 
 	Tuplestorestate *tupstore = winstate->buffer;
 
@@ -3520,6 +3888,160 @@ show_windowagg_info(WindowAggState *winstate, ExplainState *es)
 
 	tuplestore_get_stats(tupstore, &maxStorageType, &maxSpaceUsed);
 	show_storage_info(maxStorageType, maxSpaceUsed, es);
+
+	/* Show NFA statistics for Row Pattern Recognition */
+	if (wagg->rpPattern != NULL)
+		show_rpr_nfa_stats(winstate, es);
+}
+
+/*
+ * Show NFA statistics for Row Pattern Recognition on WindowAgg node.
+ */
+static void
+show_rpr_nfa_stats(WindowAggState *winstate, ExplainState *es)
+{
+	if (es->format != EXPLAIN_FORMAT_TEXT)
+	{
+		/* State and context counters */
+		ExplainPropertyInteger("NFA States Peak", NULL, winstate->nfaStatesMax, es);
+		ExplainPropertyInteger("NFA States Total", NULL, winstate->nfaStatesTotalCreated, es);
+		ExplainPropertyInteger("NFA States Merged", NULL, winstate->nfaStatesMerged, es);
+		ExplainPropertyInteger("NFA Contexts Peak", NULL, winstate->nfaContextsMax, es);
+		ExplainPropertyInteger("NFA Contexts Total", NULL, winstate->nfaContextsTotalCreated, es);
+		ExplainPropertyInteger("NFA Contexts Absorbed", NULL, winstate->nfaContextsAbsorbed, es);
+		ExplainPropertyInteger("NFA Contexts Skipped", NULL, winstate->nfaContextsSkipped, es);
+		ExplainPropertyInteger("NFA Contexts Pruned", NULL, winstate->nfaContextsPruned, es);
+
+		/* Match/mismatch counts and length statistics */
+		ExplainPropertyInteger("NFA Matched", NULL, winstate->nfaMatchesSucceeded, es);
+		ExplainPropertyInteger("NFA Mismatched", NULL, winstate->nfaMatchesFailed, es);
+		if (winstate->nfaMatchesSucceeded > 0)
+		{
+			ExplainPropertyInteger("NFA Match Length Min", NULL, winstate->nfaMatchLen.min, es);
+			ExplainPropertyInteger("NFA Match Length Max", NULL, winstate->nfaMatchLen.max, es);
+			ExplainPropertyFloat("NFA Match Length Avg", NULL,
+								 (double) winstate->nfaMatchLen.total / winstate->nfaMatchesSucceeded, 1,
+								 es);
+		}
+		if (winstate->nfaMatchesFailed > 0)
+		{
+			ExplainPropertyInteger("NFA Mismatch Length Min", NULL, winstate->nfaFailLen.min, es);
+			ExplainPropertyInteger("NFA Mismatch Length Max", NULL, winstate->nfaFailLen.max, es);
+			ExplainPropertyFloat("NFA Mismatch Length Avg", NULL,
+								 (double) winstate->nfaFailLen.total / winstate->nfaMatchesFailed, 1,
+								 es);
+		}
+
+		/* Absorbed/skipped context length statistics */
+		if (winstate->nfaContextsAbsorbed > 0)
+		{
+			ExplainPropertyInteger("NFA Absorbed Length Min", NULL, winstate->nfaAbsorbedLen.min, es);
+			ExplainPropertyInteger("NFA Absorbed Length Max", NULL, winstate->nfaAbsorbedLen.max, es);
+			ExplainPropertyFloat("NFA Absorbed Length Avg", NULL,
+								 (double) winstate->nfaAbsorbedLen.total / winstate->nfaContextsAbsorbed, 1,
+								 es);
+		}
+		if (winstate->nfaContextsSkipped > 0)
+		{
+			ExplainPropertyInteger("NFA Skipped Length Min", NULL, winstate->nfaSkippedLen.min, es);
+			ExplainPropertyInteger("NFA Skipped Length Max", NULL, winstate->nfaSkippedLen.max, es);
+			ExplainPropertyFloat("NFA Skipped Length Avg", NULL,
+								 (double) winstate->nfaSkippedLen.total / winstate->nfaContextsSkipped, 1,
+								 es);
+		}
+	}
+	else
+	{
+		/* State and context counters */
+		ExplainIndentText(es);
+		appendStringInfo(es->str,
+						 "NFA States: " INT64_FORMAT " peak, " INT64_FORMAT " total, " INT64_FORMAT " merged\n",
+						 winstate->nfaStatesMax,
+						 winstate->nfaStatesTotalCreated,
+						 winstate->nfaStatesMerged);
+		ExplainIndentText(es);
+		appendStringInfo(es->str,
+						 "NFA Contexts: " INT64_FORMAT " peak, " INT64_FORMAT " total, " INT64_FORMAT " pruned\n",
+						 winstate->nfaContextsMax,
+						 winstate->nfaContextsTotalCreated,
+						 winstate->nfaContextsPruned);
+
+		/* Match/mismatch counts with length min/max/avg */
+		ExplainIndentText(es);
+		appendStringInfo(es->str, "NFA: ");
+		if (winstate->nfaMatchesSucceeded > 0)
+		{
+			double		avgLen = (double) winstate->nfaMatchLen.total / winstate->nfaMatchesSucceeded;
+
+			appendStringInfo(es->str,
+							 INT64_FORMAT " matched (len " INT64_FORMAT "/" INT64_FORMAT "/%.1f)",
+							 winstate->nfaMatchesSucceeded,
+							 winstate->nfaMatchLen.min,
+							 winstate->nfaMatchLen.max,
+							 avgLen);
+		}
+		else
+		{
+			appendStringInfo(es->str, "0 matched");
+		}
+		if (winstate->nfaMatchesFailed > 0)
+		{
+			double		avgFail = (double) winstate->nfaFailLen.total / winstate->nfaMatchesFailed;
+
+			appendStringInfo(es->str,
+							 ", " INT64_FORMAT " mismatched (len " INT64_FORMAT "/" INT64_FORMAT "/%.1f)",
+							 winstate->nfaMatchesFailed,
+							 winstate->nfaFailLen.min,
+							 winstate->nfaFailLen.max,
+							 avgFail);
+		}
+		else
+		{
+			appendStringInfo(es->str, ", 0 mismatched");
+		}
+		appendStringInfoChar(es->str, '\n');
+
+		/* Absorbed/skipped context length statistics */
+		if (winstate->nfaContextsAbsorbed > 0 || winstate->nfaContextsSkipped > 0)
+		{
+			ExplainIndentText(es);
+			appendStringInfo(es->str, "NFA: ");
+
+			if (winstate->nfaContextsAbsorbed > 0)
+			{
+				double		avgAbsorbed = (double) winstate->nfaAbsorbedLen.total / winstate->nfaContextsAbsorbed;
+
+				appendStringInfo(es->str,
+								 INT64_FORMAT " absorbed (len " INT64_FORMAT "/" INT64_FORMAT "/%.1f)",
+								 winstate->nfaContextsAbsorbed,
+								 winstate->nfaAbsorbedLen.min,
+								 winstate->nfaAbsorbedLen.max,
+								 avgAbsorbed);
+			}
+			else
+			{
+				appendStringInfo(es->str, "0 absorbed");
+			}
+
+			if (winstate->nfaContextsSkipped > 0)
+			{
+				double		avgSkipped = (double) winstate->nfaSkippedLen.total / winstate->nfaContextsSkipped;
+
+				appendStringInfo(es->str,
+								 ", " INT64_FORMAT " skipped (len " INT64_FORMAT "/" INT64_FORMAT "/%.1f)",
+								 winstate->nfaContextsSkipped,
+								 winstate->nfaSkippedLen.min,
+								 winstate->nfaSkippedLen.max,
+								 avgSkipped);
+			}
+			else
+			{
+				appendStringInfo(es->str, ", 0 skipped");
+			}
+
+			appendStringInfoChar(es->str, '\n');
+		}
+	}
 }
 
 /*
diff --git a/src/backend/executor/Makefile b/src/backend/executor/Makefile
index 11118d0ce02..2b257427795 100644
--- a/src/backend/executor/Makefile
+++ b/src/backend/executor/Makefile
@@ -25,6 +25,7 @@ OBJS = \
 	execParallel.o \
 	execPartition.o \
 	execProcnode.o \
+	execRPR.o \
 	execReplication.o \
 	execSRF.o \
 	execScan.o \
diff --git a/src/backend/executor/README.rpr b/src/backend/executor/README.rpr
new file mode 100644
index 00000000000..52bcd77390c
--- /dev/null
+++ b/src/backend/executor/README.rpr
@@ -0,0 +1,1578 @@
+============================================================================
+  PostgreSQL Row Pattern Recognition: Flat-Array Stream NFA Guide
+============================================================================
+
+  Target audience: Developers with a basic understanding of the PostgreSQL
+                   executor and planner architecture
+
+  Scope: The entire process from PATTERN/DEFINE clause parsing to NFA
+         runtime execution
+
+  Related code:
+    - src/backend/parser/parse_rpr.c          (parser phase)
+    - src/backend/optimizer/plan/rpr.c        (optimizer phase)
+    - src/backend/executor/nodeWindowAgg.c    (executor phase, window agg)
+    - src/backend/executor/execRPR.c          (executor phase, NFA engine)
+    - src/include/executor/execRPR.h          (NFA public API)
+    - src/include/nodes/plannodes.h           (plan node definitions)
+    - src/include/nodes/execnodes.h           (execution state definitions)
+    - src/include/optimizer/rpr.h             (types and constants)
+    - src/backend/optimizer/plan/createplan.c (nav offset computation)
+
+============================================================================
+
+What is a Flat-Array Stream NFA?
+
+  The NFA in this implementation is not a traditional state-transition graph
+  but a flat array of fixed-size 16-byte elements. At runtime, it processes
+  the row stream in a forward-only manner, expanding epsilon transitions
+  eagerly without backtracking.
+
+  - Flat-Array: Pattern compiled into a flat array,
+                not a graph (Chapter IV)
+  - Stream:     Rows consumed sequentially in one direction,
+                never revisited (Chapter XII)
+  - NFA:        Nondeterministic execution where multiple states
+                coexist within a single context (Chapter VI)
+
+Chapter I  Row Pattern Recognition Overview
+============================================================================
+
+Row Pattern Recognition (hereafter RPR) is a feature introduced in SQL:2016
+that matches regex-based patterns against ordered row sets.
+
+The SQL standard defines two forms:
+
+  Feature R010: MATCH_RECOGNIZE (FROM clause)
+    - Dedicated table operator
+    - Provides dedicated functions such as MATCH_NUMBER(), CLASSIFIER()
+    - Supports ONE ROW PER MATCH / ALL ROWS PER MATCH
+
+  Feature R020: RPR in a window (WINDOW clause)
+    - Integrated into the existing window function framework
+    - Supports ALL ROWS PER MATCH only
+    - No MATCH_NUMBER()
+
+This implementation targets Feature R020.
+
+The basic syntax is as follows:
+
+  SELECT ...
+  OVER (
+    PARTITION BY ...
+    ORDER BY ...
+    ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+    [INITIAL | SEEK]   -- SEEK is defined in the standard but not implemented
+    AFTER MATCH SKIP TO NEXT ROW | SKIP PAST LAST ROW
+    PATTERN ( <regex> )
+    DEFINE <variable> AS <condition>, ...
+  )
+
+The PATTERN clause is a regular expression over row pattern variables.
+The DEFINE clause specifies boolean conditions that determine whether each
+variable evaluates to true for the current row.
+
+Example:
+
+  PATTERN (A+ B)
+  DEFINE A AS price > PREV(price),
+         B AS price < PREV(price)
+
+This pattern matches "a span where prices rise consecutively then drop."
+
+Chapter II  Overall Processing Pipeline
+============================================================================
+
+RPR processing is divided into three phases:
+
+  +------------------------------------------------------------+
+  |  1. Parsing (Parser)                                       |
+  |     SQL text -> PATTERN AST + DEFINE expression tree       |
+  |                                                            |
+  |  2. Compilation (Optimizer/Planner)                        |
+  |     PATTERN AST -> optimization -> flat NFA element array  |
+  |                                                            |
+  |  3. Execution (Executor)                                   |
+  |     Row-by-row matching via NFA simulation                 |
+  +------------------------------------------------------------+
+
+Each phase uses independent data structures, and the interfaces between
+phases are well-defined:
+
+  Parser -> Planner:    WindowClause.rpPattern (RPRPatternNode tree)
+                        WindowClause.defineClause (TargetEntry list)
+
+  Planner -> Executor:  WindowAgg.rpPattern (RPRPattern struct)
+                        WindowAgg.defineClause (TargetEntry list)
+
+Chapter III  Parsing Phase
+============================================================================
+
+III-1. Entry Point
+
+  transformWindowDefinitions() (parse_clause.c)
+    +-- transformRPR() (parse_rpr.c)
+
+transformRPR() is invoked when RPCommonSyntax is present and performs the
+following:
+
+  (1) Frame option validation
+      - Only ROWS is allowed (RANGE, GROUPS are not)
+      - The start boundary must be CURRENT ROW
+      - EXCLUDE option is not allowed
+
+  (2) Transcription to WindowClause
+      - Copies rpPattern, rpSkipTo, initial fields
+
+  (3) DEFINE clause transformation (transformDefineClause)
+
+III-2. PATTERN AST
+
+The parser transforms the PATTERN clause into an RPRPatternNode tree.
+Each node has one of the following four types:
+
+  RPR_PATTERN_VAR    Variable reference. Name stored in varName field.
+  RPR_PATTERN_SEQ    Concatenation. Children node list in children.
+  RPR_PATTERN_ALT    Alternation. Branch node list in children.
+  RPR_PATTERN_GROUP  Group (parentheses). Body node list in children.
+
+All nodes have min/max fields to express quantifiers:
+
+  A       -> VAR(A, min=1, max=1)
+  A+      -> VAR(A, min=1, max=INF)
+  A*      -> VAR(A, min=0, max=INF)
+  A?      -> VAR(A, min=0, max=1)
+  A{3,5}  -> VAR(A, min=3, max=5)
+
+If the reluctant field is true, the quantifier is reluctant (non-greedy).
+(RPRPatternNode.reluctant is bool; reluctant_location is the separate
+ParseLoc field holding the '?' token position, or -1 if absent.)
+
+Example: PATTERN ((A+ B) | C*)
+
+  ALT
+  +-- SEQ
+  |   +-- VAR(A, 1, INF)
+  |   +-- VAR(B, 1, 1)
+  +-- VAR(C, 0, INF)
+
+III-3. DEFINE Clause Transformation
+
+transformDefineClause() processes each DEFINE variable as follows:
+
+  (1) Checks for duplicate variable names
+  (2) Transforms the expression via transformExpr()
+  (3) Extracts Var nodes via pull_var_clause() and ensures each is
+      present in the query targetlist, so the planner propagates the
+      referenced columns through the plan tree
+  (4) Wraps in a TargetEntry with the variable name set in resname
+
+After all variables are processed:
+  (5) Coerces each expression to Boolean type (coerce_to_boolean)
+
+Variables that are used in PATTERN but not defined in DEFINE are implicitly
+evaluated as TRUE (matching all rows).
+
+Chapter IV  Compilation Phase
+============================================================================
+
+IV-1. Entry Point
+
+  create_windowagg_plan() (createplan.c)
+    +-- buildDefineVariableList()    Build variable name list from DEFINE
+    +-- buildRPRPattern()           NFA compilation (6 phases)
+
+IV-2. The 6 Phases of buildRPRPattern()
+
+  Phase 1: AST optimization (optimizeRPRPattern)
+  Phase 2: Statistics collection (scanRPRPattern)
+  Phase 3: Memory allocation (allocateRPRPattern)
+  Phase 4: NFA element fill (fillRPRPattern)
+  Phase 5: Finalization (finalizeRPRPattern)
+  Phase 6: Absorbability analysis (computeAbsorbability)
+
+IV-3. Phase 1: AST Optimization
+
+After copying the parser-generated AST, the following optimizations are
+applied:
+
+  (a) SEQ flattening: Unwrap nested SEQ nodes
+      SEQ(A, SEQ(B, C)) -> SEQ(A, B, C)
+
+  (b) Consecutive variable merging: Merge consecutive occurrences of the
+      same variable into a single quantifier
+      A A -> A{2}
+      A{2,3} A{1,2} -> A{3,5}
+
+  (c) Consecutive group merging: Merge repeated identical groups
+      (A B)+ (A B)+ -> (A B){2,INF}
+
+  (d) Consecutive ALT merging: Merge repeated identical ALT nodes
+      (A | B) (A | B) (A | B) -> (A | B){3}
+
+  (e) Prefix/suffix absorption: Absorb identical sequences before/after
+      a group
+      A B (A B)+ -> (A B){2,INF}
+
+  (f) ALT flattening and deduplication
+      (A | (B | C)) -> (A | B | C)
+      (A | B | A) -> (A | B)
+
+  (g) Quantifier multiplication: Collapse nested quantifiers when safe
+      (A+)+ -> A+
+      (A{2,3}){5} -> A{10,15}
+
+  (h) Single-child unwrap
+      SEQ(A) -> A,  (A){1,1} -> A
+
+IV-4. Phase 4: NFA Element Array Generation
+
+Transforms the optimized AST into a flat array of RPRPatternElement.
+This is the core data structure used for NFA simulation at runtime.
+
+RPRPatternElement struct (16 bytes):
+
+  Field      Size     Description
+  ---------------------------------------------------------
+  varId      1B      Variable ID (0-251) or control code (252-255)
+  depth      1B      Group nesting depth
+  flags      1B      Bit flags (see below)
+  reserved   1B      Padding
+  min        4B      Quantifier lower bound
+  max        4B      Quantifier upper bound
+  next       2B      Next element index (sequential flow)
+  jump       2B      Branch target index (for ALT/GROUP)
+
+Control codes:
+
+  RPR_VARID_BEGIN (252)  Group start marker
+  RPR_VARID_END   (253)  Group end marker
+  RPR_VARID_ALT   (254)  Alternation start marker
+  RPR_VARID_FIN   (255)  Pattern completion marker
+
+Element flags (1 byte, bitmask):
+
+  0x01  RPR_ELEM_RELUCTANT          (VAR, BEGIN, END)
+        Non-greedy quantifier.  Prefers shorter match: try exit-loop
+        first, then repeat.  Set on VAR for simple (A+?),
+        on BEGIN+END for group ((...)+?).
+
+  0x02  RPR_ELEM_EMPTY_LOOP         (END)
+        Group body can produce empty match (all children nullable).
+        Creates a fast-forward exit clone alongside the normal
+        loop-back so cycle detection doesn't kill legitimate
+        matches. (IV-4b)
+
+  0x04  RPR_ELEM_ABSORBABLE_BRANCH  (VAR, BEGIN, END, ALT)
+        Element lies within an absorbable region.  Used at runtime
+        to track whether the current NFA state is in an absorbable
+        context.
+
+  0x08  RPR_ELEM_ABSORBABLE         (VAR, END)
+        Absorption judgment point.  Where to compare consecutive
+        iterations for absorption.
+          - Simple unbounded VAR (A+): set on the VAR itself
+          - Unbounded GROUP ((A B)+): set on the END element only
+
+  Accessor macros:
+    RPRElemIsReluctant(e)        (e)->flags & 0x01
+    RPRElemCanEmptyLoop(e)       (e)->flags & 0x02
+    RPRElemIsAbsorbableBranch(e) (e)->flags & 0x04
+    RPRElemIsAbsorbable(e)       (e)->flags & 0x08
+
+Example: PATTERN (A+ B | C)
+
+  AST: ALT(SEQ(VAR(A,1,INF), VAR(B,1,1)), VAR(C,1,1))
+
+  Compilation result:
+
+  idx  varId  depth  min  max  next  jump  Description
+  ------------------------------------------------------------
+   0   ALT    0      1    1    1     -1    Alternation start
+   1   A(0)   1      1    INF  2     3     Branch 1: A+
+   2   B(1)   1      1    1    4     -1    Branch 1: B -> FIN
+   3   C(2)   1      1    1    4     -1    Branch 2: C -> FIN
+   4   FIN    0      1    1    -1    -1    Pattern completion
+
+  - idx 0: ALT marker. next(=1) is the start of the first branch
+  - idx 1: Variable A. next(=2) is B, jump(=3) is the start of the second
+           branch
+  - idx 2: Variable B. next(=4) is FIN
+  - idx 3: Variable C. next(=4) is FIN
+  - idx 4: FIN marker. Match completion signal
+
+Roles of next and jump:
+
+  - next: The next element to move to "after consuming" the current element.
+          For VAR, the next position after a successful match.
+          For BEGIN/END, the next position inside/outside the group.
+
+  - jump: The element to "skip to."
+          In ALT, a jump from one branch to the next branch.
+          In BEGIN, a skip path to END+1 (for groups with min=0).
+          In END, a loop-back to the start of the group body.
+
+Example: PATTERN ((A B)+)
+
+  idx  varId    depth  min  max  next  jump  Description
+  --------------------------------------------------------------
+   0   BEGIN    0      1    INF  1     4     Group start
+   1   A(0)     1      1    1    2     -1    A
+   2   B(1)     1      1    1    3     -1    B
+   3   END      0      1    INF  4     1     Group end
+   4   FIN      0      1    1    -1    -1    Pattern completion
+
+  - idx 0: BEGIN. next(=1) enters the group body.
+           jump(=4) skips to after END = FIN (used when min=0).
+  - idx 3: END. next(=4) exits the group.
+           jump(=1) loops back to the start of the group body.
+
+IV-4a. Reluctant Flag (RPR_ELEM_RELUCTANT)
+
+The reluctant flag is set during Phase 4 (fillRPRPattern) when the AST node
+has reluctant == true. It reverses the priority of quantifier expansion at
+runtime:
+
+  Greedy (default):  try loop-back first, then exit  (prefer longer match)
+  Reluctant:         try exit first, then loop-back   (prefer shorter match)
+
+The flag is set on all elements that carry the quantifier:
+
+  Simple VAR (A+?):     RPR_ELEM_RELUCTANT on the VAR element
+  Group ((...)+?):      RPR_ELEM_RELUCTANT on BEGIN and END elements
+
+At runtime (nfa_advance), the flag controls DFS exploration order:
+
+  VAR with quantifier:
+    Greedy:    primary path = next (continue), clone = jump (skip)
+    Reluctant: primary path = jump (skip), clone = next (continue)
+
+  END element:
+    Greedy:    primary path = jump (loop-back), clone = next (exit)
+    Reluctant: primary path = next (exit), clone = jump (loop-back)
+
+  BEGIN with min=0:
+    Greedy:    primary path = next (enter group), clone = jump (skip)
+    Reluctant: primary path = jump (skip), clone = next (enter group)
+
+The absorption optimization requires greedy quantifiers. Reluctant
+quantifiers are excluded from absorbability analysis (see IV-5).
+
+IV-4b. Empty Loop Flag (RPR_ELEM_EMPTY_LOOP)
+
+The empty-loop flag is set during Phase 4 (fillRPRPatternGroup) on the END
+element when the group body is nullable -- i.e., every path through the
+body can match zero rows (all children are nullable).
+
+Example patterns that trigger this flag:
+
+  (A?)*    A is nullable (min=0), so group body is nullable -> END gets flag
+  (A? B?)+ Both children nullable -> body nullable -> END gets flag
+  (A | B*) B* is nullable, making the ALT nullable -> END gets flag
+
+The flag works in conjunction with the empty match cycle detection
+(elemIdx visited bitmap). Without this flag, cycle detection alone would
+cause legitimate matches to fail.
+
+Problem example: (A*){2,3}
+  - Iteration 1: A* consumes all available rows -> count=1, END reached
+  - Loop-back for iteration 2: A* matches zero rows -> END reached again
+  - Cycle detection sees the same elemIdx on the same row -> state killed
+  - count never reaches min(2) -> match fails (incorrect)
+
+With the RPR_ELEM_EMPTY_LOOP flag, nfa_advance_end creates two paths:
+the normal loop-back (which cycle detection will eventually kill) and
+a fast-forward exit clone that bypasses the loop entirely.
+(See IX-4(c) for detailed runtime behavior.)
+
+IV-5. Absorbability Analysis (RPR_ELEM_ABSORBABLE)
+
+Context absorption is an optimization technique that reduces O(n^2) to O(n).
+(Runtime behavior is described in Chapter VIII.)
+
+This phase determines whether the pattern has a structure suitable for the
+absorption optimization and sets flags on the relevant elements:
+
+  RPR_ELEM_ABSORBABLE         Absorption comparison point
+  RPR_ELEM_ABSORBABLE_BRANCH  Element within an absorbable region
+
+Eligibility conditions:
+
+  (1) SKIP PAST LAST ROW (not NEXT ROW)
+  (2) Frame end is UNBOUNDED FOLLOWING
+
+Structural conditions (isUnboundedStart + computeAbsorbabilityRecursive):
+
+  Case 1: Simple VAR+ (e.g., A+)
+          -> ABSORBABLE | ABSORBABLE_BRANCH set on the VAR
+  Case 2: GROUP+ with fixed-length children (min == max, recursively)
+          e.g., (A B)+, (A B{2})+, ((A (B C){2}){2})+
+          -> ABSORBABLE_BRANCH on all elements within the group,
+             ABSORBABLE | ABSORBABLE_BRANCH on END
+
+          Why this is safe: when every child has min == max, the group
+          is semantically equivalent to unrolling its body into {1,1}
+          elements.  E.g., (A B{2})+ behaves like (A B B)+.  Each
+          iteration consumes a fixed number of rows, so an earlier
+          context's count always dominates a later one's (monotonicity).
+
+  Case 3: GROUP+ whose body starts with VAR+ (e.g., (A+ B)+)
+          -> Recurses from BEGIN into the body, applying Case 1.
+            ABSORBABLE | ABSORBABLE_BRANCH set on A.
+            B and END get no flags -> absorption stops once past A.
+
+Absorbability is determined per-element, not per-pattern.
+Absorption comparison is performed only when a state resides at an
+element with the RPR_ELEM_ABSORBABLE flag. Once a state leaves the
+flagged region, absorption is permanently disabled for that state.
+
+Through this mechanism, the runtime guarantees monotonicity:
+"a context that started earlier always subsumes a context that
+started later."
+
+Chapter V  NFA Runtime Data Structures
+============================================================================
+
+V-1. RPRNFAState -- NFA State
+
+A single NFA state represents "how far the pattern has progressed."
+
+  Field         Description
+  -----------------------------------------------------------
+  elemIdx       Index of the current pattern element
+  counts[]      Repetition count per group depth
+  isAbsorbable  Whether the state is in an absorbable region
+  next          Next state in the linked list
+
+The size of the counts array is rpPattern->maxDepth (= maximum group
+nesting depth + 1), allocated as a flexible array member at the end of
+the struct.
+
+Example: In PATTERN ((A B)+ C), a state waiting for B in the 3rd iteration
+
+  Element array: [0:BEGIN(d0) 1:A(d1) 2:B(d1) 3:END(d0) 4:C(d0) 5:FIN]
+
+  elemIdx = 2 (B, depth 1)
+  counts[0] = 2 (depth 0: depth of END. Group completed 2 iterations)
+  counts[1] = 1 (depth 1: depth of B. A matched in current iteration)
+
+  Counts are indexed by depth, not by elemIdx.
+  counts[0] is incremented when passing through END(depth 0),
+  and the group repetition count is preserved even when
+  the state is at B(depth 1).
+
+Definition of two states being "equal":
+
+  Two states are equal if they have the same elemIdx and the same counts
+  up to the depth of that element.
+  nfa_states_equal() compares counts[0..elem->depth] using memcmp.
+  Only counts at or below the depth of the current element are meaningful.
+
+V-2. RPRNFAContext -- Matching Context
+
+A single context represents "a matching attempt started from a specific
+start row."
+
+  Field                 Description
+  ---------------------------------------------------------------------
+  states                Linked list of active NFA states
+  matchStartRow         Row number where matching started
+  matchEndRow           Row number where matching completed
+                        (-1 if incomplete)
+  lastProcessedRow      Last row processed
+  matchedState          State that reached FIN (for greedy fallback)
+  hasAbsorbableState    Whether this context can absorb other contexts
+  allStatesAbsorbable   Whether this context can be absorbed
+  next, prev            Doubly-linked list
+
+Since the NFA is nondeterministic, multiple states can coexist
+simultaneously within a single context.
+
+Example: In PATTERN (A | B) C, if the first row matches both A and B,
+two states coexist within the context:
+
+  State 1: elemIdx=3 (waiting for C, via branch A)
+  State 2: elemIdx=3 (waiting for C, via branch B)
+
+In this case, since the (elemIdx, counts) of the two states are equal,
+nfa_add_state_unique() retains only State 1 (branch A), which was
+added first.
+Because DFS processes the first branch of ALT first, the state via A
+is registered first, and the state via B is discarded as a duplicate.
+This is the preferment guarantee.
+
+V-3. RPR Fields of WindowAggState
+
+  nfaContext / nfaContextTail   Doubly-linked list of active contexts
+  nfaContextFree                Reuse pool for contexts
+  nfaStateFree                  Reuse pool for states
+  nfaVarMatched                 Per-row cache: varMatched[varId]
+  nfaVisitedElems               Bitmap for cycle detection
+  nfaStateSize                  Precomputed size of RPRNFAState
+
+Memory management:
+
+  States and contexts are managed through their own free lists.
+  Instead of palloc, they are obtained from the reuse pool, and
+  returned to the pool upon deallocation.
+  This reduces the overhead of frequent allocation/deallocation.
+
+Chapter VI  NFA Execution: 3-Phase Model
+============================================================================
+
+VI-1. Entry Point and Overall Flow
+
+When the window function processes each row, row_is_in_reduced_frame()
+is called. This function determines whether the current row belongs to
+a matched frame, and if necessary, calls update_reduced_frame() to
+drive the NFA.
+
+Flow of update_reduced_frame():
+
+  (1) Find or create a context for the target row
+  (2) Enter the row processing loop
+  (3) After the loop ends, record the match result
+
+Pseudocode of the row processing loop:
+
+  targetCtx = ExecRPRGetHeadContext(pos)
+  if targetCtx == NULL:
+      targetCtx = ExecRPRStartContext(pos)
+
+  for currentPos = startPos; targetCtx->states != NULL; currentPos++:
+      if not nfa_evaluate_row(currentPos):  -- row does not exist
+          ExecRPRFinalizeAllContexts()      -- finalize all contexts
+          ExecRPRCleanupDeadContexts()      -- clean up after finalization
+          break
+
+      ExecRPRProcessRow(currentPos)         -- 3-phase processing
+      ExecRPRStartContext(currentPos + 1)   -- pre-create next start point
+      ExecRPRCleanupDeadContexts()          -- remove dead contexts
+
+Key point: Processing a single row may require processing multiple rows
+ahead. Due to the nature of window functions, determining the frame for
+row N requires looking at rows beyond N.
+
+VI-2. Context Creation: ExecRPRStartContext()
+
+Creates a new context and performs the initial advance.
+
+  (1) Allocate context via nfa_context_alloc()
+  (2) Set matchStartRow = pos
+  (3) Create initial state: elemIdx=0 (first pattern element),
+      counts=all zero
+  (4) Call nfa_advance(initialAdvance=true)
+
+The initial advance expands epsilon transitions at the beginning of
+the pattern. For example, the initial advance for PATTERN ((A | B) C):
+
+  Start: elemIdx=0 (ALT)
+    -> Expand ALT branches
+      -> elemIdx=1 (A) -- VAR, so add state; stop here
+      -> elemIdx=2 (B) -- VAR, so add state; stop here
+
+  Result: Two states in the context {waiting for A, waiting for B}
+
+During the initial advance, reaching FIN is not recorded as a match.
+This is to prevent empty matches.
+
+VI-3. Row Evaluation: nfa_evaluate_row()
+
+Evaluates all variable conditions in the DEFINE clause at once for
+the current row.
+
+  for each defineClause[i]:
+      result = ExecEvalExpr(defineClause[i])
+      varMatched[i] = (not null and true)
+
+To support row navigation operators (PREV, NEXT, FIRST, LAST),
+a 1-slot model is used: only ecxt_outertuple is set to the current
+row.  Navigation is handled by EEOP_RPR_NAV_SET/RESTORE opcodes
+emitted during DEFINE expression compilation:
+
+  NAV_SET:     save ecxt_outertuple, swap in target row via nav_slot
+  (evaluate):  argument expression reads from swapped slot
+  NAV_RESTORE: restore original ecxt_outertuple
+
+Compound navigation (PREV(FIRST()), NEXT(FIRST()), PREV(LAST()),
+NEXT(LAST())) is flattened by the parser into a single RPRNavExpr
+with a compound kind (RPR_NAV_PREV_FIRST, etc.).  The executor
+computes the target position in two steps: first the inner reference
+point (match_start + N or currentpos - N) with match-range validation,
+then the outer adjustment (+/- M) with partition-range validation.
+If either step is out of range, the result is NULL.
+
+nav_slot caches the last fetched position (nav_slot_pos) to avoid
+redundant tuplestore lookups when multiple navigation calls target
+the same row.
+
+The varMatched array is referenced later in Phase 1 (Match).
+
+VI-4. Per-Context Re-evaluation (match_start-dependent variables)
+
+DEFINE variables that depend on match_start (those containing FIRST,
+LAST-with-offset, or compound PREV_FIRST/NEXT_FIRST/PREV_LAST/NEXT_LAST)
+are identified at plan time via defineMatchStartDependent.  The shared
+evaluation in nfa_evaluate_row() uses the head context's matchStartRow
+for FIRST/LAST base position.
+
+When processing a context whose matchStartRow differs from the shared
+value, nfa_reevaluate_dependent_vars() temporarily sets nav_match_start
+to that context's matchStartRow and re-evaluates only the dependent
+variables.  The original nav_match_start and currentpos are saved and
+restored after re-evaluation.
+
+Summary of evaluation strategy by navigation content:
+
+  Navigation content               evaluation
+  -------------------------------------------------------
+  No navigation                    shared (once per row)
+  PREV/NEXT only                   shared (once per row)
+  LAST (no offset)                 shared (once per row)
+  LAST (with offset)               per-context
+  FIRST (any)                      per-context
+  Compound (inner FIRST)           per-context
+  Compound (inner LAST, no off.)   shared (once per row)
+  Compound (inner LAST, w/off.)    per-context
+
+VI-5. Tuplestore Mark and Trim (nodeWindowAgg.c)
+
+Navigation functions require access to past rows via the tuplestore.
+To allow tuplestore_trim() to free rows that are no longer reachable,
+the planner computes two offsets (see compute_nav_offsets):
+
+  navMaxOffset (Nav Mark Lookback):
+    Maximum backward reach from currentpos.  Contributed by PREV,
+    LAST-with-offset, and compound PREV_LAST/NEXT_LAST.
+    Mark position: currentpos - navMaxOffset.
+
+  navFirstOffset (Nav Mark Lookahead):
+    Minimum forward offset from match_start.  Contributed by FIRST
+    and compound PREV_FIRST/NEXT_FIRST.  Can be negative when
+    compound PREV_FIRST looks before match_start.
+    Mark position: oldest_context->matchStartRow + navFirstOffset.
+
+The actual mark is set to: min(lookback_mark, lookahead_mark).
+This ensures all rows reachable by any navigation function are retained.
+
+When offsets contain non-constant expressions (Param), the planner sets
+navMaxOffsetKind/navFirstOffsetKind to RPR_NAV_OFFSET_NEEDS_EVAL and the
+executor evaluates them at init time.  On overflow, the kind is set to
+RPR_NAV_OFFSET_RETAIN_ALL, disabling trim for that dimension.
+
+VI-6. ExecRPRProcessRow(): 3-Phase Processing
+
+NFA processing for a single row is divided into three phases:
+
+  +--------------------------------------------+
+  |  Phase 1: MATCH (convergence)              |
+  |  Compare the current row against each VAR  |
+  |  state. Remove states that fail to match.  |
+  |                                            |
+  |  Phase 2: ABSORB (absorption)              |
+  |  Merge duplicate contexts to prevent       |
+  |  state explosion.                          |
+  |                                            |
+  |  Phase 3: ADVANCE (expansion)              |
+  |  Expand epsilon transitions to prepare     |
+  |  for the next row.                         |
+  +--------------------------------------------+
+
+This ordering is important:
+
+  - Match executes first to "consume the current row."
+  - Absorb executes immediately after Match, when states have been updated.
+  - Advance executes last to prepare "states waiting for the next row."
+
+Chapter VII  Phase 1: Match
+============================================================================
+
+nfa_match() iterates through each state in the context:
+
+  (1) Check whether the state's elemIdx is a VAR element
+  (2) Compare against the current row using nfa_eval_var_match()
+  (3) Match success: increment repetition count, retain state
+  (4) Match failure: remove state
+
+Match determination (nfa_eval_var_match):
+
+  If varId is within the range of defineVariableList:
+      Use the value of varMatched[varId]
+
+  If varId exceeds the range (variable not defined in DEFINE):
+      Unconditionally true (matches all rows)
+
+Immediate advance for simple VARs:
+
+  For a VAR with min=1, max=1 where the next element is END,
+  the Match phase processes through END immediately.
+  This is necessary for accurate state comparison in Phase 2 (Absorb).
+
+  Example: In PATTERN ((A B)+), when A matches, it immediately advances
+  to B, and when B matches, it immediately advances through END to
+  complete the group count. This enables absorption comparison with
+  other contexts.
+
+Chapter VIII  Phase 2: Absorb (Context Absorption)
+============================================================================
+
+VIII-1. Problem
+
+In the current implementation, a new context is started for each row
+processed.
+Applying PATTERN (A+) to 10 rows produces 10 contexts,
+each of which tracks state independently.
+
+If there are N rows, the total number of states becomes O(N^2):
+
+  Context 1 (started at row 1): can match A up to N times
+  Context 2 (started at row 2): can match A up to N-1 times
+  ...
+  Context N (started at row N): can match A 1 time
+
+VIII-2. Solution: Context Absorption
+
+Key observation: a context started earlier contains
+all matches of a later-started context (monotonicity principle).
+
+If Context 1 started at row 1 and matched A 5 times,
+the state where Context 2 (started at row 2) matched A 4 times
+is already contained within Context 1.
+
+Therefore Context 2 can be "absorbed" into Context 1.
+
+VIII-3. Absorption Conditions
+
+Planner-time prerequisites (all must hold for absorption to be enabled):
+
+  (a) SKIP PAST LAST ROW.  SKIP TO NEXT ROW creates overlapping
+      contexts that cannot be safely absorbed.
+  (b) Unbounded frame (ROWS BETWEEN CURRENT ROW AND UNBOUNDED
+      FOLLOWING).  Limited frames apply differently to each context,
+      breaking the monotonicity principle.
+  (c) No match_start-dependent navigation in DEFINE.
+
+      Mechanism: each context has a different matchStartRow, so FIRST
+      resolves to a different row for each context at the same
+      currentpos.  An earlier context's DEFINE result no longer
+      subsumes a later one's, making count-dominance comparison
+      invalid.  Rather than comparing matchStartRow at runtime
+      (which would complicate the absorb path), any match_start
+      dependency disables absorption entirely.
+
+      Navigation content              match_start dep.  absorption
+      ------------------------------------------------------------
+      No navigation                   none              safe
+      PREV/NEXT only                  none              safe
+      LAST (no offset)                none              safe
+      LAST (with offset)              boundary check    unsafe
+      FIRST (any)                     direct            unsafe
+      Compound (inner FIRST)          direct            unsafe
+      Compound (inner LAST, no off.)  none              safe
+      Compound (inner LAST, w/off.)   boundary chk      unsafe
+
+Runtime conditions (evaluated per context pair):
+
+  (1) The pattern is marked as isAbsorbable (see IV-5)
+  (2) allStatesAbsorbable of the target context is true
+  (3) An earlier context "covers" all states of the target
+
+Cover condition (nfa_states_covered):
+
+  A state with the same elemIdx exists in the earlier context,
+  and the count at that depth is greater than or equal -- then it is covered.
+
+VIII-4. Dual-Flag Design
+
+Two boolean flags make the absorption decision efficient:
+
+  hasAbsorbableState (monotonic: only true->false transition possible)
+    "Does this context have the ability to absorb other contexts?"
+    true if at least one absorbable state exists.
+    Transitions to false when states are removed leaving no absorbable
+    states.
+    Once false, it never becomes true again.
+
+  allStatesAbsorbable (dynamic: can fluctuate)
+    "Can this context be absorbed?"
+    true if all states are in an absorbable region.
+    Becomes false when a non-absorbable state is added; reverts to true
+    when it is removed.
+
+VIII-5. Absorption Order
+
+nfa_absorb_contexts() traverses from tail (newest) to head (oldest).
+
+  for ctx = tail to head:
+      if ctx.allStatesAbsorbable:
+          for older = ctx.prev to head:
+              if older.hasAbsorbableState:
+                  if nfa_states_covered(older, ctx):
+                      free(ctx)  -- absorbed
+                      break
+
+Since inspection starts from the newest context, the most recently started
+(= having the shortest match) context is absorbed first.
+
+Chapter IX  Phase 3: Advance (Epsilon Transition Expansion)
+============================================================================
+
+IX-1. Overview
+
+nfa_advance() expands epsilon transitions from each state after Match,
+generating "new states waiting for the next row."
+
+An epsilon transition is a transition that moves without consuming a row:
+
+  - ALT: branch to each alternative
+  - BEGIN: enter group (or skip if min=0)
+  - END: loop-back within group (or exit when condition is met)
+  - FIN: record match completion
+  - VAR loop/exit: repeat/exit according to the quantifier
+
+Expansion stops upon reaching a VAR element, and the state is added.
+This is because VAR is the element that "will consume the next row."
+
+IX-2. Processing Order: DFS and Preferment
+
+advance processes states in lexicographic order,
+performing Depth-First Search (DFS) on each state.
+
+This DFS order is what guarantees the SQL standard's "preferment":
+
+  The branch that appears first in the PATTERN text takes precedence.
+
+Example: PATTERN (A | B) C
+
+  The first branch A of the ALT takes precedence over the second branch B.
+  When both A and B can match, the match via A is selected.
+
+nfa_add_state_unique() prevents duplicate addition of the same state,
+so the state added first (= from the preferred branch) is retained.
+
+IX-3. Routing Function: nfa_route_to_elem()
+
+All inter-element transitions in the advance phase go through
+nfa_route_to_elem().
+This function branches its behavior based on the type of the next element:
+
+  If the next element is VAR:
+    (1) Add the state to the context (nfa_add_state_unique)
+    (2) If the VAR has min=0, also add a skip path (recurse via next)
+    -> Expansion stops here (VAR is the element that "will consume the next
+       row")
+
+  If the next element is non-VAR (ALT, BEGIN, END, FIN):
+    -> Recursively call nfa_advance_state() to continue expansion
+
+With this structure, advance recursively follows epsilon transitions
+until reaching a VAR, consistently stopping only at VAR elements.
+
+IX-4. Per-Element advance Behavior
+
+(a) ALT (nfa_advance_alt)
+
+  Upon encountering an ALT element, all branches are expanded in order.
+  The first element of each branch is connected via a jump pointer.
+
+  idx=0 (ALT) -> branch 1 start (next) -> branch 2 start (jump) -> ...
+
+  nfa_advance_state() is recursively called for each branch.
+
+(b) BEGIN (nfa_advance_begin)
+
+  Handles group entry.
+  jump points to the element after END (= first element outside the group).
+
+  Greedy (default):
+    (1) Enter the group body (move via next, reset the count at that depth)
+    (2) If min=0, also add a group skip path (move via jump)
+
+  Reluctant:
+    Order reversed -- skip path first, group entry second.
+    If the skip path reaches FIN, the group entry path is not generated
+    (shortest match preferred).
+
+(c) END (nfa_advance_end)
+
+  Handles group termination. This is the core of the repetition logic.
+
+  Let count be the count at the current depth:
+
+  count < min:
+    Loop-back (move via jump, repeat the group body)
+
+    If the RPR_ELEM_EMPTY_LOOP flag is set:
+      In addition to loop-back, also add a fast-forward exit path.
+      This is because the body may produce an empty match, causing count
+      to never reach min. fast-forward resets counts[depth] to 0
+      and exits via next (treating the remaining required iterations
+      as empty matches).
+
+  min <= count < max:
+    Greedy: loop-back first, exit second
+    Reluctant: exit first, loop-back second
+               If the exit path reaches FIN, loop-back is omitted.
+
+  count >= max:
+    Unconditional exit (move via next)
+
+  On exit: reset counts[depth] = 0, and if the next element is an outer END,
+  increment the count at the outer depth.
+
+(d) VAR (nfa_advance_var)
+
+  Handles repeat/exit for a VAR element with a quantifier.
+
+  Let count be the count at the current depth:
+
+  count < min:
+    Unconditional loop (stay at the same elemIdx, wait for the next row)
+
+  min <= count < max:
+    Greedy: loop first, exit (next) second
+    Reluctant: exit first, loop second
+               If the exit path reaches FIN, loop is omitted.
+
+  count >= max:
+    Unconditional exit (move via next)
+
+  On exit: reset counts[depth] = 0.
+
+(e) FIN
+
+  Match success. The current state is moved to matchedState for recording,
+  and matchEndRow is set to the current row.
+
+  Upon reaching FIN, all remaining unprocessed states are removed
+  (early termination). By DFS order, the path that reached FIN first
+  has the highest preferment, so the rest are inferior paths.
+  This is the core mechanism that guarantees preferment.
+
+  In SKIP PAST LAST ROW mode, upon reaching FIN, subsequent contexts
+  that started within the match range are immediately pruned.
+
+IX-5. State Deduplication: nfa_add_state_unique()
+
+When adding a new state to a context, it is compared against existing
+states;
+if an identical state already exists, it is not added.
+
+Comparison criteria: elemIdx + counts[0..elem->depth] (see V-1)
+
+This deduplication is the core mechanism that suppresses NFA state
+explosion.
+Because DFS order causes preferred-branch states to be added first,
+identical states from lower-priority branches are automatically discarded.
+
+IX-6. Cycle Detection: nfaVisitedElems
+
+When a group body can produce an empty match,
+looping back from END may cause an infinite loop.
+
+Example: PATTERN ((A?)*)
+
+  A? has min=0, so it can pass through without matching.
+  If the outer group repeats: BEGIN -> A? skip -> END -> BEGIN -> ...
+
+To prevent this:
+
+  (1) At compile time: set the RPR_ELEM_EMPTY_LOOP flag on the END
+      of groups whose body is nullable.
+      The runtime effect of this flag is described in IX-4(c):
+      when count < min, a fast-forward exit path is added,
+      resolving the deadlock where count cannot increase due to empty
+      matches.
+
+  (2) At runtime: initialize the nfaVisitedElems bitmap immediately before
+      DFS expansion of each state within advance (once per state).
+      During DFS, epsilon elements (END, ALT, BEGIN) are marked in the
+      bitmap at nfa_advance_state entry.  VAR elements are marked later
+      when added to the state list (nfa_add_state_unique), so that
+      legitimate loop-back to the same VAR in a new group iteration
+      (e.g., END -> ALT -> same VAR) is not blocked.
+      If a previously visited elemIdx is revisited, that path is terminated.
+
+  Note: the bitmap tracks only elemIdx and does not consider counts.
+  Therefore, legitimate revisits to the same elemIdx but with different
+  counts may also be blocked.  This only occurs when the group body is
+  nullable (all paths can match empty), causing END -> loop-back ->
+  skip -> END within a single DFS.  In such cases the END element has
+  the RPR_ELEM_EMPTY_LOOP flag, so the fast-forward exit (IX-4(c))
+  provides an alternative path that bypasses the cycle.
+
+Chapter X  Match Result Processing
+============================================================================
+
+X-1. Match Result
+
+RPR tracks the current match result as a single entry in WindowAggState
+with four fields: rpr_match_valid, rpr_match_matched, rpr_match_start,
+and rpr_match_length.  When rpr_match_valid is true, the entry describes
+the match result for the position at rpr_match_start: rpr_match_matched
+indicates success or failure, and rpr_match_length gives the number of
+rows consumed.  A match with rpr_match_length 0 represents an empty match
+(pattern matched but consumed no rows).  When rpr_match_valid is false,
+the position has not been evaluated yet (RF_NOT_DETERMINED).
+
+A row's status against the current match result can be obtained by
+calling get_reduced_frame_status().
+
+X-2. AFTER MATCH SKIP
+
+Determines the starting point for the next match attempt after a successful
+match:
+
+  SKIP TO NEXT ROW:
+    New match attempt begins from the row after the match start row.
+    Overlapping matches are possible.
+
+  SKIP PAST LAST ROW:
+    New match attempt begins from the row after the match end row.
+    Only non-overlapping matches are possible.
+
+X-3. INITIAL vs SEEK
+
+  Standard definition (section 6.12):
+  INITIAL: "is used to look for a match whose first row is R."
+  SEEK:    "is used to permit a search for the first match anywhere
+           from R through the end of the full window frame."
+  In either case, if there is no match, the reduced window frame is empty.
+  The default is INITIAL.
+
+  Current implementation:
+  SEEK is not supported (the parser raises an error).
+  Only INITIAL is supported, searching only for matches starting at each
+  row position pos.
+
+X-4. Bounded Frame Handling
+
+  When the frame is bounded (e.g., ROWS BETWEEN CURRENT ROW AND 5
+  FOLLOWING), ExecRPRProcessRow receives hasLimitedFrame=true and
+  frameOffset indicating the upper bound.  Before the match phase,
+  any context whose match has exceeded the frame boundary
+  (currentPos >= matchStartRow + frameOffset + 1) is finalized early
+  by forcing a mismatch.  This prevents matches from extending beyond
+  the window frame.  The sum is clamped to PG_INT64_MAX on overflow.
+
+  Note that bounded frames also disable context absorption at the
+  planner level (see VIII-3(b)), since the frame boundary breaks the
+  monotonicity assumption required for correct absorption.
+
+Chapter XI  Worked Example: Full Execution Trace
+============================================================================
+
+XI-1. Query
+
+  SELECT company, tdate, price,
+         first_value(price) OVER w AS start_price,
+         last_value(price) OVER w AS end_price
+  FROM stock
+  WINDOW w AS (
+    PARTITION BY company
+    ORDER BY tdate
+    ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
+    AFTER MATCH SKIP PAST LAST ROW
+    PATTERN (A+ B)
+    DEFINE A AS price > PREV(price),
+           B AS price < PREV(price)
+  );
+
+XI-2. Data
+
+  Row#    tdate       price
+  --------------------------
+  0       2024-01-01  100
+  1       2024-01-02  110
+  2       2024-01-03  120
+  3       2024-01-04  115
+  4       2024-01-05  130
+
+XI-3. Compilation Result
+
+  PATTERN (A+ B) -> unchanged after optimization
+
+  idx  varId  depth  min  max  next  jump
+  -----------------------------------------
+   0   A(0)   0      1    INF  1     -1     A+
+   1   B(1)   0      1    1    2     -1     B
+   2   FIN    0      1    1    -1    -1
+
+  DEFINE: A -> "price > PREV(price)", B -> "price < PREV(price)"
+  isAbsorbable = true (A+ is a simple unbounded VAR)
+
+XI-4. Execution Trace
+
+--- Row 0 (price=100) ---
+
+  update_reduced_frame(0) called.
+
+  Context C0 created (matchStartRow=0).
+  Initial advance: elemIdx=0(A) -> VAR, so state is added.
+  C0.states = [{elemIdx=0, counts=[0]}]
+
+  nfa_evaluate_row(0):
+    A: price(100) > PREV(price) -> no PREV -> false
+    B: price(100) < PREV(price) -> no PREV -> false
+    varMatched = [false, false]
+
+  ExecRPRProcessRow(0):
+    Phase 1 (Match): A(0) state vs varMatched[0]=false -> state removed
+    C0.states = [] (empty)
+
+    Phase 2 (Absorb): skipped (no states)
+    Phase 3 (Advance): skipped (no states)
+
+  C0.states is empty, so the loop terminates.
+  matchEndRow < matchStartRow -> unmatched.
+
+--- Row 1 (price=110) ---
+
+  update_reduced_frame(1) called.
+
+  Context C1 created (matchStartRow=1).
+  Initial advance: C1.states = [{elemIdx=0, counts=[0]}]
+
+  nfa_evaluate_row(1):
+    A: 110 > PREV(100) -> true
+    B: 110 < PREV(100) -> false
+    varMatched = [true, false]
+
+  ExecRPRProcessRow(1):
+    Phase 1 (Match): A(0) match succeeds -> counts[0]++ -> counts=[1]
+    C1.states = [{elemIdx=0, counts=[1]}]
+
+    Phase 3 (Advance):
+      State {elemIdx=0, counts=[1]}: A+ (min=1, count=1, max=INF)
+        count >= min, so:
+        Greedy -> loop first: keep {elemIdx=0, counts=[1]}
+                  exit: reset counts[0]=0, next(=1) -> {elemIdx=1,
+                        counts=[0]}
+    C1.states = [{elemIdx=0, counts=[1]}, {elemIdx=1, counts=[0]}]
+
+--- Row 2 (price=120) ---
+
+  Context C2 created (matchStartRow=2).
+  Initial advance: C2.states = [{elemIdx=0, counts=[0]}]
+
+  nfa_evaluate_row(2):
+    A: 120 > PREV(110) -> true
+    B: 120 < PREV(110) -> false
+    varMatched = [true, false]
+
+  C1 ExecRPRProcessRow(2):
+    Phase 1 (Match):
+      {elemIdx=0, counts=[1]}: A matches -> counts=[2]
+      {elemIdx=1, counts=[0]}: B does not match -> removed
+    C1.states = [{elemIdx=0, counts=[2]}]
+
+  C2 ExecRPRProcessRow(2):
+    Phase 1 (Match):
+      {elemIdx=0, counts=[0]}: A matches -> counts=[1]
+    C2.states = [{elemIdx=0, counts=[1]}]
+
+    Phase 2 (Absorb):
+      Does C1 (started earlier) cover C2?
+        C1: {elemIdx=0, counts=[2]}, C2: {elemIdx=0, counts=[1]}
+        Same elemIdx, C1.counts >= C2.counts -> covered
+      C2 absorbed. -> removed.
+
+    Phase 3 (Advance):
+      {elemIdx=0, counts=[2]}: Greedy -> loop + exit
+        Loop: {elemIdx=0, counts=[2]}
+        Exit: reset counts[0]=0, next(=1) -> {elemIdx=1, counts=[0]}
+    C1.states = [{elemIdx=0, counts=[2]}, {elemIdx=1, counts=[0]}]
+
+  Context C3 created (matchStartRow=3).
+
+--- Row 3 (price=115) ---
+
+  nfa_evaluate_row(3):
+    A: 115 > PREV(120) -> false
+    B: 115 < PREV(120) -> true
+    varMatched = [false, true]
+
+  ExecRPRProcessRow(3):
+    Phase 1 (Match):
+      {elemIdx=0, counts=[2]}: A does not match -> removed
+      {elemIdx=1, counts=[0]}: B matches -> counts=[1]
+    C1.states = [{elemIdx=1, counts=[1]}]
+
+    Phase 3 (Advance):
+      {elemIdx=1, counts=[1]}: B (min=1, max=1)
+        count(1) >= max(1) -> unconditional exit
+        Reset counts[0]=0, next = 2 (FIN)
+      FIN reached -> matchEndRow = 3, matchedState recorded.
+      Early termination: no remaining states, so completed immediately.
+    C1.states = [] (empty after reaching FIN)
+
+  C1.states is empty and matchEndRow=3 >= matchStartRow=1 -> match succeeds.
+
+  rpr_match_start = 1, rpr_match_length = 3
+
+--- Row 4 (price=130) ---
+
+  update_reduced_frame(4) called.
+  C3 was already created but matchStartRow=3, so it is not applicable.
+  New context C4 created (matchStartRow=4).
+
+  nfa_evaluate_row(4):
+    A: 130 > PREV(115) -> true
+    B: 130 < PREV(115) -> false
+
+  ... No subsequent rows, so ExecRPRFinalizeAllContexts() is called.
+  Match incomplete -> unmatched.
+
+XI-5. Final Result
+
+  Row 0: unmatched     -> frame = the row itself
+  Row 1: match head    -> frame = rows 1 through 3
+  Row 2: inside match  -> skipped
+  Row 3: inside match  -> skipped
+  Row 4: unmatched     -> frame = the row itself
+
+Chapter XII  Summary of Key Design Decisions
+============================================================================
+
+XII-1. Flat Array vs Tree-Based NFA
+
+  Choice: Flat array (RPRPatternElement[])
+
+  Rationale:
+  - Cache-friendly: 16-byte fixed size, contiguous memory
+  - Index-based references: 2-byte indices instead of pointers
+  - Easy to serialize: can use memcpy when passing to plan nodes
+
+XII-2. Forward-only Execution vs Backtracking
+
+  Choice: Forward-only (state set tracking)
+
+  Rationale:
+  - Backtracking takes exponential time in the worst case
+  - NFA simulation guarantees polynomial time
+  - DFS order naturally guarantees preferment.
+    Greedy/reluctant per quantifier requires only reversing the DFS order
+  - Window functions receive sorted rows sequentially.
+    Forward-only fits directly into this pipeline,
+    whereas backtracking requires re-fetching previous rows
+  - DEFINE conditions are SQL expressions (PREV, RUNNING aggregates, etc.)
+    with high re-evaluation cost. Forward-only requires only one evaluation
+    per row
+
+XII-3. Per-Context Management
+
+  Choice: Independent context per start row
+
+  Rationale:
+  - Supports overlapping matches under SKIP TO NEXT ROW
+  - Determines the frame for each row independently
+  - Absorption optimization can eliminate redundant contexts in O(n)
+
+XII-4. Memory Pool Management
+
+  Choice: Custom free list
+
+  Rationale:
+  - NFA states are created and destroyed in large numbers per row
+  - Avoids palloc/pfree overhead
+  - State size is variable (counts[] array), but within a single query
+    maxDepth is fixed, so all states have the same size
+
+XII-5. Execution Optimization Summary
+
+  The following optimizations make the NFA simulation practical.
+
+  -- Compile-time --
+
+  (1) AST Optimization (IV-3)
+
+    Simplifies the AST before converting the pattern to an NFA.
+    Reduces the number of NFA elements through consecutive variable
+    merging (A A -> A{2}), SEQ flattening, quantifier multiplication,
+    and other transformations.
+
+    Significance: Reducing the element count directly shrinks the state
+    space, decreasing the cost of all subsequent runtime phases (match,
+    absorb, advance).
+
+  -- Runtime: advance phase --
+
+  (2) Group Skip (IX-4(b))
+
+    At the BEGIN of a group with min=0, uses jump to skip the entire
+    group. Moves directly to the first element outside the group without
+    exploring the group body. Greedy enters then skips; Reluctant skips
+    then enters.
+
+    Significance: For optional groups (min=0), immediately generates
+    a skip path without exploring the body, avoiding unnecessary DFS
+    expansion.
+
+  (3) State Deduplication (IX-5)
+
+    During advance, DFS may generate states with the same (elemIdx,
+    counts) combination through multiple paths. Additionally, for
+    group absorption, nfa_match performs inline advance from bounded
+    VARs (count >= max) within the absorbable region (ABSORBABLE_BRANCH)
+    through END chains to reach the judgment point (ABSORBABLE END).
+    This process can also produce duplicate states reaching the same END.
+    nfa_add_state_unique() blocks duplicate addition of identical states
+    in both cases.
+
+    Significance: Prevents exponential growth of the state count in
+    ALT branches and quantifier expansion. Since DFS order causes the
+    preferred branch's state to be registered first, identical states
+    from lower-priority branches are automatically discarded, thereby
+    also guaranteeing preferment.
+
+  (4) Cycle Detection and Fast-Forward (IX-6, IX-4(c))
+
+    When a nullable group body (e.g., A?) repeats empty matches,
+    the END -> BEGIN loop-back can continue indefinitely.
+
+    Two mechanisms resolve this:
+    - A visited bitmap (nfaVisitedElems) blocks revisitation of the
+      same element, preventing infinite loops (safety)
+    - At an END with the RPR_ELEM_EMPTY_LOOP flag set, when
+      count < min, the remaining required iterations are treated as
+      empty matches and a fast-forward exit path out of the group is
+      added (correctness)
+
+    Significance: Cycle detection guarantees termination, and
+    fast-forward guarantees that the min condition is satisfied.
+    Without these, patterns containing nullable groups would fall
+    into infinite loops or fail to match.
+
+  (5) Match Pruning (IX-4(e))
+
+    When a state reaches FIN during advance, all remaining unprocessed
+    states of that context are removed. Because of DFS order, the path
+    that reaches FIN first has the highest preferment, so the remaining
+    paths are inferior.
+
+    Significance: Once the best match is determined, exploration of
+    inferior paths is immediately terminated. This mechanism achieves
+    both preferment guarantees and performance optimization.
+
+  -- Runtime: inter-context --
+
+  (6) Early Termination (SKIP PAST LAST ROW)
+
+    In SKIP PAST LAST ROW mode, when a match is found, subsequent
+    contexts whose start rows fall within the match range are pruned
+    immediately without further processing.
+    In SKIP TO NEXT ROW mode, overlapping contexts are preserved
+    because each row requires its own independent match.
+
+    Significance: Prunes subsequent contexts whose start rows overlap
+    with a prior match range, avoiding unnecessary processing.
+
+  (7) Context Absorption (Chapter VIII)
+
+    If an independent context is created for each row, O(n^2) states
+    accumulate. By exploiting the monotonicity that an earlier-started
+    context subsumes the states of a later-started context, redundant
+    contexts are eliminated early.
+
+    Absorbability is determined per-element; comparison is performed
+    only at elements with the RPR_ELEM_ABSORBABLE flag (see IV-5).
+
+    Significance: Keeps the number of active contexts at a constant
+    level, achieving O(n^2) -> O(n) time complexity. Without this,
+    performance degrades sharply on long partitions.
+
+Appendix A. Key Function Index
+============================================================================
+
+  Function                      File                  Role
+  --------------------------------------------------------------------------
+  transformRPR                  parse_rpr.c           Parser entry point
+  transformDefineClause         parse_rpr.c           DEFINE transformation
+  collectPatternVariables       rpr.c                 Variable collection
+  buildDefineVariableList       rpr.c                 DEFINE variable list
+  buildRPRPattern               rpr.c                 NFA compilation main
+  optimizeRPRPattern            rpr.c                 AST optimization
+  fillRPRPattern                rpr.c                 NFA element generation
+  finalizeRPRPattern            rpr.c                 Finalization
+  computeAbsorbability          rpr.c                 Absorption analysis
+  update_reduced_frame          nodeWindowAgg.c       Execution main loop
+  nfa_evaluate_row              nodeWindowAgg.c       DEFINE evaluation
+  ExecRPRStartContext           execRPR.c             Context creation
+  ExecRPRProcessRow             execRPR.c             3-phase processing
+  nfa_match                     execRPR.c             Phase 1
+  nfa_absorb_contexts           execRPR.c             Phase 2
+  nfa_advance                   execRPR.c             Phase 3
+  nfa_advance_state             execRPR.c             Per-state branching
+  nfa_route_to_elem             execRPR.c             Element routing
+  nfa_advance_alt               execRPR.c             ALT handling
+  nfa_advance_begin             execRPR.c             BEGIN handling
+  nfa_advance_end               execRPR.c             END handling
+  nfa_advance_var               execRPR.c             VAR handling
+  nfa_add_state_unique          execRPR.c             Deduplication
+  nfa_states_covered            execRPR.c             Absorption check
+  nfa_reevaluate_dependent_vars execRPR.c             Per-context re-eval
+  ExecRPRGetHeadContext         execRPR.c             Context lookup
+  ExecRPRFreeContext            execRPR.c             Context deallocation
+  ExecRPRCleanupDeadContexts    execRPR.c             Dead context cleanup
+  ExecRPRFinalizeAllContexts    execRPR.c             Partition-end finalize
+  ExecRPRRecordContextSuccess   execRPR.c             Stats: match success
+  ExecRPRRecordContextFailure   execRPR.c             Stats: match failure
+  compute_nav_offsets           createplan.c          Trim offset computation
+
+Appendix B. Data Structure Relationship Diagram
+============================================================================
+
+  Parser Layer
+  --------
+  RPCommonSyntax
+    |--- rpSkipTo: RPSkipTo
+    |--- initial: bool
+    +--- rpPattern: RPRPatternNode* (tree)
+         |--- nodeType: VAR | SEQ | ALT | GROUP
+         |--- min, max: quantifier
+         |--- varName: variable name (VAR only)
+         +--- children: List* (SEQ/ALT/GROUP only)
+
+  Planner Layer
+  ----------
+  WindowAgg (plan node)
+    |--- rpSkipTo: RPSkipTo
+    |--- defineClause: List<TargetEntry>
+    +--- rpPattern: RPRPattern*
+         |--- numVars: int
+         |--- varNames: char**
+         |--- maxDepth: RPRDepth
+         |--- isAbsorbable: bool
+         |--- numElements: int
+         +--- elements: RPRPatternElement[]  (flat array)
+              |--- varId      (1B)
+              |--- depth      (1B)
+              |--- flags      (1B)
+              |--- reserved   (1B)
+              |--- min, max   (4B + 4B)
+              +--- next, jump (2B + 2B)
+
+  Executor Layer
+  ----------
+  WindowAggState
+    |--- rpSkipTo: RPSkipTo (AFTER MATCH SKIP mode)
+    |--- rpPattern: RPRPattern* (copied from plan)
+    |--- defineVariableList: List<String> (variable names, DEFINE order)
+    |--- defineClauseList: List<ExprState>
+    |--- nfaVarMatched: bool[] (per-row cache)
+    |--- nfaVisitedElems: bitmapword* (cycle detection)
+    |--- nfaStateSize: Size (pre-calculated RPRNFAState allocation size)
+    |--- nfaContext <-> nfaContextTail (doubly-linked list)
+    |   +--- RPRNFAContext
+    |       |--- states: RPRNFAState* (linked list)
+    |       |   |--- elemIdx
+    |       |   |--- counts[]
+    |       |   +--- isAbsorbable
+    |       |--- matchStartRow, matchEndRow
+    |       |--- lastProcessedRow
+    |       |--- matchedState (cloned on FIN arrival)
+    |       |--- hasAbsorbableState
+    |       +--- allStatesAbsorbable
+    |--- nfaContextFree (recycling pool)
+    +--- nfaStateFree (recycling pool)
+
+Appendix C. NFA Element Array Examples
+============================================================================
+
+C-1. PATTERN (A B C)
+
+  idx  varId  depth  min  max  next  jump
+  ------------------------------------------
+   0   A      0      1    1    1     -1
+   1   B      0      1    1    2     -1
+   2   C      0      1    1    3     -1
+   3   FIN    0      1    1    -1    -1
+
+C-2. PATTERN (A+ B*)
+
+  idx  varId  depth  min  max  next  jump  flags
+  ------------------------------------------------------------------------
+   0   A      0      1    INF  1     -1    ABSORBABLE | ABSORBABLE_BRANCH
+   1   B      0      0    INF  2     -1
+   2   FIN    0      1    1    -1    -1
+
+  Only A+ is the absorption point (Case 1). Once past A,
+  absorption is permanently disabled for that state.
+
+C-3. PATTERN (A | B | C)
+
+  idx  varId  depth  min  max  next  jump
+  ----------------------------------------
+   0   ALT    0      1    1    1     -1    alternation start
+   1   A      1      1    1    4     2     branch 1 -> FIN, jump -> branch 2
+   2   B      1      1    1    4     3     branch 2 -> FIN, jump -> branch 3
+   3   C      1      1    1    4     -1    branch 3 -> FIN
+   4   FIN    0      1    1    -1    -1
+
+C-4. PATTERN ((A B)+ C)
+
+  idx  varId    depth  min  max  next  jump  flags
+  --------------------------------------------------------------------------
+   0   BEGIN    0      1    INF  1     4     ABSORBABLE_BRANCH
+   1   A        1      1    1    2     -1    ABSORBABLE_BRANCH
+   2   B        1      1    1    3     -1    ABSORBABLE_BRANCH
+   3   END      0      1    INF  4     1     ABSORBABLE | ABSORBABLE_BRANCH
+   4   C        0      1    1    5     -1
+   5   FIN      0      1    1    -1    -1
+
+  Case 2: GROUP+ with {1,1} body VARs. A, B are branches;
+  END is the absorption point. Compare with C-6 (Case 3).
+
+C-5. PATTERN ((A | B)+? C)
+
+  idx  varId    depth  min  max   next  jump  flags
+  -------------------------------------------------------------------
+   0   BEGIN    0      1    INF   1     5     RELUCTANT, group start
+   1   ALT      1      1    1     2     -1    alternation start
+   2   A        2      1    1     4     3     branch 1
+   3   B        2      1    1     4     -1    branch 2
+   4   END      0      1    INF   5     1     RELUCTANT, group end
+   5   C        0      1    1     6     -1
+   6   FIN      0      1    1     -1    -1
+
+C-6. PATTERN ((A+ B)+ C)  -- Absorbability flag example
+
+  idx  varId    depth  min  max   next  jump  flags
+  ---------------------------------------------------------------------------
+   0   BEGIN    0      1    INF   1     4     ABSORBABLE_BRANCH, group start
+   1   A        1      1    INF   2     -1    ABSORBABLE | ABSORBABLE_BRANCH
+   2   B        1      1    1     3     -1
+   3   END      0      1    INF   4     1     group end
+   4   C        0      1    1     5     -1
+   5   FIN      0      1    1     -1    -1
+
+  Recurses from BEGIN into the body -> A matches Case 1 (simple VAR+).
+  A gets ABSORBABLE | ABSORBABLE_BRANCH, BEGIN gets ABSORBABLE_BRANCH.
+  B and END get no flags -> absorption stops once the state advances to B.
+  (See IV-5 Case 3)
+
+C-7. PATTERN ((A+ B | C*)+ D)  -- Per-branch absorption in ALT
+
+  idx  varId    depth  min  max   next  jump  flags
+  ---------------------------------------------------------------------------
+   0   BEGIN    0      1    INF   1     6     ABSORBABLE_BRANCH
+   1   ALT      1      1    1     2     -1    ABSORBABLE_BRANCH
+   2   A        2      1    INF   3     4     ABSORBABLE | ABSORBABLE_BRANCH
+   3   B        2      1    1     5     -1
+   4   C        2      0    INF   5     -1    ABSORBABLE | ABSORBABLE_BRANCH
+   5   END      0      1    INF   6     1     EMPTY_LOOP
+   6   D        0      1    1     7     -1
+   7   FIN      0      1    1     -1    -1
+
+  ALT branches are checked independently for absorbability.
+  Branch 1: A+ matches Case 1 -> A gets ABSORBABLE. B has no flag.
+  Branch 2: C* matches Case 1 -> C gets ABSORBABLE.
+  Both A and C get ABSORBABLE_BRANCH as part of their respective branch
+  paths.
+  END has EMPTY_LOOP: branch 2 (C*) is nullable, making the group body
+  nullable.
+  BEGIN and ALT get ABSORBABLE_BRANCH (on the path to absorbable elements).
+
+============================================================================
+  End of document
+============================================================================
diff --git a/src/backend/executor/execExpr.c b/src/backend/executor/execExpr.c
index 77229141b38..8d8da67e79f 100644
--- a/src/backend/executor/execExpr.c
+++ b/src/backend/executor/execExpr.c
@@ -1222,6 +1222,98 @@ ExecInitExprRec(Expr *node, ExprState *state,
 				break;
 			}
 
+		case T_RPRNavExpr:
+			{
+				/*
+				 * RPR navigation functions (PREV/NEXT/FIRST/LAST) are
+				 * compiled into EEOP_RPR_NAV_SET / EEOP_RPR_NAV_RESTORE
+				 * opcodes instead of a normal function call.  The SET opcode
+				 * swaps ecxt_outertuple to the target row, the argument
+				 * expression is compiled normally (reads from the swapped
+				 * slot), and the RESTORE opcode restores the original slot.
+				 *
+				 * Default offset when offset_arg is NULL: PREV/NEXT: 1
+				 * (physical offset from currentpos) FIRST/LAST: 0 (logical
+				 * offset from match boundary)
+				 */
+				RPRNavExpr *nav = (RPRNavExpr *) node;
+				WindowAggState *winstate;
+
+				Assert(state->parent && IsA(state->parent, WindowAggState));
+				winstate = (WindowAggState *) state->parent;
+
+				/* Emit SET opcode: swap slot to target row */
+				scratch.opcode = EEOP_RPR_NAV_SET;
+				scratch.d.rpr_nav.winstate = winstate;
+				scratch.d.rpr_nav.kind = nav->kind;
+
+				if (nav->kind >= RPR_NAV_PREV_FIRST)
+				{
+					/*
+					 * Compound navigation: allocate array of 2 for inner [0]
+					 * and outer [1] offsets.
+					 */
+					Datum	   *offset_values = palloc_array(Datum, 2);
+					bool	   *offset_isnulls = palloc_array(bool, 2);
+
+					/* Inner offset (default 0 for FIRST/LAST) */
+					if (nav->offset_arg != NULL)
+						ExecInitExprRec(nav->offset_arg, state,
+										&offset_values[0], &offset_isnulls[0]);
+					else
+					{
+						offset_values[0] = Int64GetDatum(0);
+						offset_isnulls[0] = false;
+					}
+
+					/* Outer offset (default 1 for PREV/NEXT) */
+					if (nav->compound_offset_arg != NULL)
+						ExecInitExprRec(nav->compound_offset_arg, state,
+										&offset_values[1], &offset_isnulls[1]);
+					else
+					{
+						offset_values[1] = Int64GetDatum(1);
+						offset_isnulls[1] = false;
+					}
+
+					scratch.d.rpr_nav.offset_value = offset_values;
+					scratch.d.rpr_nav.offset_isnull = offset_isnulls;
+				}
+				else if (nav->offset_arg != NULL)
+				{
+					/* Simple navigation with explicit offset */
+					Datum	   *offset_value = palloc_object(Datum);
+					bool	   *offset_isnull = palloc_object(bool);
+
+					ExecInitExprRec(nav->offset_arg, state,
+									offset_value, offset_isnull);
+					scratch.d.rpr_nav.offset_value = offset_value;
+					scratch.d.rpr_nav.offset_isnull = offset_isnull;
+				}
+				else
+				{
+					/* Simple navigation with default offset */
+					scratch.d.rpr_nav.offset_value = NULL;
+					scratch.d.rpr_nav.offset_isnull = NULL;
+				}
+
+				ExprEvalPushStep(state, &scratch);
+
+				/* Compile the argument expression normally */
+				ExecInitExprRec(nav->arg, state, resv, resnull);
+
+				/* Emit RESTORE opcode: restore original slot */
+				scratch.opcode = EEOP_RPR_NAV_RESTORE;
+				scratch.resvalue = resv;
+				scratch.resnull = resnull;
+				scratch.d.rpr_nav.winstate = winstate;
+				get_typlenbyval(nav->resulttype,
+								&scratch.d.rpr_nav.resulttyplen,
+								&scratch.d.rpr_nav.resulttypbyval);
+				ExprEvalPushStep(state, &scratch);
+				break;
+			}
+
 		case T_FuncExpr:
 			{
 				FuncExpr   *func = (FuncExpr *) node;
diff --git a/src/backend/executor/execExprInterp.c b/src/backend/executor/execExprInterp.c
index 0634af964a9..324b9a962a8 100644
--- a/src/backend/executor/execExprInterp.c
+++ b/src/backend/executor/execExprInterp.c
@@ -57,11 +57,13 @@
 #include "postgres.h"
 
 #include "access/heaptoast.h"
+#include "common/int.h"
 #include "access/tupconvert.h"
 #include "catalog/pg_type.h"
 #include "commands/sequence.h"
 #include "executor/execExpr.h"
 #include "executor/nodeSubplan.h"
+#include "executor/nodeWindowAgg.h"
 #include "funcapi.h"
 #include "miscadmin.h"
 #include "nodes/miscnodes.h"
@@ -586,6 +588,8 @@ ExecInterpExpr(ExprState *state, ExprContext *econtext, bool *isnull)
 		&&CASE_EEOP_WINDOW_FUNC,
 		&&CASE_EEOP_MERGE_SUPPORT_FUNC,
 		&&CASE_EEOP_SUBPLAN,
+		&&CASE_EEOP_RPR_NAV_SET,
+		&&CASE_EEOP_RPR_NAV_RESTORE,
 		&&CASE_EEOP_AGG_STRICT_DESERIALIZE,
 		&&CASE_EEOP_AGG_DESERIALIZE,
 		&&CASE_EEOP_AGG_STRICT_INPUT_CHECK_ARGS,
@@ -2013,6 +2017,24 @@ ExecInterpExpr(ExprState *state, ExprContext *econtext, bool *isnull)
 			EEO_NEXT();
 		}
 
+		/* RPR navigation: swap slot to target row */
+		EEO_CASE(EEOP_RPR_NAV_SET)
+		{
+			ExecEvalRPRNavSet(state, op, econtext);
+			outerslot = econtext->ecxt_outertuple;
+
+			EEO_NEXT();
+		}
+
+		/* RPR navigation: restore slot to original row */
+		EEO_CASE(EEOP_RPR_NAV_RESTORE)
+		{
+			ExecEvalRPRNavRestore(state, op, econtext);
+			outerslot = econtext->ecxt_outertuple;
+
+			EEO_NEXT();
+		}
+
 		/* evaluate a strict aggregate deserialization function */
 		EEO_CASE(EEOP_AGG_STRICT_DESERIALIZE)
 		{
@@ -5988,3 +6010,248 @@ ExecAggPlainTransByRef(AggState *aggstate, AggStatePerTrans pertrans,
 
 	MemoryContextSwitchTo(oldContext);
 }
+
+/*
+ * Extract compound (outer) offset from step data.
+ * For compound nav, offset_value is an array: [0]=inner, [1]=outer.
+ * Returns the outer offset; errors on NULL or negative.
+ * Default is 1 (like PREV/NEXT implicit offset).
+ */
+static int64
+rpr_nav_get_compound_offset(ExprEvalStep *op)
+{
+	int64		val;
+
+	Assert(op->d.rpr_nav.offset_value != NULL);
+
+	if (op->d.rpr_nav.offset_isnull[1])
+		ereport(ERROR,
+				(errcode(ERRCODE_NULL_VALUE_NOT_ALLOWED),
+				 errmsg("row pattern navigation offset must not be null")));
+
+	val = DatumGetInt64(op->d.rpr_nav.offset_value[1]);
+
+	if (val < 0)
+		ereport(ERROR,
+				(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
+				 errmsg("row pattern navigation offset must not be negative")));
+
+	return val;
+}
+
+/*
+ * Evaluate RPR navigation (PREV/NEXT/FIRST/LAST): swap slot to target row.
+ *
+ * Saves the current outertuple into winstate for later restore, computes
+ * the target row position, fetches the corresponding slot from the
+ * tuplestore, and replaces econtext->ecxt_outertuple with it.
+ *
+ * This is called both from the interpreter inline handler and from
+ * JIT-compiled expressions via build_EvalXFunc.
+ */
+void
+ExecEvalRPRNavSet(ExprState *state, ExprEvalStep *op, ExprContext *econtext)
+{
+	WindowAggState *winstate = op->d.rpr_nav.winstate;
+	int64		offset;
+	int64		target_pos;
+	TupleTableSlot *target_slot;
+
+	/* Save current slot for later restore */
+	winstate->nav_saved_outertuple = econtext->ecxt_outertuple;
+
+	/*
+	 * Determine the inner offset.  NULL or negative offsets are errors per
+	 * the SQL standard.
+	 *
+	 * Default offset when offset_arg is NULL: PREV/NEXT: 1 (standard 5.6.2)
+	 * FIRST/LAST and compound: 0 for inner, 1 for outer
+	 */
+	if (op->d.rpr_nav.offset_value != NULL)
+	{
+		if (*op->d.rpr_nav.offset_isnull)
+			ereport(ERROR,
+					(errcode(ERRCODE_NULL_VALUE_NOT_ALLOWED),
+					 errmsg("row pattern navigation offset must not be null")));
+
+		offset = DatumGetInt64(*op->d.rpr_nav.offset_value);
+
+		if (offset < 0)
+			ereport(ERROR,
+					(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
+					 errmsg("row pattern navigation offset must not be negative")));
+	}
+	else
+	{
+		/* Default offset: 1 for simple PREV/NEXT, 0 otherwise */
+		if (op->d.rpr_nav.kind == RPR_NAV_PREV ||
+			op->d.rpr_nav.kind == RPR_NAV_NEXT)
+			offset = 1;
+		else
+			offset = 0;
+	}
+
+	/*
+	 * Calculate target position based on navigation direction.  On overflow,
+	 * use -1 so that ExecRPRNavGetSlot treats it as out of range.
+	 */
+	switch (op->d.rpr_nav.kind)
+	{
+		case RPR_NAV_PREV:
+			if (pg_sub_s64_overflow(winstate->currentpos, offset, &target_pos))
+				target_pos = -1;
+			break;
+		case RPR_NAV_NEXT:
+			if (pg_add_s64_overflow(winstate->currentpos, offset, &target_pos))
+				target_pos = -1;
+			break;
+		case RPR_NAV_FIRST:
+			/* FIRST: offset from match_start, clamped to currentpos */
+			if (pg_add_s64_overflow(winstate->nav_match_start, offset, &target_pos))
+				target_pos = -1;
+			else if (target_pos > winstate->currentpos)
+				target_pos = -1;	/* beyond current match range */
+			break;
+		case RPR_NAV_LAST:
+			/* LAST: offset backward from currentpos, clamped to match_start */
+			if (pg_sub_s64_overflow(winstate->currentpos, offset, &target_pos))
+				target_pos = -1;
+			else if (target_pos < winstate->nav_match_start)
+				target_pos = -1;	/* before match_start */
+			break;
+
+		case RPR_NAV_PREV_FIRST:
+		case RPR_NAV_NEXT_FIRST:
+			{
+				int64		compound_offset;
+				int64		inner_pos;
+
+				/* Inner: match_start + offset */
+				if (pg_add_s64_overflow(winstate->nav_match_start, offset, &inner_pos))
+				{
+					target_pos = -1;
+					break;
+				}
+				if (inner_pos > winstate->currentpos || inner_pos < 0)
+				{
+					target_pos = -1;
+					break;
+				}
+
+				/* Outer offset */
+				compound_offset = rpr_nav_get_compound_offset(op);
+
+				/* Apply outer: PREV subtracts, NEXT adds */
+				if (op->d.rpr_nav.kind == RPR_NAV_PREV_FIRST)
+				{
+					if (pg_sub_s64_overflow(inner_pos, compound_offset, &target_pos))
+						target_pos = -1;
+				}
+				else
+				{
+					if (pg_add_s64_overflow(inner_pos, compound_offset, &target_pos))
+						target_pos = -1;
+				}
+			}
+			break;
+
+		case RPR_NAV_PREV_LAST:
+		case RPR_NAV_NEXT_LAST:
+			{
+				int64		compound_offset;
+				int64		inner_pos;
+
+				/* Inner: currentpos - offset */
+				if (pg_sub_s64_overflow(winstate->currentpos, offset, &inner_pos))
+				{
+					target_pos = -1;
+					break;
+				}
+				if (inner_pos < winstate->nav_match_start)
+				{
+					target_pos = -1;
+					break;
+				}
+
+				/* Outer offset */
+				compound_offset = rpr_nav_get_compound_offset(op);
+
+				/* Apply outer: PREV subtracts, NEXT adds */
+				if (op->d.rpr_nav.kind == RPR_NAV_PREV_LAST)
+				{
+					if (pg_sub_s64_overflow(inner_pos, compound_offset, &target_pos))
+						target_pos = -1;
+				}
+				else
+				{
+					if (pg_add_s64_overflow(inner_pos, compound_offset, &target_pos))
+						target_pos = -1;
+				}
+			}
+			break;
+		default:
+			elog(ERROR, "unrecognized RPR navigation kind: %d",
+				 op->d.rpr_nav.kind);
+			break;
+	}
+
+	/*
+	 * Slot swap elision: if target_pos is the current row, skip the
+	 * tuplestore fetch and slot swap entirely.  This benefits LAST(expr),
+	 * PREV(expr, 0), NEXT(expr, 0), and similar cases.
+	 *
+	 * We must still set nav_saved_outertuple (done above) so that
+	 * EEOP_RPR_NAV_RESTORE is a harmless no-op.
+	 */
+	if (target_pos == winstate->currentpos)
+		return;
+
+	/* Fetch target row slot (returns nav_null_slot if out of range) */
+	target_slot = ExecRPRNavGetSlot(winstate, target_pos);
+
+	/*
+	 * Update econtext to point to the target slot.  Also decompress the new
+	 * slot's attributes since FETCHSOME already ran for the original slot.
+	 * The caller (interpreter or JIT) is responsible for updating any local
+	 * slot cache (e.g. outerslot) from econtext after we return.
+	 */
+	slot_getallattrs(target_slot);
+	econtext->ecxt_outertuple = target_slot;
+}
+
+/*
+ * Evaluate RPR navigation: restore slot to original row.
+ *
+ * Restores econtext->ecxt_outertuple from the saved slot in winstate.
+ * When slot swap was elided (target == currentpos), this is a harmless
+ * no-op since saved and current slots are identical.
+ * The caller is responsible for updating any local slot cache.
+ *
+ * For pass-by-reference result types, the result datum points into
+ * nav_slot's tuple memory.  If a subsequent navigation in the same
+ * expression re-fetches nav_slot for a different position, the old
+ * tuple is freed, leaving a dangling pointer.  We prevent this by
+ * copying pass-by-ref results into per-tuple memory, which survives
+ * until the next ResetExprContext.
+ */
+void
+ExecEvalRPRNavRestore(ExprState *state, ExprEvalStep *op,
+					  ExprContext *econtext)
+{
+	WindowAggState *winstate = op->d.rpr_nav.winstate;
+
+	econtext->ecxt_outertuple = winstate->nav_saved_outertuple;
+
+	/* Stabilize pass-by-ref result against nav_slot re-fetch */
+	if (!op->d.rpr_nav.resulttypbyval &&
+		!*op->resnull)
+	{
+		MemoryContext oldContext;
+
+		oldContext = MemoryContextSwitchTo(econtext->ecxt_per_tuple_memory);
+		*op->resvalue = datumCopy(*op->resvalue,
+								  false,
+								  op->d.rpr_nav.resulttyplen);
+		MemoryContextSwitchTo(oldContext);
+	}
+}
diff --git a/src/backend/executor/execRPR.c b/src/backend/executor/execRPR.c
new file mode 100644
index 00000000000..242ae9c6dcf
--- /dev/null
+++ b/src/backend/executor/execRPR.c
@@ -0,0 +1,1772 @@
+/*-------------------------------------------------------------------------
+ *
+ * execRPR.c
+ *	  NFA-based Row Pattern Recognition engine for window functions.
+ *
+ * This file implements the NFA execution engine for the ROWS BETWEEN
+ * PATTERN clause (SQL Standard Feature R020: Row Pattern Recognition in
+ * Window Functions).
+ *
+ * The engine executes the compiled RPRPattern structure directly, avoiding
+ * regex compilation overhead.  It is called by nodeWindowAgg.c and exposes
+ * the interface declared in executor/execRPR.h.
+ *
+ *
+ * Portions Copyright (c) 1996-2026, PostgreSQL Global Development Group
+ * Portions Copyright (c) 1994, Regents of the University of California
+ *
+ * IDENTIFICATION
+ *	  src/backend/executor/execRPR.c
+ *
+ *-------------------------------------------------------------------------
+ */
+#include "postgres.h"
+
+#include "common/int.h"
+#include "executor/execRPR.h"
+#include "executor/executor.h"
+#include "miscadmin.h"
+#include "optimizer/rpr.h"
+#include "utils/memutils.h"
+
+/*
+ * For the design and execution model of the NFA engine implemented
+ * in this file, see src/backend/executor/README.rpr.
+ */
+
+/* Bitmap macros for NFA cycle detection (cf. bitmapset.c, tidbitmap.c) */
+#define WORDNUM(x)	((x) / BITS_PER_BITMAPWORD)
+#define BITNUM(x)	((x) % BITS_PER_BITMAPWORD)
+
+/* Forward declarations - NFA state management */
+static RPRNFAState *nfa_state_alloc(WindowAggState *winstate);
+static void nfa_state_free(WindowAggState *winstate, RPRNFAState *state);
+static void nfa_state_free_list(WindowAggState *winstate, RPRNFAState *list);
+static RPRNFAState *nfa_state_create(WindowAggState *winstate, int16 elemIdx,
+									 int32 *counts, bool sourceAbsorbable);
+static bool nfa_states_equal(WindowAggState *winstate, RPRNFAState *s1,
+							 RPRNFAState *s2);
+static bool nfa_add_state_unique(WindowAggState *winstate, RPRNFAContext *ctx,
+								 RPRNFAState *state);
+static void nfa_add_matched_state(WindowAggState *winstate, RPRNFAContext *ctx,
+								  RPRNFAState *state, int64 matchEndRow);
+
+/* Forward declarations - NFA context management (internal) */
+static RPRNFAContext *nfa_context_alloc(WindowAggState *winstate);
+static void nfa_unlink_context(WindowAggState *winstate, RPRNFAContext *ctx);
+
+/* Forward declarations - NFA statistics */
+static void nfa_update_length_stats(int64 count, NFALengthStats *stats, int64 newLen);
+static void nfa_record_context_skipped(WindowAggState *winstate, int64 skippedLen);
+static void nfa_record_context_absorbed(WindowAggState *winstate, int64 absorbedLen);
+
+/* Forward declarations - NFA absorption */
+static void nfa_update_absorption_flags(RPRNFAContext *ctx);
+static bool nfa_states_covered(RPRPattern *pattern, RPRNFAContext *older,
+							   RPRNFAContext *newer);
+static void nfa_try_absorb_context(WindowAggState *winstate, RPRNFAContext *ctx);
+static void nfa_absorb_contexts(WindowAggState *winstate);
+
+/* Forward declarations - NFA match and advance */
+static bool nfa_eval_var_match(WindowAggState *winstate,
+							   RPRPatternElement *elem, bool *varMatched);
+static void nfa_match(WindowAggState *winstate, RPRNFAContext *ctx,
+					  bool *varMatched);
+static void nfa_advance_state(WindowAggState *winstate, RPRNFAContext *ctx,
+							  RPRNFAState *state, int64 currentPos);
+static void nfa_route_to_elem(WindowAggState *winstate, RPRNFAContext *ctx,
+							  RPRNFAState *state, RPRPatternElement *nextElem,
+							  int64 currentPos);
+static void nfa_advance_alt(WindowAggState *winstate, RPRNFAContext *ctx,
+							RPRNFAState *state, RPRPatternElement *elem,
+							int64 currentPos);
+static void nfa_advance_begin(WindowAggState *winstate, RPRNFAContext *ctx,
+							  RPRNFAState *state, RPRPatternElement *elem,
+							  int64 currentPos);
+static void nfa_advance_end(WindowAggState *winstate, RPRNFAContext *ctx,
+							RPRNFAState *state, RPRPatternElement *elem,
+							int64 currentPos);
+static void nfa_advance_var(WindowAggState *winstate, RPRNFAContext *ctx,
+							RPRNFAState *state, RPRPatternElement *elem,
+							int64 currentPos);
+static void nfa_advance(WindowAggState *winstate, RPRNFAContext *ctx,
+						int64 currentPos);
+
+/*
+ * NFA-based pattern matching implementation
+ *
+ * These functions implement direct NFA execution using the compiled
+ * RPRPattern structure, avoiding regex compilation overhead.
+ *
+ * Execution Flow: match -> absorb -> advance
+ * -----------------------------------------
+ * The NFA execution follows a three-phase cycle for each row:
+ *
+ * 1. MATCH (convergence): Evaluate all waiting states against current row.
+ *    States on VAR elements are checked against their defining conditions.
+ *    Failed matches are removed, successful ones may transition forward.
+ *    This is a "convergence" phase - the number of states tends to decrease.
+ *
+ * 2. ABSORB: After matching, check if any context can absorb another.
+ *    Context absorption is an optimization that merges equivalent contexts.
+ *    A context can only be absorbed if ALL its states are absorbable.
+ *
+ * 3. ADVANCE (divergence): Expand states through epsilon transitions.
+ *    States advance through ALT (alternation), END (group end), and
+ *    optional elements until reaching VAR or FIN elements where they wait.
+ *    This is a "divergence" phase - ALT creates multiple branch states.
+ *
+ * Key Design Decisions:
+ * ---------------------
+ * - VAR->END transition in match phase: When a simple VAR (max=1) matches
+ *   and the next element is END, we transition immediately in the match
+ *   phase rather than waiting for advance. This is necessary for correct
+ *   absorption: states must be at END to be marked absorbable before the
+ *   absorption check occurs.
+ *
+ * - Optional VAR skip paths: When advance lands on a VAR with min=0,
+ *   we create both a waiting state AND a skip state (like ALT branches).
+ *   This ensures patterns like "A B? C" work correctly - we need a state
+ *   waiting for B AND a state that has already skipped to C.
+ *
+ * - END->END count increment: When transitioning from one END to another
+ *   END within advance, we must increment the outer END's count. This
+ *   handles nested groups like "((A|B)+)+" correctly - exiting the inner
+ *   group counts as one iteration of the outer group.
+ *
+ * - Empty match handling: The initial advance uses currentPos =
+ *   startPos - 1 (before any row is consumed). If FIN is reached via
+ *   epsilon transitions alone, matchEndRow = startPos - 1 < matchStartRow.
+ *   If matchedState is set (FIN was reached), this is an empty match
+ *   (RF_EMPTY_MATCH); otherwise it is unmatched (RF_UNMATCHED).
+ *   For reluctant min=0 patterns (A*?, A??), the skip path reaches
+ *   FIN first and early termination prunes enter paths, yielding an
+ *   immediate empty match result. For greedy patterns (A*), the enter
+ *   path adds VAR states first, then the skip FIN is recorded but VAR
+ *   states survive for later matching.
+ *
+ * Context Absorption Runtime:
+ * ---------------------------
+ * Absorption uses flags computed at planning time (in rpr.c) and two
+ * context-level flags maintained at runtime:
+ *
+ * State-level:
+ *   state.isAbsorbable: true if state is in the absorbable region.
+ *     - Set at creation: elem->flags & RPR_ELEM_ABSORBABLE_BRANCH
+ *     - At transition: prevAbsorbable && (newElem->flags & ABSORBABLE_BRANCH)
+ *     - Monotonic: once false, stays false forever
+ *
+ * Context-level:
+ *   ctx.hasAbsorbableState: can this context absorb others?
+ *     - True if at least one state has isAbsorbable=true
+ *     - Monotonic: true->false only (optimization: skip recalc when false)
+ *
+ *   ctx.allStatesAbsorbable: can this context be absorbed?
+ *     - True if ALL states have isAbsorbable=true
+ *     - Dynamic: can change false->true (when non-absorbable states die)
+ *
+ * Absorption Algorithm:
+ *   For each pair (older Ctx1, newer Ctx2):
+ *   1. Pre-check: Ctx1.hasAbsorbableState && Ctx2.allStatesAbsorbable
+ *      -> If false, skip (fast filter)
+ *   2. Coverage check: For each Ctx2 state with isAbsorbable=true,
+ *      find Ctx1 state with same elemIdx and count >= Ctx2.count
+ *   3. If all Ctx2 absorbable states are covered, absorb Ctx2
+ *
+ * Example: Pattern A+ B
+ *   Row 1: Ctx1 at A (count=1)
+ *   Row 2: Ctx1 at A (count=2), Ctx2 at A (count=1)
+ *   -> Both at same elemIdx (A), Ctx1.count >= Ctx2.count
+ *   -> Ctx2 absorbed
+ *
+ * The asymmetric design (Ctx1 needs hasAbsorbable, Ctx2 needs allAbsorbable)
+ * allows absorption even when Ctx1 has extra non-absorbable states.
+ */
+
+/*
+ * nfa_state_alloc
+ *
+ * Allocate an NFA state, reusing from freeList if available.
+ * freeList is stored in WindowAggState for reuse across match attempts.
+ * Uses flexible array member for counts[].
+ */
+static RPRNFAState *
+nfa_state_alloc(WindowAggState *winstate)
+{
+	RPRNFAState *state;
+
+	/* Try to reuse from free list first */
+	if (winstate->nfaStateFree != NULL)
+	{
+		state = winstate->nfaStateFree;
+		winstate->nfaStateFree = state->next;
+	}
+	else
+	{
+		/* Allocate in partition context for proper lifetime */
+		state = MemoryContextAlloc(winstate->partcontext, winstate->nfaStateSize);
+	}
+
+	/* Initialize entire state to zero */
+	memset(state, 0, winstate->nfaStateSize);
+
+	/* Update statistics */
+	winstate->nfaStatesActive++;
+	winstate->nfaStatesTotalCreated++;
+	winstate->nfaStatesMax = Max(winstate->nfaStatesMax,
+								 winstate->nfaStatesActive);
+
+	return state;
+}
+
+/*
+ * nfa_state_free
+ *
+ * Return a state to the free list for later reuse.
+ */
+static void
+nfa_state_free(WindowAggState *winstate, RPRNFAState *state)
+{
+	winstate->nfaStatesActive--;
+	state->next = winstate->nfaStateFree;
+	winstate->nfaStateFree = state;
+}
+
+/*
+ * nfa_state_free_list
+ *
+ * Return all states in a list to the free list.
+ */
+static void
+nfa_state_free_list(WindowAggState *winstate, RPRNFAState *list)
+{
+	RPRNFAState *next;
+
+	for (; list != NULL; list = next)
+	{
+		next = list->next;
+		nfa_state_free(winstate, list);
+	}
+}
+
+/*
+ * nfa_state_create
+ *
+ * Create a new state with given elemIdx and counts.
+ * isAbsorbable is computed immediately: inherited AND new element's flag.
+ * Monotonic property: once false, stays false through all transitions.
+ *
+ * Caller is responsible for linking the returned state.
+ */
+static RPRNFAState *
+nfa_state_create(WindowAggState *winstate, int16 elemIdx,
+				 int32 *counts, bool sourceAbsorbable)
+{
+	RPRPattern *pattern = winstate->rpPattern;
+	int			maxDepth = pattern->maxDepth;
+	RPRNFAState *state = nfa_state_alloc(winstate);
+	RPRPatternElement *elem = &pattern->elements[elemIdx];
+
+	state->elemIdx = elemIdx;
+	if (counts != NULL && maxDepth > 0)
+		memcpy(state->counts, counts, sizeof(int32) * maxDepth);
+
+	/*
+	 * Compute isAbsorbable immediately at transition time. isAbsorbable =
+	 * sourceAbsorbable && (elem->flags & ABSORBABLE_BRANCH) Monotonic: once
+	 * false, stays false (can't re-enter absorbable region).
+	 */
+	state->isAbsorbable = sourceAbsorbable && RPRElemIsAbsorbableBranch(elem);
+
+	return state;
+}
+
+/*
+ * nfa_states_equal
+ *
+ * Check if two states are equivalent (same elemIdx and counts).
+ */
+static bool
+nfa_states_equal(WindowAggState *winstate, RPRNFAState *s1, RPRNFAState *s2)
+{
+	RPRPattern *pattern = winstate->rpPattern;
+	RPRPatternElement *elem;
+	int			compareDepth;
+
+	if (s1->elemIdx != s2->elemIdx)
+		return false;
+
+	/* Compare counts up to current element's depth */
+	elem = &pattern->elements[s1->elemIdx];
+	compareDepth = elem->depth + 1; /* depth 0 needs 1 count, etc. */
+
+	if (memcmp(s1->counts, s2->counts, sizeof(int32) * compareDepth) != 0)
+		return false;
+
+	return true;
+}
+
+/*
+ * nfa_add_state_unique
+ *
+ * Add a state to ctx->states at the END, only if no duplicate exists.
+ * Returns true if state was added, false if duplicate found (state is freed).
+ * Earlier states have better lexical order (DFS traversal order), so existing wins.
+ */
+static bool
+nfa_add_state_unique(WindowAggState *winstate, RPRNFAContext *ctx, RPRNFAState *state)
+{
+	RPRNFAState *s;
+	RPRNFAState *tail = NULL;
+
+	/* Mark VAR in visited before duplicate check to prevent DFS loops */
+	winstate->nfaVisitedElems[WORDNUM(state->elemIdx)] |=
+		((bitmapword) 1 << BITNUM(state->elemIdx));
+
+	/* Check for duplicate and find tail */
+	for (s = ctx->states; s != NULL; s = s->next)
+	{
+		CHECK_FOR_INTERRUPTS();
+
+		if (nfa_states_equal(winstate, s, state))
+		{
+			/*
+			 * Duplicate found - existing has better lexical order, discard
+			 * new
+			 */
+			nfa_state_free(winstate, state);
+			winstate->nfaStatesMerged++;
+			return false;
+		}
+		tail = s;
+	}
+
+	/* No duplicate, add at end */
+	state->next = NULL;
+	if (tail == NULL)
+		ctx->states = state;
+	else
+		tail->next = state;
+
+	return true;
+}
+
+/*
+ * nfa_add_matched_state
+ *
+ * Record a state that reached FIN, replacing any previous match.
+ *
+ * For SKIP PAST LAST ROW, also prune subsequent contexts whose start row
+ * falls within the match range, as they cannot produce output rows.
+ */
+static void
+nfa_add_matched_state(WindowAggState *winstate, RPRNFAContext *ctx,
+					  RPRNFAState *state, int64 matchEndRow)
+{
+	if (ctx->matchedState != NULL)
+		nfa_state_free(winstate, ctx->matchedState);
+
+	ctx->matchedState = state;
+	state->next = NULL;
+	ctx->matchEndRow = matchEndRow;
+
+	/* Prune contexts that started within this match's range */
+	if (winstate->rpSkipTo == ST_PAST_LAST_ROW)
+	{
+		int64		skippedLen;
+
+		while (ctx->next != NULL &&
+			   ctx->next->matchStartRow <= matchEndRow)
+		{
+			RPRNFAContext *nextCtx = ctx->next;
+
+			Assert(nextCtx->lastProcessedRow >= nextCtx->matchStartRow);
+			skippedLen = nextCtx->lastProcessedRow - nextCtx->matchStartRow + 1;
+			nfa_record_context_skipped(winstate, skippedLen);
+
+			ExecRPRFreeContext(winstate, nextCtx);
+		}
+	}
+}
+
+/*
+ * nfa_context_alloc
+ *
+ * Allocate an NFA context, reusing from free list if available.
+ */
+static RPRNFAContext *
+nfa_context_alloc(WindowAggState *winstate)
+{
+	RPRNFAContext *ctx;
+
+	if (winstate->nfaContextFree != NULL)
+	{
+		ctx = winstate->nfaContextFree;
+		winstate->nfaContextFree = ctx->next;
+	}
+	else
+	{
+		/* Allocate in partition context for proper lifetime */
+		ctx = MemoryContextAlloc(winstate->partcontext, sizeof(RPRNFAContext));
+	}
+
+	ctx->next = NULL;
+	ctx->prev = NULL;
+	ctx->states = NULL;
+	ctx->matchStartRow = -1;
+	ctx->matchEndRow = -1;
+	ctx->lastProcessedRow = -1;
+	ctx->matchedState = NULL;
+
+	/* Initialize two-flag absorption design based on pattern */
+	ctx->hasAbsorbableState = winstate->rpPattern->isAbsorbable;
+	ctx->allStatesAbsorbable = winstate->rpPattern->isAbsorbable;
+
+	/* Update statistics */
+	winstate->nfaContextsActive++;
+	winstate->nfaContextsTotalCreated++;
+	winstate->nfaContextsMax = Max(winstate->nfaContextsMax,
+								   winstate->nfaContextsActive);
+
+	return ctx;
+}
+
+/*
+ * nfa_unlink_context
+ *
+ * Remove a context from the doubly-linked active context list.
+ * Updates head (nfaContext) and tail (nfaContextTail) as needed.
+ */
+static void
+nfa_unlink_context(WindowAggState *winstate, RPRNFAContext *ctx)
+{
+	if (ctx->prev != NULL)
+		ctx->prev->next = ctx->next;
+	else
+		winstate->nfaContext = ctx->next;	/* was head */
+
+	if (ctx->next != NULL)
+		ctx->next->prev = ctx->prev;
+	else
+		winstate->nfaContextTail = ctx->prev;	/* was tail */
+
+	ctx->next = NULL;
+	ctx->prev = NULL;
+}
+
+/*
+ * nfa_update_length_stats
+ *
+ * Helper function to update min/max/total length statistics.
+ * Called when tracking match/mismatch/absorbed/skipped lengths.
+ */
+static void
+nfa_update_length_stats(int64 count, NFALengthStats *stats, int64 newLen)
+{
+	if (count == 1)
+	{
+		stats->min = newLen;
+		stats->max = newLen;
+	}
+	else
+	{
+		stats->min = Min(stats->min, newLen);
+		stats->max = Max(stats->max, newLen);
+	}
+	stats->total += newLen;
+}
+
+/*
+ * nfa_record_context_skipped
+ *
+ * Record a skipped context in statistics.
+ */
+static void
+nfa_record_context_skipped(WindowAggState *winstate, int64 skippedLen)
+{
+	winstate->nfaContextsSkipped++;
+	nfa_update_length_stats(winstate->nfaContextsSkipped,
+							&winstate->nfaSkippedLen,
+							skippedLen);
+}
+
+/*
+ * nfa_record_context_absorbed
+ *
+ * Record an absorbed context in statistics.
+ */
+static void
+nfa_record_context_absorbed(WindowAggState *winstate, int64 absorbedLen)
+{
+	winstate->nfaContextsAbsorbed++;
+	nfa_update_length_stats(winstate->nfaContextsAbsorbed,
+							&winstate->nfaAbsorbedLen,
+							absorbedLen);
+}
+
+/*
+ * nfa_update_absorption_flags
+ *
+ * Update context's absorption flags after state changes.
+ *
+ * Two flags control absorption behavior:
+ *   hasAbsorbableState: true if context has at least one absorbable state.
+ *     This flag is monotonic (true -> false only). Once all absorbable states
+ *     die, no new absorbable states can be created through transitions.
+ *   allStatesAbsorbable: true if ALL states in context are absorbable.
+ *     This flag is dynamic and can change false -> true when non-absorbable
+ *     states die off.
+ *
+ * Optimization: Once hasAbsorbableState becomes false, both flags remain false
+ * permanently, so we skip recalculation.
+ */
+static void
+nfa_update_absorption_flags(RPRNFAContext *ctx)
+{
+	RPRNFAState *state;
+	bool		hasAbsorbable = false;
+	bool		allAbsorbable = true;
+
+	/*
+	 * Optimization: Once hasAbsorbableState becomes false, it stays false. No
+	 * need to recalculate - both flags remain false permanently.
+	 */
+	if (!ctx->hasAbsorbableState)
+	{
+		ctx->allStatesAbsorbable = false;
+		return;
+	}
+
+	/* No states means no absorbable states */
+	if (ctx->states == NULL)
+	{
+		ctx->hasAbsorbableState = false;
+		ctx->allStatesAbsorbable = false;
+		return;
+	}
+
+	/*
+	 * Iterate through all states to check absorption status. Uses
+	 * state->isAbsorbable which tracks if state is in absorbable region. This
+	 * is different from RPRElemIsAbsorbable(elem) which checks judgment
+	 * point.
+	 */
+	for (state = ctx->states; state != NULL; state = state->next)
+	{
+		CHECK_FOR_INTERRUPTS();
+
+		if (state->isAbsorbable)
+			hasAbsorbable = true;
+		else
+			allAbsorbable = false;
+	}
+
+	ctx->hasAbsorbableState = hasAbsorbable;
+	ctx->allStatesAbsorbable = allAbsorbable;
+}
+
+/*
+ * nfa_states_covered
+ *
+ * Check if all states in newer context are "covered" by older context.
+ *
+ * A newer state is covered when older context has an absorbable state at the
+ * same pattern element (elemIdx) with count >= newer's count at that depth.
+ * The covering state must be absorbable because only absorbable states can
+ * guarantee to produce superset matches.
+ *
+ * If all newer states are covered, newer context's eventual matches will be
+ * a subset of older context's matches, making newer redundant.
+ */
+static bool
+nfa_states_covered(RPRPattern *pattern, RPRNFAContext *older, RPRNFAContext *newer)
+{
+	RPRNFAState *newerState;
+
+	for (newerState = newer->states; newerState != NULL; newerState = newerState->next)
+	{
+		RPRNFAState *olderState;
+		RPRPatternElement *elem;
+		int			depth;
+		bool		found = false;
+
+		/* All states are absorbable (caller checks allStatesAbsorbable) */
+		elem = &pattern->elements[newerState->elemIdx];
+		depth = elem->depth;
+
+		/*
+		 * Only compare at absorption judgment points (RPR_ELEM_ABSORBABLE).
+		 * Judgment points are where count-dominance guarantees the newer
+		 * context's future matches are a subset of the older's.
+		 */
+		if (!RPRElemIsAbsorbable(elem))
+			return false;
+
+		for (olderState = older->states; olderState != NULL; olderState = olderState->next)
+		{
+			CHECK_FOR_INTERRUPTS();
+
+			/* Covering state must also be absorbable */
+			if (olderState->isAbsorbable &&
+				olderState->elemIdx == newerState->elemIdx &&
+				olderState->counts[depth] >= newerState->counts[depth])
+			{
+				found = true;
+				break;
+			}
+		}
+
+		if (!found)
+			return false;
+	}
+
+	return true;
+}
+
+/*
+ * nfa_try_absorb_context
+ *
+ * Try to absorb ctx (newer) into an older in-progress context.
+ * Returns true if ctx was absorbed and freed.
+ *
+ * Absorption requires three conditions:
+ *   1. ctx must have all states absorbable (allStatesAbsorbable).
+ *      If ctx has any non-absorbable state, it may produce unique matches.
+ *   2. older must have at least one absorbable state (hasAbsorbableState).
+ *      Without absorbable states, older cannot cover newer's states.
+ *   3. All ctx states must be covered by older's absorbable states.
+ *      This ensures older will produce all matches that ctx would produce.
+ *
+ * Context list is ordered by creation time (oldest first via prev chain).
+ * Each row creates at most one context, so earlier contexts have smaller
+ * matchStartRow values.
+ */
+static void
+nfa_try_absorb_context(WindowAggState *winstate, RPRNFAContext *ctx)
+{
+	RPRPattern *pattern = winstate->rpPattern;
+	RPRNFAContext *older;
+
+	/* Early exit: ctx must have all states absorbable */
+	if (!ctx->allStatesAbsorbable)
+		return;
+
+	for (older = ctx->prev; older != NULL; older = older->prev)
+	{
+		CHECK_FOR_INTERRUPTS();
+
+		/*
+		 * By invariant: ctx->prev chain is in creation order (oldest first),
+		 * and each row creates at most one context. So all contexts in this
+		 * chain have matchStartRow < ctx->matchStartRow.
+		 */
+
+		/* Older must also be in-progress */
+		if (older->states == NULL)
+			continue;
+
+		/* Older must have at least one absorbable state */
+		if (!older->hasAbsorbableState)
+			continue;
+
+		/* Check if all newer states are covered by older */
+		if (nfa_states_covered(pattern, older, ctx))
+		{
+			int64		absorbedLen = ctx->lastProcessedRow - ctx->matchStartRow + 1;
+
+			ExecRPRFreeContext(winstate, ctx);
+			nfa_record_context_absorbed(winstate, absorbedLen);
+			return;
+		}
+	}
+}
+
+/*
+ * nfa_absorb_contexts
+ *
+ * Absorb redundant contexts to reduce memory usage and computation.
+ *
+ * For patterns like A+, newer contexts starting later will produce subset
+ * matches of older contexts with higher counts. By absorbing these redundant
+ * contexts early, we avoid duplicate work.
+ *
+ * Iterates from tail (newest) toward head (oldest) via prev chain.
+ * Only in-progress contexts (states != NULL) are candidates for absorption;
+ * completed contexts represent valid match results.
+ */
+static void
+nfa_absorb_contexts(WindowAggState *winstate)
+{
+	RPRNFAContext *ctx;
+	RPRNFAContext *nextCtx;
+
+	for (ctx = winstate->nfaContextTail; ctx != NULL; ctx = nextCtx)
+	{
+		nextCtx = ctx->prev;
+
+		/*
+		 * Only absorb in-progress contexts; completed contexts are valid
+		 * results
+		 */
+		if (ctx->states != NULL)
+			nfa_try_absorb_context(winstate, ctx);
+	}
+}
+
+/*
+ * nfa_eval_var_match
+ *
+ * Evaluate if a VAR element matches the current row.
+ *
+ * varMatched is a pre-evaluated boolean array indexed by varId, computed
+ * once per row by evaluating all DEFINE expressions.  NULL means no DEFINE
+ * clauses exist (only possible during early development/testing).
+ *
+ * Per SQL:2016 R020, pattern variables not listed in DEFINE are implicitly
+ * TRUE -- they match every row.  This is checked via varId >= list_length.
+ */
+static bool
+nfa_eval_var_match(WindowAggState *winstate, RPRPatternElement *elem,
+				   bool *varMatched)
+{
+	/* This function should only be called for VAR elements */
+	Assert(RPRElemIsVar(elem));
+
+	if (varMatched == NULL)
+		return false;
+	if (elem->varId >= list_length(winstate->defineVariableList))
+		return true;
+	return varMatched[elem->varId];
+}
+
+/*
+ * nfa_match
+ *
+ * Match phase (convergence): evaluate VAR elements against current row.
+ * Only updates counts and removes dead states. Minimal transitions.
+ *
+ * For VAR elements:
+ *   - matched: count++, keep state (unless count > max)
+ *   - not matched: remove state (exit alternatives already exist from
+ *     previous advance when count >= min was satisfied)
+ *
+ * For VARs that reached max count followed by END:
+ *   - Advance through END chain to reach absorption judgment point
+ *   - Only deterministic exits (count >= max, max != INF) are handled
+ *   - Chains through END elements while count >= max (must-exit path)
+ *
+ * Non-VAR elements (ALT, END, FIN) are kept as-is for advance phase.
+ */
+static void
+nfa_match(WindowAggState *winstate, RPRNFAContext *ctx, bool *varMatched)
+{
+	RPRPattern *pattern = winstate->rpPattern;
+	RPRPatternElement *elements = pattern->elements;
+	RPRNFAState **prevPtr = &ctx->states;
+	RPRNFAState *state;
+	RPRNFAState *nextState;
+
+	/*
+	 * Evaluate VAR elements against current row. For VARs that reach max
+	 * count with END next, advance through END chain inline so absorb phase
+	 * can compare states at judgment points.
+	 */
+	for (state = ctx->states; state != NULL; state = nextState)
+	{
+		RPRPatternElement *elem = &elements[state->elemIdx];
+
+		CHECK_FOR_INTERRUPTS();
+
+		nextState = state->next;
+
+		if (RPRElemIsVar(elem))
+		{
+			bool		matched;
+			int			depth = elem->depth;
+			int32		count = state->counts[depth];
+
+			matched = nfa_eval_var_match(winstate, elem, varMatched);
+
+			if (matched)
+			{
+				/* Increment count */
+				if (count < RPR_COUNT_MAX)
+					count++;
+
+				/* Max constraint should not be exceeded */
+				Assert(elem->max == RPR_QUANTITY_INF || count <= elem->max);
+
+				state->counts[depth] = count;
+
+				/*
+				 * For VAR at max count with END next, advance through END
+				 * chain to reach the absorption judgment point.  Only
+				 * deterministic exits (count >= max, max finite) are handled;
+				 * unbounded VARs stay for advance phase.
+				 *
+				 * In nested patterns like ((A B){2}){3}, a VAR reaching its
+				 * max triggers an exit cascade: inner END increments inner
+				 * group count, which may itself reach max, requiring an exit
+				 * to the next outer END.  The loop below walks this chain.
+				 *
+				 * ABSORBABLE_BRANCH marks elements inside the absorbable
+				 * region; ABSORBABLE marks the outermost judgment point where
+				 * count-dominance is evaluated.  We chain through BRANCH
+				 * elements until reaching the ABSORBABLE point or an element
+				 * that can still loop (count < max).
+				 */
+				if (RPRElemIsAbsorbableBranch(elem) &&
+					!RPRElemIsAbsorbable(elem) &&
+					count >= elem->max &&
+					RPRElemIsEnd(&elements[elem->next]))
+				{
+					RPRPatternElement *endElem = &elements[elem->next];
+					int			endDepth = endElem->depth;
+					int32		endCount = state->counts[endDepth];
+
+					/* Increment group count */
+					if (endCount < RPR_COUNT_MAX)
+						endCount++;
+					Assert(endElem->max == RPR_QUANTITY_INF ||
+						   endCount <= endElem->max);
+
+					state->elemIdx = elem->next;
+					state->counts[endDepth] = endCount;
+
+					/*
+					 * Chain through END elements within the absorbable region
+					 * (ABSORBABLE_BRANCH) until reaching the judgment point
+					 * (ABSORBABLE).  Continue only on must-exit path (count
+					 * >= max) with END next.
+					 */
+					while (RPRElemIsAbsorbableBranch(endElem) &&
+						   !RPRElemIsAbsorbable(endElem) &&
+						   endCount >= endElem->max &&
+						   RPRElemIsEnd(&elements[endElem->next]))
+					{
+						RPRPatternElement *outerEnd = &elements[endElem->next];
+						int			outerDepth = outerEnd->depth;
+						int32		outerCount = state->counts[outerDepth];
+
+						/* Reset exited group's count */
+						state->counts[endDepth] = 0;
+
+						/* Increment outer group count */
+						if (outerCount < RPR_COUNT_MAX)
+							outerCount++;
+						Assert(outerEnd->max == RPR_QUANTITY_INF ||
+							   outerCount <= outerEnd->max);
+
+						state->elemIdx = endElem->next;
+						state->counts[outerDepth] = outerCount;
+
+						/* Advance to next END in chain */
+						endElem = outerEnd;
+						endDepth = outerDepth;
+						endCount = outerCount;
+					}
+				}
+				/* else: stay at VAR for advance phase */
+			}
+			else
+			{
+				/*
+				 * Not matched - remove state. Exit alternatives were already
+				 * created by advance phase when count >= min was satisfied.
+				 */
+				*prevPtr = nextState;
+				nfa_state_free(winstate, state);
+				continue;
+			}
+		}
+		/* Non-VAR elements: keep as-is for advance phase */
+
+		prevPtr = &state->next;
+	}
+}
+
+/*
+ * nfa_route_to_elem
+ *
+ * Route state to next element. If VAR, add to ctx->states and process
+ * skip path if optional. Otherwise, continue epsilon expansion via recursion.
+ */
+static void
+nfa_route_to_elem(WindowAggState *winstate, RPRNFAContext *ctx,
+				  RPRNFAState *state, RPRPatternElement *nextElem,
+				  int64 currentPos)
+{
+	if (RPRElemIsVar(nextElem))
+	{
+		RPRNFAState *skipState = NULL;
+
+		/* Create skip state before add_unique, which may free state */
+		if (RPRElemCanSkip(nextElem))
+			skipState = nfa_state_create(winstate, nextElem->next,
+										 state->counts, state->isAbsorbable);
+
+		nfa_add_state_unique(winstate, ctx, state);
+
+		if (skipState != NULL)
+			nfa_advance_state(winstate, ctx, skipState, currentPos);
+	}
+	else
+	{
+		nfa_advance_state(winstate, ctx, state, currentPos);
+	}
+}
+
+/*
+ * nfa_advance_alt
+ *
+ * Handle ALT element: expand all branches in lexical order via DFS.
+ */
+static void
+nfa_advance_alt(WindowAggState *winstate, RPRNFAContext *ctx,
+				RPRNFAState *state, RPRPatternElement *elem,
+				int64 currentPos)
+{
+	RPRPattern *pattern = winstate->rpPattern;
+	RPRPatternElement *elements = pattern->elements;
+	RPRElemIdx	altIdx = elem->next;
+
+	while (altIdx >= 0 && altIdx < pattern->numElements)
+	{
+		RPRPatternElement *altElem = &elements[altIdx];
+		RPRNFAState *newState;
+
+		/* Stop if element is outside ALT scope (not a branch) */
+		if (altElem->depth <= elem->depth)
+			break;
+
+		/* Create independent state for each branch */
+		newState = nfa_state_create(winstate, altIdx,
+									state->counts, state->isAbsorbable);
+
+		/* Recursively process this branch before next */
+		nfa_advance_state(winstate, ctx, newState, currentPos);
+		altIdx = altElem->jump;
+	}
+
+	nfa_state_free(winstate, state);
+}
+
+/*
+ * nfa_advance_begin
+ *
+ * Handle BEGIN element: group entry logic.
+ * BEGIN is only visited at initial group entry (count is always 0).
+ * If min=0, creates a skip path past the group.
+ * Loop-back from END goes directly to first child, bypassing BEGIN.
+ */
+static void
+nfa_advance_begin(WindowAggState *winstate, RPRNFAContext *ctx,
+				  RPRNFAState *state, RPRPatternElement *elem,
+				  int64 currentPos)
+{
+	RPRPattern *pattern = winstate->rpPattern;
+	RPRPatternElement *elements = pattern->elements;
+	RPRNFAState *skipState = NULL;
+
+	state->counts[elem->depth] = 0;
+
+	/* Optional group: create skip path (but don't route yet) */
+	if (elem->min == 0)
+	{
+		skipState = nfa_state_create(winstate, elem->jump,
+									 state->counts, state->isAbsorbable);
+	}
+
+	if (skipState != NULL && RPRElemIsReluctant(elem))
+	{
+		RPRNFAState *savedMatch = ctx->matchedState;
+
+		/* Reluctant: skip first (prefer fewer iterations), enter second */
+		nfa_route_to_elem(winstate, ctx, skipState,
+						  &elements[elem->jump], currentPos);
+
+		/*
+		 * If skip path reached FIN, shortest match is found. Skip group entry
+		 * to prevent longer matches.
+		 */
+		if (ctx->matchedState != savedMatch)
+		{
+			nfa_state_free(winstate, state);
+			return;
+		}
+
+		state->elemIdx = elem->next;
+		nfa_route_to_elem(winstate, ctx, state,
+						  &elements[state->elemIdx], currentPos);
+	}
+	else
+	{
+		/* Greedy: enter first, skip second */
+		state->elemIdx = elem->next;
+		nfa_route_to_elem(winstate, ctx, state,
+						  &elements[state->elemIdx], currentPos);
+
+		if (skipState != NULL)
+		{
+			nfa_route_to_elem(winstate, ctx, skipState,
+							  &elements[elem->jump], currentPos);
+		}
+	}
+}
+
+/*
+ * nfa_advance_end
+ *
+ * Handle END element: group repetition logic.
+ * Decides whether to loop back or exit based on count vs min/max.
+ */
+static void
+nfa_advance_end(WindowAggState *winstate, RPRNFAContext *ctx,
+				RPRNFAState *state, RPRPatternElement *elem,
+				int64 currentPos)
+{
+	RPRPattern *pattern = winstate->rpPattern;
+	RPRPatternElement *elements = pattern->elements;
+	int			depth = elem->depth;
+	int32		count = state->counts[depth];
+
+	if (count < elem->min)
+	{
+		RPRPatternElement *jumpElem;
+		RPRNFAState *ffState = NULL;
+
+		/*----------
+		 * Two paths are explored in parallel when the group body is nullable
+		 * (RPR_ELEM_EMPTY_LOOP):
+		 *
+		 * 1. Primary path: loop back and attempt real matches in the
+		 *    next iteration (state, modified below).
+		 *
+		 * 2. Fast-forward path: skip directly to after the group,
+		 *    treating all remaining required iterations as empty
+		 *    matches (ffState, handled after the primary path).
+		 *
+		 * The snapshot must be taken BEFORE modifying state for the loop-back,
+		 * since both paths diverge from the same point.
+		 *----------
+		 */
+		if (RPRElemCanEmptyLoop(elem))
+			ffState = nfa_state_create(winstate, state->elemIdx,
+									   state->counts, state->isAbsorbable);
+
+		/* Primary path: loop back for real matches */
+		for (int d = depth + 1; d < pattern->maxDepth; d++)
+			state->counts[d] = 0;
+		state->elemIdx = elem->jump;
+		jumpElem = &elements[state->elemIdx];
+		nfa_route_to_elem(winstate, ctx, state, jumpElem,
+						  currentPos);
+
+		/*
+		 * Fast-forward path for nullable bodies.  E.g. (A?){2,3} when A
+		 * doesn't match: the primary loop-back produces empty iterations that
+		 * cycle detection would kill.  Instead, exit directly with count
+		 * satisfied.  Route to elem->next (not nfa_advance_end) to avoid
+		 * creating competing greedy/reluctant loop states.
+		 */
+		if (ffState != NULL)
+		{
+			RPRPatternElement *nextElem;
+
+			ffState->counts[depth] = 0;
+			ffState->elemIdx = elem->next;
+			nextElem = &elements[ffState->elemIdx];
+
+			/* END->END: increment outer END's count */
+			if (RPRElemIsEnd(nextElem) &&
+				ffState->counts[nextElem->depth] < RPR_COUNT_MAX)
+				ffState->counts[nextElem->depth]++;
+
+			nfa_route_to_elem(winstate, ctx, ffState, nextElem,
+							  currentPos);
+		}
+	}
+	else if (elem->max != RPR_QUANTITY_INF && count >= elem->max)
+	{
+		/* Must exit: reached max iterations. */
+		RPRPatternElement *nextElem;
+
+		state->counts[depth] = 0;
+		state->elemIdx = elem->next;
+		nextElem = &elements[state->elemIdx];
+
+		/* Update isAbsorbable for target element (monotonic) */
+		state->isAbsorbable = state->isAbsorbable &&
+			RPRElemIsAbsorbableBranch(nextElem);
+
+		/* END->END: increment outer END's count */
+		if (RPRElemIsEnd(nextElem) && state->counts[nextElem->depth] < RPR_COUNT_MAX)
+			state->counts[nextElem->depth]++;
+
+		nfa_route_to_elem(winstate, ctx, state, nextElem, currentPos);
+	}
+	else
+	{
+		/*
+		 * Between min and max (with at least one iteration) - can exit or
+		 * loop. Greedy: loop first (prefer more iterations). Reluctant: exit
+		 * first (prefer fewer iterations).
+		 */
+		RPRNFAState *exitState;
+		RPRPatternElement *jumpElem;
+		RPRPatternElement *nextElem;
+
+		/*
+		 * Create exit state first (need original counts before modifying
+		 * state)
+		 */
+		exitState = nfa_state_create(winstate, elem->next,
+									 state->counts, state->isAbsorbable);
+		exitState->counts[depth] = 0;
+		nextElem = &elements[exitState->elemIdx];
+
+		/* END->END: increment outer END's count */
+		if (RPRElemIsEnd(nextElem) && exitState->counts[nextElem->depth] < RPR_COUNT_MAX)
+			exitState->counts[nextElem->depth]++;
+
+		/* Prepare loop state */
+		for (int d = depth + 1; d < pattern->maxDepth; d++)
+			state->counts[d] = 0;
+		state->elemIdx = elem->jump;
+		jumpElem = &elements[state->elemIdx];
+
+		if (RPRElemIsReluctant(elem))
+		{
+			RPRNFAState *savedMatch = ctx->matchedState;
+
+			/* Exit first (preferred for reluctant) */
+			nfa_route_to_elem(winstate, ctx, exitState, nextElem,
+							  currentPos);
+
+			/*
+			 * If exit path reached FIN, shortest match is found. Skip loop to
+			 * prevent longer matches from replacing it.
+			 */
+			if (ctx->matchedState != savedMatch)
+			{
+				nfa_state_free(winstate, state);
+				return;
+			}
+
+			/* Loop second */
+			nfa_route_to_elem(winstate, ctx, state, jumpElem,
+							  currentPos);
+		}
+		else
+		{
+			/* Loop first (preferred for greedy) */
+			nfa_route_to_elem(winstate, ctx, state, jumpElem,
+							  currentPos);
+			/* Exit second */
+			nfa_route_to_elem(winstate, ctx, exitState, nextElem,
+							  currentPos);
+		}
+	}
+}
+
+/*
+ * nfa_advance_var
+ *
+ * Handle VAR element: loop/exit transitions.
+ * After match phase, all VAR states have matched - decide next action.
+ */
+static void
+nfa_advance_var(WindowAggState *winstate, RPRNFAContext *ctx,
+				RPRNFAState *state, RPRPatternElement *elem,
+				int64 currentPos)
+{
+	RPRPattern *pattern = winstate->rpPattern;
+	RPRPatternElement *elements = pattern->elements;
+	int			depth = elem->depth;
+	int32		count = state->counts[depth];
+	bool		canLoop = (elem->max == RPR_QUANTITY_INF || count < elem->max);
+	bool		canExit = (count >= elem->min);
+
+	/* After a successful match, count >= 1, so at least one must be true */
+	Assert(canLoop || canExit);
+
+	if (canLoop && canExit)
+	{
+		/*
+		 * Both loop and exit possible. Greedy: loop first (prefer longer
+		 * match). Reluctant: exit first (prefer shorter match).
+		 */
+		RPRNFAState *cloneState;
+		RPRPatternElement *nextElem;
+		bool		reluctant = RPRElemIsReluctant(elem);
+
+		/*
+		 * Clone state for the first-priority path. For greedy, clone is the
+		 * loop state; for reluctant, clone is the exit state.
+		 */
+		if (reluctant)
+		{
+			RPRNFAState *savedMatch = ctx->matchedState;
+
+			/* Clone for exit, original stays for loop */
+			cloneState = nfa_state_create(winstate, elem->next,
+										  state->counts, state->isAbsorbable);
+			cloneState->counts[depth] = 0;
+			nextElem = &elements[cloneState->elemIdx];
+
+			/* When exiting directly to an outer END, increment its count */
+			if (RPRElemIsEnd(nextElem))
+			{
+				if (cloneState->counts[nextElem->depth] < RPR_COUNT_MAX)
+					cloneState->counts[nextElem->depth]++;
+			}
+
+			/* Exit first (preferred for reluctant) */
+			nfa_route_to_elem(winstate, ctx, cloneState, nextElem,
+							  currentPos);
+
+			/*
+			 * If exit path reached FIN, the shortest match is found. Skip
+			 * loop state to prevent longer matches from replacing it.
+			 */
+			if (ctx->matchedState != savedMatch)
+			{
+				nfa_state_free(winstate, state);
+				return;
+			}
+
+			/* Loop second */
+			nfa_add_state_unique(winstate, ctx, state);
+		}
+		else
+		{
+			/* Clone for loop, original used for exit */
+			cloneState = nfa_state_create(winstate, state->elemIdx,
+										  state->counts, state->isAbsorbable);
+
+			/* Loop first (preferred for greedy) */
+			nfa_add_state_unique(winstate, ctx, cloneState);
+
+			/* Exit second */
+			state->counts[depth] = 0;
+			state->elemIdx = elem->next;
+			nextElem = &elements[state->elemIdx];
+
+			/*
+			 * Update isAbsorbable for target element (monotonic: AND
+			 * preserves false)
+			 */
+			state->isAbsorbable = state->isAbsorbable &&
+				RPRElemIsAbsorbableBranch(nextElem);
+
+			/*
+			 * When exiting directly to an outer END, increment its iteration
+			 * count.  Simple VARs (min=max=1) handle this via inline advance
+			 * in nfa_match, but quantified VARs bypass that path.
+			 */
+			if (RPRElemIsEnd(nextElem))
+			{
+				if (state->counts[nextElem->depth] < RPR_COUNT_MAX)
+					state->counts[nextElem->depth]++;
+			}
+
+			nfa_route_to_elem(winstate, ctx, state, nextElem,
+							  currentPos);
+		}
+	}
+	else if (canLoop)
+	{
+		/* Loop only: keep state as-is */
+		nfa_add_state_unique(winstate, ctx, state);
+	}
+	else if (canExit)
+	{
+		/* Exit only: advance to next element */
+		RPRPatternElement *nextElem;
+
+		state->counts[depth] = 0;
+		state->elemIdx = elem->next;
+		nextElem = &elements[state->elemIdx];
+
+		/*
+		 * Update isAbsorbable for target element (monotonic: AND preserves
+		 * false)
+		 */
+		state->isAbsorbable = state->isAbsorbable &&
+			RPRElemIsAbsorbableBranch(nextElem);
+
+		/* See comment above: increment outer END count for quantified VARs */
+		if (RPRElemIsEnd(nextElem))
+		{
+			if (state->counts[nextElem->depth] < RPR_COUNT_MAX)
+				state->counts[nextElem->depth]++;
+		}
+
+		nfa_route_to_elem(winstate, ctx, state, nextElem, currentPos);
+	}
+}
+
+/*
+ * nfa_advance_state
+ *
+ * Recursively process a single state through epsilon transitions.
+ * DFS traversal ensures states are added to ctx->states in lexical order.
+ */
+static void
+nfa_advance_state(WindowAggState *winstate, RPRNFAContext *ctx,
+				  RPRNFAState *state, int64 currentPos)
+{
+	RPRPattern *pattern = winstate->rpPattern;
+	RPRPatternElement *elem;
+
+	Assert(state->elemIdx >= 0 && state->elemIdx < pattern->numElements);
+
+	/* Protect against stack overflow for deeply complex patterns */
+	check_stack_depth();
+
+	/* Cycle detection: if this elemIdx was already visited in this DFS, bail */
+	if (winstate->nfaVisitedElems[WORDNUM(state->elemIdx)] &
+		((bitmapword) 1 << BITNUM(state->elemIdx)))
+	{
+		nfa_state_free(winstate, state);
+		return;
+	}
+
+	elem = &pattern->elements[state->elemIdx];
+
+	/*
+	 * Mark epsilon elements (END, ALT, BEGIN, FIN) in visited to prevent
+	 * infinite epsilon cycles.  VAR elements are marked later when added to
+	 * the state list (nfa_add_state_unique), allowing legitimate loop-back to
+	 * the same VAR in a new iteration.
+	 */
+	if (!RPRElemIsVar(elem))
+		winstate->nfaVisitedElems[WORDNUM(state->elemIdx)] |=
+			((bitmapword) 1 << BITNUM(state->elemIdx));
+
+	switch (elem->varId)
+	{
+		case RPR_VARID_FIN:
+			/* FIN: record match */
+			nfa_add_matched_state(winstate, ctx, state, currentPos);
+			break;
+
+		case RPR_VARID_ALT:
+			nfa_advance_alt(winstate, ctx, state, elem, currentPos);
+			break;
+
+		case RPR_VARID_BEGIN:
+			nfa_advance_begin(winstate, ctx, state, elem, currentPos);
+			break;
+
+		case RPR_VARID_END:
+			nfa_advance_end(winstate, ctx, state, elem, currentPos);
+			break;
+
+		default:
+			/* VAR element */
+			nfa_advance_var(winstate, ctx, state, elem, currentPos);
+			break;
+	}
+}
+
+/*
+ * nfa_advance
+ *
+ * Advance phase (divergence): transition from all surviving states.
+ * Called after match phase with matched VAR states, or at context creation
+ * for initial epsilon expansion (with currentPos = startPos - 1).
+ *
+ * Processes states in order, using recursive DFS to maintain lexical order.
+ */
+static void
+nfa_advance(WindowAggState *winstate, RPRNFAContext *ctx, int64 currentPos)
+{
+	RPRNFAState *states = ctx->states;
+	RPRNFAState *state;
+	RPRNFAState *savedMatchedState;
+
+	ctx->states = NULL;			/* Will rebuild */
+
+	/* Process each state in lexical order (DFS order from previous advance) */
+	while (states != NULL)
+	{
+		CHECK_FOR_INTERRUPTS();
+		savedMatchedState = ctx->matchedState;
+
+		/* Clear visited bitmap before each state's DFS expansion */
+		memset(winstate->nfaVisitedElems, 0,
+			   sizeof(bitmapword) * winstate->nfaVisitedNWords);
+
+		state = states;
+		states = states->next;
+		state->next = NULL;
+
+		nfa_advance_state(winstate, ctx, state, currentPos);
+
+		/*
+		 * Early termination: if a FIN was newly reached in this advance,
+		 * remaining old states have worse lexical order and can be pruned.
+		 * Only check for new FIN arrivals (not ones from previous rows).
+		 */
+		if (ctx->matchedState != savedMatchedState && states != NULL)
+		{
+			nfa_state_free_list(winstate, states);
+			break;
+		}
+	}
+}
+
+
+/***********************************************************************
+ * API exposed to nodeWindowAgg.c
+ ***********************************************************************/
+
+/*
+ * ExecRPRStartContext
+ *
+ * Start a new match context at given position.
+ * Initializes context, state absorption flags, and performs initial advance
+ * to expand epsilon transitions (ALT branches, optional elements).
+ * Adds context to the tail of winstate->nfaContext list.
+ */
+RPRNFAContext *
+ExecRPRStartContext(WindowAggState *winstate, int64 startPos)
+{
+	RPRNFAContext *ctx;
+	RPRPattern *pattern = winstate->rpPattern;
+	RPRPatternElement *elem;
+
+	ctx = nfa_context_alloc(winstate);
+	ctx->matchStartRow = startPos;
+	ctx->states = nfa_state_alloc(winstate);	/* initial state at elem 0 */
+
+	elem = &pattern->elements[0];
+
+	if (RPRElemIsAbsorbableBranch(elem))
+	{
+		ctx->states->isAbsorbable = true;
+	}
+	else
+	{
+		ctx->hasAbsorbableState = false;
+		ctx->allStatesAbsorbable = false;
+		ctx->states->isAbsorbable = false;
+	}
+
+	/* Add to tail of active context list (doubly-linked, oldest-first) */
+	ctx->prev = winstate->nfaContextTail;
+	ctx->next = NULL;
+	if (winstate->nfaContextTail != NULL)
+		winstate->nfaContextTail->next = ctx;
+	else
+		winstate->nfaContext = ctx; /* first context becomes head */
+	winstate->nfaContextTail = ctx;
+
+	/*
+	 * Initial advance (divergence): expand ALT branches and create exit
+	 * states for VAR elements with min=0. This prepares the context for the
+	 * first row's match phase.
+	 *
+	 * Use startPos - 1 as currentPos since no row has been consumed yet. If
+	 * FIN is reached via epsilon transitions, matchEndRow = startPos - 1
+	 * which is less than matchStartRow, resulting in UNMATCHED treatment.
+	 */
+	nfa_advance(winstate, ctx, startPos - 1);
+
+	return ctx;
+}
+
+/*
+ * ExecRPRGetHeadContext
+ *
+ * Return the head context if its start position matches pos.
+ * Returns NULL if no context exists or head doesn't match pos.
+ */
+RPRNFAContext *
+ExecRPRGetHeadContext(WindowAggState *winstate, int64 pos)
+{
+	RPRNFAContext *ctx = winstate->nfaContext;
+
+	/*
+	 * Contexts are sorted by matchStartRow ascending.  If the head context
+	 * doesn't match pos, no context exists for this position.
+	 */
+	if (ctx == NULL || ctx->matchStartRow != pos)
+		return NULL;
+
+	return ctx;
+}
+
+/*
+ * ExecRPRFreeContext
+ *
+ * Unlink context from active list and return it to free list.
+ * Also frees any states in the context.
+ */
+void
+ExecRPRFreeContext(WindowAggState *winstate, RPRNFAContext *ctx)
+{
+	/* Unlink from active list first */
+	nfa_unlink_context(winstate, ctx);
+
+	/* Update statistics */
+	winstate->nfaContextsActive--;
+
+	if (ctx->states != NULL)
+		nfa_state_free_list(winstate, ctx->states);
+	if (ctx->matchedState != NULL)
+		nfa_state_free(winstate, ctx->matchedState);
+
+	ctx->states = NULL;
+	ctx->matchedState = NULL;
+	ctx->next = winstate->nfaContextFree;
+	winstate->nfaContextFree = ctx;
+}
+
+/*
+ * ExecRPRRecordContextSuccess
+ *
+ * Record a successful context in statistics.
+ */
+void
+ExecRPRRecordContextSuccess(WindowAggState *winstate, int64 matchLen)
+{
+	winstate->nfaMatchesSucceeded++;
+	nfa_update_length_stats(winstate->nfaMatchesSucceeded,
+							&winstate->nfaMatchLen,
+							matchLen);
+}
+
+/*
+ * ExecRPRRecordContextFailure
+ *
+ * Record a failed context in statistics.
+ * If failedLen == 1, count as pruned (failed on first row).
+ * If failedLen > 1, count as mismatched and update length stats.
+ */
+void
+ExecRPRRecordContextFailure(WindowAggState *winstate, int64 failedLen)
+{
+	if (failedLen == 1)
+	{
+		winstate->nfaContextsPruned++;
+	}
+	else
+	{
+		winstate->nfaMatchesFailed++;
+		nfa_update_length_stats(winstate->nfaMatchesFailed,
+								&winstate->nfaFailLen,
+								failedLen);
+	}
+}
+
+/*
+ * nfa_reevaluate_dependent_vars
+ *		Re-evaluate match_start-dependent DEFINE variables for a specific
+ *		context whose matchStartRow differs from the shared evaluation's
+ *		nav_match_start.
+ *
+ * Only variables in defineMatchStartDependent are re-evaluated.  The
+ * current row's slot (ecxt_outertuple) must already be set up by
+ * nfa_evaluate_row().
+ */
+static void
+nfa_reevaluate_dependent_vars(WindowAggState *winstate, RPRNFAContext *ctx,
+							  int64 currentPos)
+{
+	ExprContext *econtext = winstate->ss.ps.ps_ExprContext;
+	int64		saved_match_start = winstate->nav_match_start;
+	int64		saved_pos = winstate->currentpos;
+	int			varIdx = 0;
+	ListCell   *lc;
+
+	/* Temporarily set nav_match_start and currentpos for FIRST/LAST */
+	winstate->nav_match_start = ctx->matchStartRow;
+	winstate->currentpos = currentPos;
+
+	/* Invalidate nav_slot cache since match_start changed */
+	winstate->nav_slot_pos = -1;
+
+	foreach(lc, winstate->defineClauseList)
+	{
+		if (bms_is_member(varIdx, winstate->defineMatchStartDependent))
+		{
+			ExprState  *exprState = (ExprState *) lfirst(lc);
+			Datum		result;
+			bool		isnull;
+
+			result = ExecEvalExpr(exprState, econtext, &isnull);
+			winstate->nfaVarMatched[varIdx] = (!isnull && DatumGetBool(result));
+		}
+
+		varIdx++;
+		if (varIdx >= list_length(winstate->defineVariableList))
+			break;
+	}
+
+	/* Restore original match_start, currentpos, and invalidate cache */
+	winstate->nav_match_start = saved_match_start;
+	winstate->currentpos = saved_pos;
+	winstate->nav_slot_pos = -1;
+}
+
+/*
+ * ExecRPRProcessRow
+ *
+ * Process all contexts for one row:
+ *   1. Match all contexts (convergence) - evaluate VARs, prune dead states
+ *   2. Absorb redundant contexts - ideal timing after convergence
+ *   3. Advance all contexts (divergence) - create new states for next row
+ */
+void
+ExecRPRProcessRow(WindowAggState *winstate, int64 currentPos,
+				  bool hasLimitedFrame, int64 frameOffset)
+{
+	RPRNFAContext *ctx;
+	bool	   *varMatched = winstate->nfaVarMatched;
+	bool		hasDependent = !bms_is_empty(winstate->defineMatchStartDependent);
+
+	/* Allow query cancellation once per row for simple/low-state patterns */
+	CHECK_FOR_INTERRUPTS();
+
+	/*
+	 * Phase 1: Match all contexts (convergence).  Evaluate VAR elements,
+	 * update counts, remove dead states.
+	 */
+	for (ctx = winstate->nfaContext; ctx != NULL; ctx = ctx->next)
+	{
+		if (ctx->states == NULL)
+			continue;
+
+		/* Check frame boundary - finalize if exceeded */
+		if (hasLimitedFrame)
+		{
+			int64		ctxFrameEnd;
+
+			/* Clamp to INT64_MAX on overflow */
+			if (pg_add_s64_overflow(ctx->matchStartRow, frameOffset + 1,
+									&ctxFrameEnd))
+				ctxFrameEnd = PG_INT64_MAX;
+
+			if (currentPos >= ctxFrameEnd)
+			{
+				/* Frame boundary exceeded: force mismatch */
+				nfa_match(winstate, ctx, NULL);
+				continue;
+			}
+		}
+
+		/*
+		 * If this context has a different matchStartRow than the one used in
+		 * the shared evaluation, re-evaluate match_start-dependent variables
+		 * with this context's matchStartRow.
+		 */
+		if (hasDependent && ctx->matchStartRow != winstate->nav_match_start)
+			nfa_reevaluate_dependent_vars(winstate, ctx, currentPos);
+		nfa_match(winstate, ctx, varMatched);
+		ctx->lastProcessedRow = currentPos;
+	}
+
+	/*
+	 * Phase 2: Absorb redundant contexts.  After match phase, states have
+	 * converged - ideal for absorption.  First update absorption flags that
+	 * may have changed due to state removal.
+	 */
+	if (winstate->rpPattern->isAbsorbable)
+	{
+		for (ctx = winstate->nfaContext; ctx != NULL; ctx = ctx->next)
+			nfa_update_absorption_flags(ctx);
+
+		nfa_absorb_contexts(winstate);
+	}
+
+	/*
+	 * Phase 3: Advance all contexts (divergence).  Create new states
+	 * (loop/exit) from surviving matched states.
+	 */
+	for (ctx = winstate->nfaContext; ctx != NULL; ctx = ctx->next)
+	{
+		if (ctx->states == NULL)
+			continue;
+
+		/*
+		 * Phase 1 already handled frame boundary exceeded contexts by forcing
+		 * mismatch (nfa_match with NULL), which removes all states (all
+		 * states are at VAR positions after advance). So any surviving
+		 * context here must be within its frame boundary.
+		 */
+		Assert(!hasLimitedFrame ||
+			   ctx->matchStartRow > PG_INT64_MAX - frameOffset - 1 ||
+			   currentPos < ctx->matchStartRow + frameOffset + 1);
+
+		nfa_advance(winstate, ctx, currentPos);
+	}
+}
+
+/*
+ * ExecRPRCleanupDeadContexts
+ *
+ * Remove contexts that have failed (no active states and no match).
+ * These are contexts that failed during normal processing and should be
+ * counted as pruned (if length 1) or mismatched (if length > 1).
+ */
+void
+ExecRPRCleanupDeadContexts(WindowAggState *winstate, RPRNFAContext *excludeCtx)
+{
+	RPRNFAContext *ctx;
+	RPRNFAContext *next;
+
+	for (ctx = winstate->nfaContext; ctx != NULL; ctx = next)
+	{
+		CHECK_FOR_INTERRUPTS();
+
+		next = ctx->next;
+
+		/* Skip the target context and contexts still processing */
+		if (ctx == excludeCtx || ctx->states != NULL)
+			continue;
+
+		/* Skip successfully matched contexts (will be handled by SKIP logic) */
+		if (ctx->matchEndRow >= ctx->matchStartRow)
+			continue;
+
+		/*
+		 * This is a failed context - count and remove it. Only count if it
+		 * actually processed its start row. Contexts created for
+		 * beyond-partition rows are silently removed.
+		 */
+		if (ctx->lastProcessedRow >= ctx->matchStartRow)
+		{
+			int64		failedLen = ctx->lastProcessedRow - ctx->matchStartRow + 1;
+
+			ExecRPRRecordContextFailure(winstate, failedLen);
+		}
+		/* else: context was never processed (beyond-partition), just remove */
+
+		ExecRPRFreeContext(winstate, ctx);
+	}
+}
+
+/*
+ * ExecRPRFinalizeAllContexts
+ *
+ * Finalize all active contexts when partition ends.
+ * Match with NULL to force mismatch, then advance to process epsilon transitions.
+ */
+void
+ExecRPRFinalizeAllContexts(WindowAggState *winstate, int64 lastPos)
+{
+	RPRNFAContext *ctx;
+
+	for (ctx = winstate->nfaContext; ctx != NULL; ctx = ctx->next)
+	{
+		CHECK_FOR_INTERRUPTS();
+
+		if (ctx->states != NULL)
+		{
+			nfa_match(winstate, ctx, NULL);
+			nfa_advance(winstate, ctx, lastPos);
+		}
+	}
+}
diff --git a/src/backend/executor/meson.build b/src/backend/executor/meson.build
index dc45be0b2ce..0ff4a5b1d83 100644
--- a/src/backend/executor/meson.build
+++ b/src/backend/executor/meson.build
@@ -13,6 +13,7 @@ backend_sources += files(
   'execParallel.c',
   'execPartition.c',
   'execProcnode.c',
+  'execRPR.c',
   'execReplication.c',
   'execSRF.c',
   'execScan.c',
diff --git a/src/backend/executor/nodeWindowAgg.c b/src/backend/executor/nodeWindowAgg.c
index f52a7aae843..93cb9bbdd11 100644
--- a/src/backend/executor/nodeWindowAgg.c
+++ b/src/backend/executor/nodeWindowAgg.c
@@ -36,15 +36,19 @@
 #include "access/htup_details.h"
 #include "catalog/objectaccess.h"
 #include "catalog/pg_aggregate.h"
+#include "common/int.h"
 #include "catalog/pg_proc.h"
 #include "common/int.h"
 #include "executor/executor.h"
+#include "executor/execRPR.h"
 #include "executor/instrument.h"
 #include "executor/nodeWindowAgg.h"
 #include "miscadmin.h"
 #include "nodes/nodeFuncs.h"
+#include "nodes/plannodes.h"
 #include "optimizer/clauses.h"
 #include "optimizer/optimizer.h"
+#include "optimizer/rpr.h"
 #include "parser/parse_agg.h"
 #include "parser/parse_coerce.h"
 #include "utils/acl.h"
@@ -173,6 +177,7 @@ typedef struct WindowStatePerAggData
 	bool		restart;		/* need to restart this agg in this cycle? */
 } WindowStatePerAggData;
 
+
 static void initialize_windowaggregate(WindowAggState *winstate,
 									   WindowStatePerFunc perfuncstate,
 									   WindowStatePerAgg peraggstate);
@@ -209,6 +214,9 @@ static Datum GetAggInitVal(Datum textInitVal, Oid transtype);
 
 static bool are_peers(WindowAggState *winstate, TupleTableSlot *slot1,
 					  TupleTableSlot *slot2);
+static int	WinGetSlotInFrame(WindowObject winobj, TupleTableSlot *slot,
+							  int relpos, int seektype, bool set_mark,
+							  bool *isnull, bool *isout);
 static bool window_gettupleslot(WindowObject winobj, int64 pos,
 								TupleTableSlot *slot);
 
@@ -227,6 +235,19 @@ static uint8 get_notnull_info(WindowObject winobj,
 							  int64 pos, int argno);
 static void put_notnull_info(WindowObject winobj,
 							 int64 pos, int argno, bool isnull);
+static bool rpr_is_defined(WindowAggState *winstate);
+static int64 row_is_in_reduced_frame(WindowObject winobj, int64 pos);
+
+static void clear_reduced_frame(WindowAggState *winstate);
+static int	get_reduced_frame_status(WindowAggState *winstate, int64 pos);
+static void update_reduced_frame(WindowObject winobj, int64 pos);
+
+/* Forward declarations - NFA row evaluation */
+static bool nfa_evaluate_row(WindowObject winobj, int64 pos, bool *varMatched);
+
+/* Forward declarations - navigation offset evaluation */
+static void eval_nav_max_offset(WindowAggState *winstate, List *defineClause);
+static void eval_nav_first_offset(WindowAggState *winstate, List *defineClause);
 
 /*
  * Not null info bit array consists of 2-bit items
@@ -820,6 +841,9 @@ eval_windowaggregates(WindowAggState *winstate)
 	 *	   transition function, or
 	 *	 - we have an EXCLUSION clause, or
 	 *	 - if the new frame doesn't overlap the old one
+	 *   - if RPR (Row Pattern Recognition) is enabled, because the reduced
+	 *     frame depends on pattern matching results which can differ entirely
+	 *     from row to row, making inverse transition optimization inapplicable
 	 *
 	 * Note that we don't strictly need to restart in the last case, but if
 	 * we're going to remove all rows from the aggregation anyway, a restart
@@ -834,7 +858,8 @@ eval_windowaggregates(WindowAggState *winstate)
 			(winstate->aggregatedbase != winstate->frameheadpos &&
 			 !OidIsValid(peraggstate->invtransfn_oid)) ||
 			(winstate->frameOptions & FRAMEOPTION_EXCLUSION) ||
-			winstate->aggregatedupto <= winstate->frameheadpos)
+			winstate->aggregatedupto <= winstate->frameheadpos ||
+			rpr_is_defined(winstate))
 		{
 			peraggstate->restart = true;
 			numaggs_restart++;
@@ -963,6 +988,14 @@ eval_windowaggregates(WindowAggState *winstate)
 	{
 		winstate->aggregatedupto = winstate->frameheadpos;
 		ExecClearTuple(agg_row_slot);
+
+		/*
+		 * If RPR is defined, we do not use aggregatedupto_nonrestarted.  To
+		 * avoid assertion failure below, we reset aggregatedupto_nonrestarted
+		 * to frameheadpos.
+		 */
+		if (rpr_is_defined(winstate))
+			aggregatedupto_nonrestarted = winstate->frameheadpos;
 	}
 
 	/*
@@ -974,7 +1007,7 @@ eval_windowaggregates(WindowAggState *winstate)
 	 */
 	for (;;)
 	{
-		int			ret;
+		int64		ret;
 
 		/* Fetch next row if we didn't already */
 		if (TupIsNull(agg_row_slot))
@@ -992,9 +1025,40 @@ eval_windowaggregates(WindowAggState *winstate)
 							  agg_row_slot, false);
 		if (ret < 0)
 			break;
+
 		if (ret == 0)
 			goto next_tuple;
 
+		if (rpr_is_defined(winstate))
+		{
+			/*
+			 * If currentpos is already decided but aggregatedupto is not yet
+			 * determined, we've passed the last reduced frame.
+			 */
+			if (get_reduced_frame_status(winstate, winstate->currentpos)
+				!= RF_NOT_DETERMINED &&
+				get_reduced_frame_status(winstate, winstate->aggregatedupto)
+				== RF_NOT_DETERMINED)
+				break;
+
+			/*
+			 * Calculate the reduced frame for aggregatedupto.
+			 */
+			ret = row_is_in_reduced_frame(winstate->agg_winobj,
+										  winstate->aggregatedupto);
+			if (ret == -1)		/* unmatched row */
+				break;
+
+			/*
+			 * Check if current row is inside a match but not the head
+			 * (skipped), and it's the base row for aggregation.
+			 */
+			if (get_reduced_frame_status(winstate,
+										 winstate->aggregatedupto) == RF_SKIPPED &&
+				winstate->aggregatedupto == winstate->aggregatedbase)
+				break;
+		}
+
 		/* Set tuple context for evaluation of aggregate arguments */
 		winstate->tmpcontext->ecxt_outertuple = agg_row_slot;
 
@@ -1023,6 +1087,7 @@ next_tuple:
 		ExecClearTuple(agg_row_slot);
 	}
 
+
 	/* The frame's end is not supposed to move backwards, ever */
 	Assert(aggregatedupto_nonrestarted <= winstate->aggregatedupto);
 
@@ -1190,6 +1255,28 @@ prepare_tuplestore(WindowAggState *winstate)
 		}
 	}
 
+	/* Create read/mark pointers for RPR navigation if needed */
+	if (winstate->nav_winobj)
+	{
+		/*
+		 * Allocate mark and read pointers for RPR navigation.
+		 *
+		 * If navMaxOffsetKind == RPR_NAV_OFFSET_FIXED, we advance the mark
+		 * based on (currentpos - navMaxOffset) and optionally
+		 * (nfaContext->matchStartRow + navFirstOffset), allowing
+		 * tuplestore_trim() to free rows that are no longer reachable.
+		 *
+		 * RPR_NAV_OFFSET_NEEDS_EVAL is resolved at executor init; by this
+		 * point it is either FIXED or RETAIN_ALL.
+		 */
+		winstate->nav_winobj->markptr =
+			tuplestore_alloc_read_pointer(winstate->buffer, 0);
+		winstate->nav_winobj->readptr =
+			tuplestore_alloc_read_pointer(winstate->buffer,
+										  EXEC_FLAG_BACKWARD);
+		winstate->nav_winobj->markpos = 0;
+	}
+
 	/*
 	 * If we are in RANGE or GROUPS mode, then determining frame boundaries
 	 * requires physical access to the frame endpoint rows, except in certain
@@ -1246,6 +1333,8 @@ begin_partition(WindowAggState *winstate)
 	winstate->framehead_valid = false;
 	winstate->frametail_valid = false;
 	winstate->grouptail_valid = false;
+	if (rpr_is_defined(winstate))
+		clear_reduced_frame(winstate);
 	winstate->spooled_rows = 0;
 	winstate->currentpos = 0;
 	winstate->frameheadpos = 0;
@@ -1299,6 +1388,13 @@ begin_partition(WindowAggState *winstate)
 		winstate->aggregatedupto = 0;
 	}
 
+	/* reset mark and seek positions for RPR navigation */
+	if (winstate->nav_winobj)
+	{
+		winstate->nav_winobj->markpos = -1;
+		winstate->nav_winobj->seekpos = -1;
+	}
+
 	/* reset mark and seek positions for each real window function */
 	for (int i = 0; i < numfuncs; i++)
 	{
@@ -1467,6 +1563,18 @@ release_partition(WindowAggState *winstate)
 		tuplestore_clear(winstate->buffer);
 	winstate->partition_spooled = false;
 	winstate->next_partition = true;
+
+	/* Reset RPR match results */
+	clear_reduced_frame(winstate);
+
+	/* Reset NFA state for new partition */
+	winstate->nfaContext = NULL;
+	winstate->nfaContextTail = NULL;
+	winstate->nfaContextFree = NULL;
+	winstate->nfaStateFree = NULL;
+	winstate->nfaLastProcessedRow = -1;
+	winstate->nfaStatesActive = 0;
+	winstate->nfaContextsActive = 0;
 }
 
 /*
@@ -2395,6 +2503,16 @@ ExecWindowAgg(PlanState *pstate)
 		/* don't evaluate the window functions when we're in pass-through mode */
 		if (winstate->status == WINDOWAGG_RUN)
 		{
+			/*
+			 * If RPR is defined and skip mode is next row, clear the current
+			 * match so the next row triggers re-evaluation.
+			 */
+			if (rpr_is_defined(winstate))
+			{
+				if (winstate->rpSkipTo == ST_NEXT_ROW)
+					clear_reduced_frame(winstate);
+			}
+
 			/*
 			 * Evaluate true window functions
 			 */
@@ -2434,6 +2552,45 @@ ExecWindowAgg(PlanState *pstate)
 		if (winstate->grouptail_ptr >= 0)
 			update_grouptailpos(winstate);
 
+		/*
+		 * Advance RPR navigation mark pointer if possible, so that
+		 * tuplestore_trim() can free rows no longer reachable by navigation.
+		 */
+		if (winstate->nav_winobj &&
+			winstate->rpPattern != NULL &&
+			winstate->navMaxOffsetKind == RPR_NAV_OFFSET_FIXED)
+		{
+			int64		navmarkpos;
+
+			/* Backward reach from PREV/LAST/compound PREV_LAST/NEXT_LAST */
+			if (winstate->currentpos > winstate->navMaxOffset)
+				navmarkpos = winstate->currentpos - winstate->navMaxOffset;
+			else
+				navmarkpos = 0;
+
+			/*
+			 * If FIRST is used, also consider match_start + navFirstOffset.
+			 * The oldest active context (nfaContext) has the smallest
+			 * matchStartRow.
+			 */
+			if (winstate->hasFirstNav &&
+				winstate->navFirstOffsetKind == RPR_NAV_OFFSET_FIXED &&
+				winstate->nfaContext != NULL)
+			{
+				int64		firstreach;
+
+				if (winstate->navFirstOffset > -winstate->nfaContext->matchStartRow)
+					firstreach = winstate->nfaContext->matchStartRow
+						+ winstate->navFirstOffset;
+				else
+					firstreach = 0;
+				navmarkpos = Min(navmarkpos, firstreach);
+			}
+
+			if (navmarkpos > winstate->nav_winobj->markpos)
+				WinSetMarkPosition(winstate->nav_winobj, navmarkpos);
+		}
+
 		/*
 		 * Truncate any no-longer-needed rows from the tuplestore.
 		 */
@@ -2659,6 +2816,20 @@ ExecInitWindowAgg(WindowAgg *node, EState *estate, int eflags)
 	winstate->temp_slot_2 = ExecInitExtraTupleSlot(estate, scanDesc,
 												   &TTSOpsMinimalTuple);
 
+	if (node->rpPattern != NULL)
+	{
+		winstate->nav_slot = ExecInitExtraTupleSlot(estate, scanDesc,
+													&TTSOpsMinimalTuple);
+		winstate->nav_slot_pos = -1;
+
+		winstate->nav_null_slot = ExecInitExtraTupleSlot(estate, scanDesc,
+														 &TTSOpsMinimalTuple);
+		winstate->nav_null_slot = ExecStoreAllNullTuple(winstate->nav_null_slot);
+
+		winstate->nav_saved_outertuple = NULL;
+		winstate->nav_match_start = 0;
+	}
+
 	/*
 	 * create frame head and tail slots only if needed (must create slots in
 	 * exactly the same cases that update_frameheadpos and update_frametailpos
@@ -2827,6 +2998,23 @@ ExecInitWindowAgg(WindowAgg *node, EState *estate, int eflags)
 		winstate->agg_winobj = agg_winobj;
 	}
 
+	/*
+	 * Set up WindowObject for RPR navigation opcodes.  This is separate from
+	 * agg_winobj because it needs its own read pointer to avoid interfering
+	 * with aggregate processing.
+	 */
+	if (node->rpPattern != NULL)
+	{
+		WindowObject nav_winobj = makeNode(WindowObjectData);
+
+		nav_winobj->winstate = winstate;
+		nav_winobj->argstates = NIL;
+		nav_winobj->localmem = NULL;
+		nav_winobj->markptr = -1;
+		nav_winobj->readptr = -1;
+		winstate->nav_winobj = nav_winobj;
+	}
+
 	/* Set the status to running */
 	winstate->status = WINDOWAGG_RUN;
 
@@ -2845,6 +3033,81 @@ ExecInitWindowAgg(WindowAgg *node, EState *estate, int eflags)
 	winstate->inRangeAsc = node->inRangeAsc;
 	winstate->inRangeNullsFirst = node->inRangeNullsFirst;
 
+	/* Set up SKIP TO type */
+	winstate->rpSkipTo = node->rpSkipTo;
+	/* Set up row pattern recognition PATTERN clause (compiled NFA) */
+	winstate->rpPattern = node->rpPattern;
+	/* Set up nav offsets for tuplestore trim */
+	winstate->navMaxOffsetKind = node->navMaxOffsetKind;
+	winstate->navMaxOffset = node->navMaxOffset;
+	if (winstate->navMaxOffsetKind == RPR_NAV_OFFSET_NEEDS_EVAL)
+		eval_nav_max_offset(winstate, node->defineClause);
+	winstate->hasFirstNav = node->hasFirstNav;
+	winstate->navFirstOffsetKind = node->navFirstOffsetKind;
+	winstate->navFirstOffset = node->navFirstOffset;
+	if (winstate->hasFirstNav &&
+		winstate->navFirstOffsetKind == RPR_NAV_OFFSET_NEEDS_EVAL)
+		eval_nav_first_offset(winstate, node->defineClause);
+
+	/* Copy match_start dependency bitmapset for per-context evaluation */
+	winstate->defineMatchStartDependent = bms_copy(node->defineMatchStartDependent);
+
+	/* Calculate NFA state size and allocate cycle detection bitmap */
+	if (node->rpPattern != NULL)
+	{
+		winstate->nfaStateSize = offsetof(RPRNFAState, counts) +
+			sizeof(int32) * node->rpPattern->maxDepth;
+		winstate->nfaVisitedNWords =
+			(node->rpPattern->numElements - 1) / BITS_PER_BITMAPWORD + 1;
+		winstate->nfaVisitedElems = palloc0(sizeof(bitmapword) *
+											winstate->nfaVisitedNWords);
+	}
+
+	/* Set up row pattern recognition DEFINE clause */
+	winstate->defineVariableList = NIL;
+	winstate->defineClauseList = NIL;
+	if (node->defineClause != NIL)
+	{
+		/*
+		 * Compile DEFINE clause expressions.  PREV/NEXT navigation is handled
+		 * by EEOP_RPR_NAV_SET/RESTORE opcodes emitted during ExecInitExpr, so
+		 * no varno rewriting is needed here.
+		 */
+		foreach(l, node->defineClause)
+		{
+			TargetEntry *te = lfirst(l);
+			char	   *name = te->resname;
+			Expr	   *expr = te->expr;
+			ExprState  *exps;
+
+			winstate->defineVariableList =
+				lappend(winstate->defineVariableList,
+						makeString(pstrdup(name)));
+			exps = ExecInitExpr(expr, (PlanState *) winstate);
+			winstate->defineClauseList =
+				lappend(winstate->defineClauseList, exps);
+		}
+	}
+
+	/* Initialize NFA free lists for row pattern matching */
+	winstate->nfaContext = NULL;
+	winstate->nfaContextTail = NULL;
+	winstate->nfaContextFree = NULL;
+	winstate->nfaStateFree = NULL;
+	winstate->nfaLastProcessedRow = -1;
+	winstate->nfaStatesActive = 0;
+	winstate->nfaContextsActive = 0;
+
+	/*
+	 * Allocate varMatched array for NFA evaluation. With the new varNames
+	 * ordering (DEFINE order first), varId == defineIdx for all defined
+	 * variables, so no mapping is needed.
+	 */
+	if (list_length(winstate->defineVariableList) > 0)
+		winstate->nfaVarMatched = palloc0(sizeof(bool) *
+										  list_length(winstate->defineVariableList));
+	else
+		winstate->nfaVarMatched = NULL;
 	winstate->all_first = true;
 	winstate->partition_spooled = false;
 	winstate->more_partitions = false;
@@ -2853,6 +3116,42 @@ ExecInitWindowAgg(WindowAgg *node, EState *estate, int eflags)
 	return winstate;
 }
 
+/*
+ * ExecRPRNavGetSlot
+ *
+ * Fetch tuple at given position for RPR navigation opcodes.
+ * Returns nav_slot with the tuple loaded, or nav_null_slot if out of range.
+ */
+TupleTableSlot *
+ExecRPRNavGetSlot(WindowAggState *winstate, int64 pos)
+{
+	WindowObject winobj = winstate->nav_winobj;
+	TupleTableSlot *slot = winstate->nav_slot;
+
+	if (pos < 0)
+		return winstate->nav_null_slot;
+
+	/*
+	 * If nav_slot already holds this position, return it without re-fetching.
+	 * This is critical when multiple PREV/NEXT calls in the same expression
+	 * navigate to the same row, because re-fetching would free the slot's
+	 * tuple memory and invalidate any pass-by-ref Datum pointers from earlier
+	 * navigation results.
+	 */
+	if (winstate->nav_slot_pos == pos)
+		return slot;
+
+	if (!window_gettupleslot(winobj, pos, slot))
+	{
+		winstate->nav_slot_pos = -1;
+		return winstate->nav_null_slot;
+	}
+
+	winstate->nav_slot_pos = pos;
+	return slot;
+}
+
+
 /* -----------------
  * ExecEndWindowAgg
  * -----------------
@@ -2910,6 +3209,8 @@ ExecReScanWindowAgg(WindowAggState *node)
 	ExecClearTuple(node->agg_row_slot);
 	ExecClearTuple(node->temp_slot_1);
 	ExecClearTuple(node->temp_slot_2);
+	if (node->nav_slot)
+		ExecClearTuple(node->nav_slot);
 	if (node->framehead_slot)
 		ExecClearTuple(node->framehead_slot);
 	if (node->frametail_slot)
@@ -3270,7 +3571,8 @@ window_gettupleslot(WindowObject winobj, int64 pos, TupleTableSlot *slot)
 		return false;
 
 	if (pos < winobj->markpos)
-		elog(ERROR, "cannot fetch row before WindowObject's mark position");
+		elog(ERROR, "cannot fetch row: " INT64_FORMAT " before WindowObject's mark position: " INT64_FORMAT,
+			 pos, winobj->markpos);
 
 	oldcontext = MemoryContextSwitchTo(winstate->ss.ps.ps_ExprContext->ecxt_per_query_memory);
 
@@ -3387,6 +3689,7 @@ ignorenulls_getfuncarginframe(WindowObject winobj, int argno,
 	int			notnull_offset;
 	int			notnull_relpos;
 	int			forward;
+	int64		num_reduced_frame;
 
 	Assert(WindowObjectIsValid(winobj));
 	winstate = winobj->winstate;
@@ -3415,6 +3718,13 @@ ignorenulls_getfuncarginframe(WindowObject winobj, int argno,
 			/* rejecting relpos > 0 is easy and simplifies code below */
 			if (relpos > 0)
 				goto out_of_frame;
+
+			/*
+			 * RPR cares about frame head pos. Need to call
+			 * update_frameheadpos
+			 */
+			update_frameheadpos(winstate);
+
 			update_frametailpos(winstate);
 			abs_pos = winstate->frametailpos - 1;
 			mark_pos = 0;		/* keep compiler quiet */
@@ -3430,6 +3740,35 @@ ignorenulls_getfuncarginframe(WindowObject winobj, int argno,
 	 * Get the next nonnull value in the frame, moving forward or backward
 	 * until we find a value or reach the frame's end.
 	 */
+
+	/*
+	 * Check whether current row is in reduced frame.
+	 */
+	num_reduced_frame = row_is_in_reduced_frame(winobj, winstate->frameheadpos);
+	if (num_reduced_frame < 0)	/* unmatched or skipped row */
+		goto out_of_frame;
+	else if (num_reduced_frame > 0) /* the first row of the reduced frame */
+	{
+		/*
+		 * Early check if row could be out of reduced frame.  When RPR is
+		 * enabled, EXCLUDE clause cannot be specified and the frame is always
+		 * contiguous.  So we can safely perform the following checks. Note,
+		 * however, it is possible that a row is out of reduced frame if
+		 * there's a NULL in the middle. So we need to check it in the
+		 * following do loop.
+		 */
+		if (seektype == WINDOW_SEEK_HEAD && relpos >= num_reduced_frame)
+			goto out_of_frame;
+		if (seektype == WINDOW_SEEK_TAIL)
+		{
+			if (notnull_relpos >= num_reduced_frame)
+				goto out_of_frame;
+
+			/* not out of reduced frame. Set abspos as a starting point */
+			abs_pos = winstate->frameheadpos + num_reduced_frame - 1;
+		}
+	}
+
 	do
 	{
 		int			inframe;
@@ -3491,6 +3830,16 @@ ignorenulls_getfuncarginframe(WindowObject winobj, int argno,
 		}
 advance:
 		abs_pos += forward;
+		if (rpr_is_defined(winstate))
+		{
+			/*
+			 * Check whether we are still in the reduced frame.  (also check
+			 * if we succeeded in getting the target row).
+			 */
+			num_reduced_frame--;
+			if (num_reduced_frame <= 0 && notnull_offset <= notnull_relpos)
+				goto out_of_frame;
+		}
 	} while (notnull_offset <= notnull_relpos);
 
 	if (set_mark)
@@ -3612,6 +3961,631 @@ put_notnull_info(WindowObject winobj, int64 pos, int argno, bool isnull)
 	mbp[bpos] = mb;
 }
 
+/*
+ * eval_nav_offset_helper
+ *		Evaluate an offset expression at executor init time for trim
+ *		optimization.  Returns the offset value, or 0 for NULL/negative
+ *		(these will cause a runtime error during actual navigation, so the
+ *		trim value is irrelevant).
+ */
+static int64
+eval_nav_offset_helper(WindowAggState *winstate, Expr *offset_expr,
+					   int64 defaultOffset)
+{
+	ExprContext *econtext = winstate->ss.ps.ps_ExprContext;
+	ExprState  *estate;
+	Datum		val;
+	bool		isnull;
+	int64		offset;
+
+	if (offset_expr == NULL)
+		return defaultOffset;
+
+	estate = ExecInitExpr(offset_expr, (PlanState *) winstate);
+	val = ExecEvalExprSwitchContext(estate, econtext, &isnull);
+
+	if (isnull)
+		return 0;
+
+	offset = DatumGetInt64(val);
+	if (offset < 0)
+		return 0;
+
+	return offset;
+}
+
+typedef struct
+{
+	WindowAggState *winstate;
+	int64		maxOffset;
+	bool		overflow;		/* true if overflow detected */
+} EvalNavMaxContext;
+
+/*
+ * eval_nav_max_offset_walker
+ *		Walk expression tree evaluating backward-reach offsets at runtime.
+ *
+ * Handles simple PREV, LAST-with-offset, and compound PREV_LAST/NEXT_LAST.
+ */
+static bool
+eval_nav_max_offset_walker(Node *node, void *ctx)
+{
+	EvalNavMaxContext *context = (EvalNavMaxContext *) ctx;
+
+	if (node == NULL)
+		return false;
+
+	/* Short-circuit if overflow already detected */
+	if (context->overflow)
+		return false;
+
+	if (IsA(node, RPRNavExpr))
+	{
+		RPRNavExpr *nav = (RPRNavExpr *) node;
+		int64		reach = 0;
+
+		if (nav->kind == RPR_NAV_PREV)
+		{
+			reach = eval_nav_offset_helper(context->winstate,
+										   nav->offset_arg, 1);
+		}
+		else if (nav->kind == RPR_NAV_LAST && nav->offset_arg != NULL)
+		{
+			reach = eval_nav_offset_helper(context->winstate,
+										   nav->offset_arg, 0);
+		}
+		else if (nav->kind == RPR_NAV_PREV_LAST ||
+				 nav->kind == RPR_NAV_NEXT_LAST)
+		{
+			int64		inner = eval_nav_offset_helper(context->winstate,
+													   nav->offset_arg, 0);
+			int64		outer = eval_nav_offset_helper(context->winstate,
+													   nav->compound_offset_arg, 1);
+
+			if (nav->kind == RPR_NAV_PREV_LAST)
+			{
+				if (pg_add_s64_overflow(inner, outer, &reach))
+				{
+					context->overflow = true;
+					return false;
+				}
+			}
+			else
+				reach = (inner > outer) ? inner - outer : 0;
+		}
+
+		context->maxOffset = Max(context->maxOffset, reach);
+
+		return false;			/* don't walk into children */
+	}
+
+	return expression_tree_walker(node, eval_nav_max_offset_walker, ctx);
+}
+
+/*
+ * eval_nav_max_offset
+ *		Evaluate non-constant backward-reach offsets at executor init time.
+ *
+ * Called when the planner set navMaxOffsetKind to RPR_NAV_OFFSET_NEEDS_EVAL
+ * because some offset in PREV, LAST-with-offset, or compound PREV_LAST/
+ * NEXT_LAST contains a parameter or non-foldable expression.
+ *
+ * On overflow, sets navMaxOffsetKind to RPR_NAV_OFFSET_RETAIN_ALL so that
+ * tuplestore trim is disabled for backward navigation.
+ */
+static void
+eval_nav_max_offset(WindowAggState *winstate, List *defineClause)
+{
+	EvalNavMaxContext ctx;
+	ListCell   *lc;
+
+	ctx.winstate = winstate;
+	ctx.maxOffset = 0;
+	ctx.overflow = false;
+
+	foreach(lc, defineClause)
+	{
+		TargetEntry *te = (TargetEntry *) lfirst(lc);
+
+		eval_nav_max_offset_walker((Node *) te->expr, &ctx);
+	}
+
+	if (ctx.overflow)
+	{
+		winstate->navMaxOffsetKind = RPR_NAV_OFFSET_RETAIN_ALL;
+		winstate->navMaxOffset = 0;
+	}
+	else
+	{
+		winstate->navMaxOffsetKind = RPR_NAV_OFFSET_FIXED;
+		winstate->navMaxOffset = ctx.maxOffset;
+	}
+}
+
+typedef struct
+{
+	WindowAggState *winstate;
+	int64		minOffset;
+	bool		found;
+} EvalNavFirstContext;
+
+/*
+ * eval_nav_first_offset_walker
+ *		Walk expression tree evaluating forward-from-match_start offsets.
+ *
+ * Handles simple FIRST and compound PREV_FIRST/NEXT_FIRST.
+ */
+static bool
+eval_nav_first_offset_walker(Node *node, void *ctx)
+{
+	EvalNavFirstContext *context = (EvalNavFirstContext *) ctx;
+
+	if (node == NULL)
+		return false;
+
+	if (IsA(node, RPRNavExpr))
+	{
+		RPRNavExpr *nav = (RPRNavExpr *) node;
+		int64		combined = INT64_MAX;
+
+		if (nav->kind == RPR_NAV_FIRST)
+		{
+			context->found = true;
+			combined = eval_nav_offset_helper(context->winstate,
+											  nav->offset_arg, 0);
+		}
+		else if (nav->kind == RPR_NAV_PREV_FIRST ||
+				 nav->kind == RPR_NAV_NEXT_FIRST)
+		{
+			int64		inner = eval_nav_offset_helper(context->winstate,
+													   nav->offset_arg, 0);
+			int64		outer = eval_nav_offset_helper(context->winstate,
+													   nav->compound_offset_arg, 1);
+
+			context->found = true;
+			if (nav->kind == RPR_NAV_PREV_FIRST)
+			{
+				/*
+				 * combined = inner - outer.  Both are non-negative, so the
+				 * result >= -INT64_MAX, which cannot underflow int64.
+				 */
+				combined = inner - outer;
+			}
+			else
+			{
+				/*
+				 * NEXT_FIRST: combined = inner + outer.  This can overflow,
+				 * but the result is always >= 0, so it never updates
+				 * minOffset (which tracks the minimum).  Clamp to INT64_MAX
+				 * on overflow.
+				 */
+				if (pg_add_s64_overflow(inner, outer, &combined))
+					combined = INT64_MAX;
+			}
+		}
+
+		context->minOffset = Min(context->minOffset, combined);
+
+		return false;
+	}
+
+	return expression_tree_walker(node, eval_nav_first_offset_walker, ctx);
+}
+
+/*
+ * eval_nav_first_offset
+ *		Evaluate non-constant forward-from-match_start offsets at executor
+ *		init time.
+ *
+ * Called when the planner set navFirstOffsetKind to RPR_NAV_OFFSET_NEEDS_EVAL
+ * because some offset in FIRST or compound PREV_FIRST/NEXT_FIRST contains
+ * a parameter or non-foldable expression.
+ */
+static void
+eval_nav_first_offset(WindowAggState *winstate, List *defineClause)
+{
+	EvalNavFirstContext ctx;
+	ListCell   *lc;
+
+	ctx.winstate = winstate;
+	ctx.minOffset = INT64_MAX;
+	ctx.found = false;
+
+	foreach(lc, defineClause)
+	{
+		TargetEntry *te = (TargetEntry *) lfirst(lc);
+
+		eval_nav_first_offset_walker((Node *) te->expr, &ctx);
+	}
+
+	if (ctx.found && ctx.minOffset < INT64_MAX)
+	{
+		winstate->navFirstOffsetKind = RPR_NAV_OFFSET_FIXED;
+		winstate->navFirstOffset = ctx.minOffset;
+	}
+	else
+	{
+		winstate->navFirstOffsetKind = RPR_NAV_OFFSET_FIXED;
+		winstate->navFirstOffset = 0;
+	}
+}
+
+/*
+ * rpr_is_defined
+ * Return true if row pattern recognition is defined.
+ */
+static bool
+rpr_is_defined(WindowAggState *winstate)
+{
+	return winstate->rpPattern != NULL;
+}
+
+/*
+ * -----------------
+ * row_is_in_reduced_frame
+ * Determine whether a row is in the current row's reduced window frame
+ * according to row pattern matching
+ *
+ * The row must have already been determined to be in a full window frame
+ * and fetched into the slot.
+ *
+ * Returns:
+ * = 0, RPR is not defined.
+ * >0, if the row is the first in the reduced frame. Return the number of rows
+ * in the reduced frame.
+ * -1, if the row is an unmatched row
+ * -2, if the row is in the reduced frame but needed to be skipped because of
+ * AFTER MATCH SKIP PAST LAST ROW
+ * -----------------
+ */
+static int64
+row_is_in_reduced_frame(WindowObject winobj, int64 pos)
+{
+	WindowAggState *winstate = winobj->winstate;
+	int			state;
+	int64		rtn;
+
+	if (!rpr_is_defined(winstate))
+	{
+		/*
+		 * RPR is not defined. Assume that we are always in the reduced window
+		 * frame.
+		 */
+		rtn = 0;
+		return rtn;
+	}
+
+	state = get_reduced_frame_status(winstate, pos);
+
+	if (state == RF_NOT_DETERMINED)
+	{
+		update_frameheadpos(winstate);
+		update_reduced_frame(winobj, pos);
+	}
+
+	state = get_reduced_frame_status(winstate, pos);
+
+	switch (state)
+	{
+		case RF_FRAME_HEAD:
+			rtn = winstate->rpr_match_length;
+			break;
+
+		case RF_SKIPPED:
+			rtn = -2;
+			break;
+
+		case RF_UNMATCHED:
+		case RF_EMPTY_MATCH:
+			rtn = -1;
+			break;
+
+		default:
+			elog(ERROR, "unrecognized state: %d at: " INT64_FORMAT,
+				 state, pos);
+			break;
+	}
+
+	return rtn;
+}
+
+/*
+ * clear_reduced_frame
+ * Clear reduced frame status
+ */
+static void
+clear_reduced_frame(WindowAggState *winstate)
+{
+	winstate->rpr_match_valid = false;
+	winstate->rpr_match_matched = false;
+	winstate->rpr_match_start = -1;
+	winstate->rpr_match_length = 0;
+}
+
+/*
+ * get_reduced_frame_status
+ *		Look up a position against the current match.
+ *
+ * Returns one of the RF_* constants:
+ *   RF_NOT_DETERMINED  pos has not been processed yet
+ *   RF_FRAME_HEAD      pos is the start of the current match
+ *   RF_SKIPPED         pos is inside the current match but not the start
+ *   RF_UNMATCHED       pos is processed but not part of any match
+ */
+static int
+get_reduced_frame_status(WindowAggState *winstate, int64 pos)
+{
+	int64		start = winstate->rpr_match_start;
+	int64		length = winstate->rpr_match_length;
+
+	if (!winstate->rpr_match_valid)
+		return RF_NOT_DETERMINED;
+
+	/* Empty match: covers only the start position */
+	if (pos == start && winstate->rpr_match_matched && length == 0)
+		return RF_EMPTY_MATCH;
+
+	/* Outside the result range */
+	if (pos < start || pos >= start + length)
+		return RF_NOT_DETERMINED;
+
+	if (!winstate->rpr_match_matched)
+		return RF_UNMATCHED;
+
+	if (pos == start)
+		return RF_FRAME_HEAD;
+
+	return RF_SKIPPED;
+}
+
+/*
+ * update_reduced_frame
+ *		Update reduced frame info using multi-context NFA pattern matching.
+ *
+ * Maintains multiple NFA contexts simultaneously, one for each potential
+ * match start position. This allows sharing row evaluations across contexts,
+ * avoiding redundant DEFINE clause evaluations when rewinding for SKIP TO
+ * NEXT ROW mode.
+ *
+ * Key optimizations:
+ * - Row evaluations (expensive DEFINE clauses) happen only once per row
+ * - All active contexts share the same evaluation results
+ * - Contexts persist across calls, enabling O(n) DEFINE evaluations
+ */
+static void
+update_reduced_frame(WindowObject winobj, int64 pos)
+{
+	WindowAggState *winstate = winobj->winstate;
+	RPRNFAContext *targetCtx;
+	int64		currentPos;
+	int64		startPos;
+	int			frameOptions = winstate->frameOptions;
+	bool		hasLimitedFrame;
+	int64		frameOffset = 0;
+	int64		matchLen;
+
+	/*
+	 * Check if we have a limited frame (ROWS ... N FOLLOWING). Each context
+	 * needs its own frame end based on matchStartRow + offset.
+	 */
+	hasLimitedFrame = (frameOptions & FRAMEOPTION_ROWS) &&
+		!(frameOptions & FRAMEOPTION_END_UNBOUNDED_FOLLOWING);
+	if (hasLimitedFrame && winstate->endOffsetValue != 0)
+		frameOffset = DatumGetInt64(winstate->endOffsetValue);
+
+	/*
+	 * Case 1: pos is before any existing context's start position. This means
+	 * the position was already processed and determined unmatched. Head is
+	 * the oldest context (lowest matchStartRow) since contexts are added at
+	 * tail with increasing positions.
+	 */
+	if (winstate->nfaContext != NULL &&
+		pos < winstate->nfaContext->matchStartRow)
+	{
+		/* already processed, unmatched */
+		winstate->rpr_match_valid = true;
+		winstate->rpr_match_matched = false;
+		winstate->rpr_match_start = pos;
+		winstate->rpr_match_length = 1;
+		return;
+	}
+
+	/*
+	 * Case 2: Find existing context for this pos, or create new one.
+	 */
+	targetCtx = ExecRPRGetHeadContext(winstate, pos);
+	if (targetCtx == NULL)
+	{
+		/*
+		 * No context exists. If pos is already processed, it means this row
+		 * was already determined to be unmatched or skipped - no need to
+		 * reprocess.
+		 */
+		if (pos <= winstate->nfaLastProcessedRow)
+		{
+			/* already processed, unmatched */
+			winstate->rpr_match_valid = true;
+			winstate->rpr_match_matched = false;
+			winstate->rpr_match_start = pos;
+			winstate->rpr_match_length = 1;
+			return;
+		}
+		/* Not yet processed - create new context and start fresh */
+		targetCtx = ExecRPRStartContext(winstate, pos);
+	}
+	else if (targetCtx->states == NULL)
+	{
+		/* Context already completed - skip to result registration */
+		goto register_result;
+	}
+
+	/*
+	 * Determine where to start processing. Usually nfaLastProcessedRow+1 >=
+	 * pos since contexts are created at currentPos+1 during processing.
+	 * However, pos can exceed this when rows are skipped (e.g., unmatched
+	 * rows don't update nfaLastProcessedRow).
+	 */
+	startPos = Max(pos, winstate->nfaLastProcessedRow + 1);
+
+	/*
+	 * Process rows until target context completes or we hit boundaries. Each
+	 * row evaluation is shared across all active contexts.
+	 */
+	for (currentPos = startPos; targetCtx->states != NULL; currentPos++)
+	{
+		bool		rowExists;
+
+		/*
+		 * Evaluate variables for this row - done only once, shared by all
+		 * contexts.
+		 *
+		 * Set nav_match_start to the head context's matchStartRow for
+		 * FIRST/LAST navigation.  Match_start-dependent variables (FIRST,
+		 * LAST-with-offset) are re-evaluated per-context in ExecRPRProcessRow
+		 * when matchStartRow differs.
+		 */
+		winstate->nav_match_start = targetCtx->matchStartRow;
+		rowExists = nfa_evaluate_row(winobj, currentPos, winstate->nfaVarMatched);
+
+		/* No more rows in partition? Finalize all contexts */
+		if (!rowExists)
+		{
+			ExecRPRFinalizeAllContexts(winstate, currentPos - 1);
+			/* Clean up dead contexts from finalization */
+			ExecRPRCleanupDeadContexts(winstate, targetCtx);
+			break;
+		}
+
+		/* Update last processed row */
+		winstate->nfaLastProcessedRow = currentPos;
+
+		/*--------------------------
+		 * Process all contexts for this row:
+		 *   1. Match all (convergence)
+		 *   2. Absorb redundant
+		 *   3. Advance all (divergence)
+		 */
+		ExecRPRProcessRow(winstate, currentPos, hasLimitedFrame, frameOffset);
+
+		/*
+		 * Create a new context for the next potential start position. This
+		 * enables overlapping match detection for SKIP TO NEXT ROW.
+		 */
+		ExecRPRStartContext(winstate, currentPos + 1);
+
+		/*
+		 * Clean up dead contexts (failed with no active states and no match).
+		 * This removes contexts that failed during processing and counts them
+		 * appropriately as pruned or mismatched.
+		 */
+		ExecRPRCleanupDeadContexts(winstate, targetCtx);
+	}
+
+register_result:
+	Assert(pos == targetCtx->matchStartRow);
+
+	/*
+	 * Record match result.
+	 */
+	winstate->rpr_match_valid = true;
+	winstate->rpr_match_start = targetCtx->matchStartRow;
+
+	if (targetCtx->matchEndRow < targetCtx->matchStartRow)
+	{
+		matchLen = targetCtx->lastProcessedRow - targetCtx->matchStartRow + 1;
+
+		if (targetCtx->matchedState != NULL)
+		{
+			/* Empty match: FIN reached but 0 rows consumed */
+			winstate->rpr_match_matched = true;
+			winstate->rpr_match_length = 0;
+			ExecRPRRecordContextSuccess(winstate, 0);
+		}
+		else
+		{
+			/* No match */
+			winstate->rpr_match_matched = false;
+			winstate->rpr_match_length = 1;
+			ExecRPRRecordContextFailure(winstate, matchLen);
+		}
+		ExecRPRFreeContext(winstate, targetCtx);
+		return;
+	}
+
+	/* Match succeeded */
+	matchLen = targetCtx->matchEndRow - targetCtx->matchStartRow + 1;
+
+	winstate->rpr_match_matched = true;
+	winstate->rpr_match_length = matchLen;
+	ExecRPRRecordContextSuccess(winstate, matchLen);
+
+	/* Remove the matched context */
+	ExecRPRFreeContext(winstate, targetCtx);
+}
+
+/*
+ * nfa_evaluate_row
+ *
+ * Evaluate all DEFINE variables for current row.
+ * Returns true if the row exists, false if out of partition.
+ * If row exists, fills varMatched array.
+ * varMatched[i] = true if variable i matched at current row.
+ *
+ * Uses 1-slot model: only ecxt_outertuple is set to the current row.
+ * PREV/NEXT/FIRST/LAST navigation is handled by EEOP_RPR_NAV_SET/RESTORE
+ * opcodes during expression evaluation, which temporarily swap the slot.
+ */
+static bool
+nfa_evaluate_row(WindowObject winobj, int64 pos, bool *varMatched)
+{
+	WindowAggState *winstate = winobj->winstate;
+	ExprContext *econtext = winstate->ss.ps.ps_ExprContext;
+	int			numDefineVars = list_length(winstate->defineVariableList);
+	ListCell   *lc;
+	int			varIdx = 0;
+	TupleTableSlot *slot;
+	int64		saved_pos;
+
+	/* Fetch current row into temp_slot_1 */
+	slot = winstate->temp_slot_1;
+	if (!window_gettupleslot(winobj, pos, slot))
+		return false;			/* No row exists */
+
+	/* Set up 1-slot context: only ecxt_outertuple */
+	econtext->ecxt_outertuple = slot;
+
+	/*
+	 * Save and set currentpos so that EEOP_RPR_NAV_SET opcodes can calculate
+	 * target positions (currentpos +/- offset).
+	 */
+	saved_pos = winstate->currentpos;
+	winstate->currentpos = pos;
+
+	/* Invalidate nav_slot cache so PREV/NEXT re-fetch for new row */
+	winstate->nav_slot_pos = -1;
+
+	foreach(lc, winstate->defineClauseList)
+	{
+		ExprState  *exprState = (ExprState *) lfirst(lc);
+		Datum		result;
+		bool		isnull;
+
+		/* Evaluate DEFINE expression */
+		result = ExecEvalExpr(exprState, econtext, &isnull);
+
+		varMatched[varIdx] = (!isnull && DatumGetBool(result));
+
+		varIdx++;
+		if (varIdx >= numDefineVars)
+			break;
+	}
+
+	winstate->currentpos = saved_pos;
+
+	return true;				/* Row exists */
+}
+
+
 /***********************************************************************
  * API exposed to window functions
  ***********************************************************************/
@@ -3972,8 +4946,6 @@ WinGetFuncArgInFrame(WindowObject winobj, int argno,
 	WindowAggState *winstate;
 	ExprContext *econtext;
 	TupleTableSlot *slot;
-	int64		abs_pos;
-	int64		mark_pos;
 
 	Assert(WindowObjectIsValid(winobj));
 	winstate = winobj->winstate;
@@ -3984,6 +4956,48 @@ WinGetFuncArgInFrame(WindowObject winobj, int argno,
 		return ignorenulls_getfuncarginframe(winobj, argno, relpos, seektype,
 											 set_mark, isnull, isout);
 
+	if (WinGetSlotInFrame(winobj, slot,
+						  relpos, seektype, set_mark,
+						  isnull, isout) == 0)
+	{
+		econtext->ecxt_outertuple = slot;
+		return ExecEvalExpr((ExprState *) list_nth(winobj->argstates, argno),
+							econtext, isnull);
+	}
+
+	if (isout)
+		*isout = true;
+	*isnull = true;
+	return (Datum) 0;
+}
+
+/*
+ * WinGetSlotInFrame
+ * slot: TupleTableSlot to store the result
+ * relpos: signed rowcount offset from the seek position
+ * seektype: WINDOW_SEEK_HEAD or WINDOW_SEEK_TAIL
+ * set_mark: If the row is found/in frame and set_mark is true, the mark is
+ *		moved to the row as a side-effect.
+ * isnull: output argument, receives isnull status of result
+ * isout: output argument, set to indicate whether target row position
+ *		is out of frame (can pass NULL if caller doesn't care about this)
+ *
+ * Returns 0 if we successfully got the slot, or nonzero if out of frame.
+ * (isout is also set in the latter case.)
+ */
+static int
+WinGetSlotInFrame(WindowObject winobj, TupleTableSlot *slot,
+				  int relpos, int seektype, bool set_mark,
+				  bool *isnull, bool *isout)
+{
+	WindowAggState *winstate;
+	int64		abs_pos;
+	int64		mark_pos;
+	int64		num_reduced_frame;
+
+	Assert(WindowObjectIsValid(winobj));
+	winstate = winobj->winstate;
+
 	switch (seektype)
 	{
 		case WINDOW_SEEK_CURRENT:
@@ -4050,11 +5064,25 @@ WinGetFuncArgInFrame(WindowObject winobj, int argno,
 						 winstate->frameOptions);
 					break;
 			}
+			num_reduced_frame = row_is_in_reduced_frame(winobj,
+														winstate->frameheadpos);
+			if (num_reduced_frame < 0)
+				goto out_of_frame;
+			else if (num_reduced_frame > 0)
+				if (relpos >= num_reduced_frame)
+					goto out_of_frame;
 			break;
 		case WINDOW_SEEK_TAIL:
 			/* rejecting relpos > 0 is easy and simplifies code below */
 			if (relpos > 0)
 				goto out_of_frame;
+
+			/*
+			 * RPR cares about frame head pos. Need to call
+			 * update_frameheadpos
+			 */
+			update_frameheadpos(winstate);
+
 			update_frametailpos(winstate);
 			abs_pos = winstate->frametailpos - 1 + relpos;
 
@@ -4121,6 +5149,18 @@ WinGetFuncArgInFrame(WindowObject winobj, int argno,
 					mark_pos = 0;	/* keep compiler quiet */
 					break;
 			}
+
+			num_reduced_frame = row_is_in_reduced_frame(winobj,
+														winstate->frameheadpos);
+			if (num_reduced_frame < 0)
+				goto out_of_frame;
+			else if (num_reduced_frame > 0)
+			{
+				if (-relpos >= num_reduced_frame)
+					goto out_of_frame;
+				abs_pos = winstate->frameheadpos + relpos +
+					num_reduced_frame - 1;
+			}
 			break;
 		default:
 			elog(ERROR, "unrecognized window seek type: %d", seektype);
@@ -4138,16 +5178,24 @@ WinGetFuncArgInFrame(WindowObject winobj, int argno,
 	if (isout)
 		*isout = false;
 	if (set_mark)
+	{
+		/*
+		 * If RPR is enabled and seek type is WINDOW_SEEK_TAIL, we set the
+		 * mark position unconditionally to frameheadpos. In this case the
+		 * frame always starts at CURRENT_ROW and never goes back, thus
+		 * setting the mark at the position is safe.
+		 */
+		if (winstate->rpPattern != NULL && seektype == WINDOW_SEEK_TAIL)
+			mark_pos = winstate->frameheadpos;
 		WinSetMarkPosition(winobj, mark_pos);
-	econtext->ecxt_outertuple = slot;
-	return ExecEvalExpr((ExprState *) list_nth(winobj->argstates, argno),
-						econtext, isnull);
+	}
+	return 0;
 
 out_of_frame:
 	if (isout)
 		*isout = true;
 	*isnull = true;
-	return (Datum) 0;
+	return -1;
 }
 
 /*
diff --git a/src/backend/jit/llvm/llvmjit_expr.c b/src/backend/jit/llvm/llvmjit_expr.c
index 0e160b8502c..e42c5d65bb6 100644
--- a/src/backend/jit/llvm/llvmjit_expr.c
+++ b/src/backend/jit/llvm/llvmjit_expr.c
@@ -129,6 +129,9 @@ llvm_compile_expr(ExprState *state)
 	LLVMValueRef v_aggvalues;
 	LLVMValueRef v_aggnulls;
 
+	/* RPR navigation: when true, EEOP_OUTER_VAR reloads from econtext */
+	bool		has_rpr_nav;
+
 	instr_time	starttime;
 	instr_time	deform_starttime;
 	instr_time	endtime;
@@ -298,6 +301,36 @@ llvm_compile_expr(ExprState *state)
 								   FIELDNO_EXPRCONTEXT_AGGNULLS,
 								   "v.econtext.aggnulls");
 
+	/*
+	 * RPR navigation opcodes (PREV/NEXT) swap ecxt_outertuple to a different
+	 * row mid-expression.  The JIT code loads v_outervalues and v_outernulls
+	 * once in the entry block and reuses them for all EEOP_OUTER_VAR steps.
+	 * After a slot swap, these cached pointers become stale because the new
+	 * slot has its own tts_values/tts_isnull arrays.
+	 *
+	 * When RPR navigation opcodes are present, EEOP_OUTER_VAR reloads the
+	 * slot pointer from econtext->ecxt_outertuple on every access instead of
+	 * using the cached entry-block values.  This avoids the SSA/PHI
+	 * complexity while keeping the rest of the expression JIT-compiled.
+	 * Expressions without RPR navigation use the cached values as before.
+	 */
+	has_rpr_nav = false;
+	if (parent && IsA(parent, WindowAggState) &&
+		((WindowAgg *) parent->plan)->rpPattern != NULL)
+	{
+		for (int opno = 0; opno < state->steps_len; opno++)
+		{
+			ExprEvalOp	opcode = ExecEvalStepOp(state, &state->steps[opno]);
+
+			if (opcode == EEOP_RPR_NAV_SET ||
+				opcode == EEOP_RPR_NAV_RESTORE)
+			{
+				has_rpr_nav = true;
+				break;
+			}
+		}
+	}
+
 	/* allocate blocks for each op upfront, so we can do jumps easily */
 	opblocks = palloc_array(LLVMBasicBlockRef, state->steps_len);
 	for (int opno = 0; opno < state->steps_len; opno++)
@@ -460,8 +493,37 @@ llvm_compile_expr(ExprState *state)
 					}
 					else if (opcode == EEOP_OUTER_VAR)
 					{
-						v_values = v_outervalues;
-						v_nulls = v_outernulls;
+						if (has_rpr_nav)
+						{
+							/*
+							 * RPR navigation swaps ecxt_outertuple
+							 * mid-expression.  Reload slot pointer from
+							 * econtext on every access so we read from the
+							 * current (possibly swapped) slot.
+							 */
+							LLVMValueRef v_tmpslot;
+
+							v_tmpslot = l_load_struct_gep(b,
+														  StructExprContext,
+														  v_econtext,
+														  FIELDNO_EXPRCONTEXT_OUTERTUPLE,
+														  "v_outerslot_reload");
+							v_values = l_load_struct_gep(b,
+														 StructTupleTableSlot,
+														 v_tmpslot,
+														 FIELDNO_TUPLETABLESLOT_VALUES,
+														 "v_outervalues_reload");
+							v_nulls = l_load_struct_gep(b,
+														StructTupleTableSlot,
+														v_tmpslot,
+														FIELDNO_TUPLETABLESLOT_ISNULL,
+														"v_outernulls_reload");
+						}
+						else
+						{
+							v_values = v_outervalues;
+							v_nulls = v_outernulls;
+						}
 					}
 					else if (opcode == EEOP_SCAN_VAR)
 					{
@@ -2434,6 +2496,18 @@ llvm_compile_expr(ExprState *state)
 				LLVMBuildBr(b, opblocks[opno + 1]);
 				break;
 
+			case EEOP_RPR_NAV_SET:
+				build_EvalXFunc(b, mod, "ExecEvalRPRNavSet",
+								v_state, op, v_econtext);
+				LLVMBuildBr(b, opblocks[opno + 1]);
+				break;
+
+			case EEOP_RPR_NAV_RESTORE:
+				build_EvalXFunc(b, mod, "ExecEvalRPRNavRestore",
+								v_state, op, v_econtext);
+				LLVMBuildBr(b, opblocks[opno + 1]);
+				break;
+
 			case EEOP_AGG_STRICT_DESERIALIZE:
 			case EEOP_AGG_DESERIALIZE:
 				{
diff --git a/src/backend/jit/llvm/llvmjit_types.c b/src/backend/jit/llvm/llvmjit_types.c
index c8a1f841293..e78b31d775f 100644
--- a/src/backend/jit/llvm/llvmjit_types.c
+++ b/src/backend/jit/llvm/llvmjit_types.c
@@ -168,6 +168,8 @@ void	   *referenced_functions[] =
 	ExecEvalScalarArrayOp,
 	ExecEvalHashedScalarArrayOp,
 	ExecEvalSubPlan,
+	ExecEvalRPRNavSet,
+	ExecEvalRPRNavRestore,
 	ExecEvalSysVar,
 	ExecEvalWholeRowVar,
 	ExecEvalXmlExpr,
diff --git a/src/backend/utils/adt/windowfuncs.c b/src/backend/utils/adt/windowfuncs.c
index 78b7f05aba2..46e7a03a666 100644
--- a/src/backend/utils/adt/windowfuncs.c
+++ b/src/backend/utils/adt/windowfuncs.c
@@ -41,7 +41,6 @@ static bool rank_up(WindowObject winobj);
 static Datum leadlag_common(FunctionCallInfo fcinfo,
 							bool forward, bool withoffset, bool withdefault);
 
-
 /*
  * utility routine for *_rank functions.
  */
@@ -724,3 +723,121 @@ window_nth_value(PG_FUNCTION_ARGS)
 
 	PG_RETURN_DATUM(result);
 }
+
+/*
+ * prev
+ * Catalog placeholder for RPR's PREV navigation operator.
+ *
+ * The parser transforms prev() calls inside DEFINE into RPRNavExpr nodes,
+ * so this function is never reached during normal RPR execution.  It exists
+ * only so that the parser can resolve the function name from pg_proc.
+ * Calls outside DEFINE are rejected by parse_func.c (EXPR_KIND_RPR_DEFINE
+ * check).  The error below is a defensive measure in case that check is
+ * bypassed (e.g., direct C-level function invocation).
+ */
+Datum
+window_prev(PG_FUNCTION_ARGS)
+{
+	ereport(ERROR,
+			(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
+			 errmsg("cannot use PREV() outside a DEFINE clause")));
+	PG_RETURN_NULL();			/* not reached */
+}
+
+/*
+ * next
+ * Catalog placeholder for RPR's NEXT navigation operator.
+ * See window_prev() for details.
+ */
+Datum
+window_next(PG_FUNCTION_ARGS)
+{
+	ereport(ERROR,
+			(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
+			 errmsg("cannot use NEXT() outside a DEFINE clause")));
+	PG_RETURN_NULL();			/* not reached */
+}
+
+/*
+ * prev(value, offset)
+ * Catalog placeholder for RPR's PREV navigation operator with offset.
+ * See window_prev() for details.
+ */
+Datum
+window_prev_offset(PG_FUNCTION_ARGS)
+{
+	ereport(ERROR,
+			(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
+			 errmsg("cannot use PREV() outside a DEFINE clause")));
+	PG_RETURN_NULL();			/* not reached */
+}
+
+/*
+ * next(value, offset)
+ * Catalog placeholder for RPR's NEXT navigation operator with offset.
+ * See window_prev() for details.
+ */
+Datum
+window_next_offset(PG_FUNCTION_ARGS)
+{
+	ereport(ERROR,
+			(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
+			 errmsg("cannot use NEXT() outside a DEFINE clause")));
+	PG_RETURN_NULL();			/* not reached */
+}
+
+/*
+ * first
+ * Catalog placeholder for RPR's FIRST navigation operator.
+ * See window_prev() for details.
+ */
+Datum
+window_first(PG_FUNCTION_ARGS)
+{
+	ereport(ERROR,
+			(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
+			 errmsg("cannot use FIRST() outside a DEFINE clause")));
+	PG_RETURN_NULL();			/* not reached */
+}
+
+/*
+ * last
+ * Catalog placeholder for RPR's LAST navigation operator.
+ * See window_prev() for details.
+ */
+Datum
+window_last(PG_FUNCTION_ARGS)
+{
+	ereport(ERROR,
+			(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
+			 errmsg("cannot use LAST() outside a DEFINE clause")));
+	PG_RETURN_NULL();			/* not reached */
+}
+
+/*
+ * first(value, offset)
+ * Catalog placeholder for RPR's FIRST navigation operator with offset.
+ * See window_prev() for details.
+ */
+Datum
+window_first_offset(PG_FUNCTION_ARGS)
+{
+	ereport(ERROR,
+			(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
+			 errmsg("cannot use FIRST() outside a DEFINE clause")));
+	PG_RETURN_NULL();			/* not reached */
+}
+
+/*
+ * last(value, offset)
+ * Catalog placeholder for RPR's LAST navigation operator with offset.
+ * See window_prev() for details.
+ */
+Datum
+window_last_offset(PG_FUNCTION_ARGS)
+{
+	ereport(ERROR,
+			(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
+			 errmsg("cannot use LAST() outside a DEFINE clause")));
+	PG_RETURN_NULL();			/* not reached */
+}
diff --git a/src/include/catalog/pg_proc.dat b/src/include/catalog/pg_proc.dat
index fa9ae79082b..02b668fc4c5 100644
--- a/src/include/catalog/pg_proc.dat
+++ b/src/include/catalog/pg_proc.dat
@@ -11017,6 +11017,30 @@
 { oid => '3114', descr => 'fetch the Nth row value',
   proname => 'nth_value', prokind => 'w', prorettype => 'anyelement',
   proargtypes => 'anyelement int4', prosrc => 'window_nth_value' },
+{ oid => '8126', descr => 'fetch the preceding row value',
+  proname => 'prev', provolatile => 's', prorettype => 'anyelement',
+  proargtypes => 'anyelement', prosrc => 'window_prev' },
+{ oid => '8128', descr => 'fetch the Nth preceding row value',
+  proname => 'prev', provolatile => 's', proisstrict => 'f', prorettype => 'anyelement',
+  proargtypes => 'anyelement int8', prosrc => 'window_prev_offset' },
+{ oid => '8127', descr => 'fetch the following row value',
+  proname => 'next', provolatile => 's', prorettype => 'anyelement',
+  proargtypes => 'anyelement', prosrc => 'window_next' },
+{ oid => '8129', descr => 'fetch the Nth following row value',
+  proname => 'next', provolatile => 's', proisstrict => 'f', prorettype => 'anyelement',
+  proargtypes => 'anyelement int8', prosrc => 'window_next_offset' },
+{ oid => '8130', descr => 'fetch the first row value within match',
+  proname => 'first', provolatile => 's', prorettype => 'anyelement',
+  proargtypes => 'anyelement', prosrc => 'window_first' },
+{ oid => '8132', descr => 'fetch the Nth row value within match',
+  proname => 'first', provolatile => 's', proisstrict => 'f', prorettype => 'anyelement',
+  proargtypes => 'anyelement int8', prosrc => 'window_first_offset' },
+{ oid => '8131', descr => 'fetch the last row value within match',
+  proname => 'last', provolatile => 's', prorettype => 'anyelement',
+  proargtypes => 'anyelement', prosrc => 'window_last' },
+{ oid => '8133', descr => 'fetch the Nth-from-last row value within match',
+  proname => 'last', provolatile => 's', proisstrict => 'f', prorettype => 'anyelement',
+  proargtypes => 'anyelement int8', prosrc => 'window_last_offset' },
 
 # functions for range types
 { oid => '3832', descr => 'I/O',
diff --git a/src/include/executor/execExpr.h b/src/include/executor/execExpr.h
index c61b3d624d5..db66ebe313c 100644
--- a/src/include/executor/execExpr.h
+++ b/src/include/executor/execExpr.h
@@ -274,6 +274,10 @@ typedef enum ExprEvalOp
 	EEOP_MERGE_SUPPORT_FUNC,
 	EEOP_SUBPLAN,
 
+	/* row pattern navigation (RPR PREV/NEXT) */
+	EEOP_RPR_NAV_SET,
+	EEOP_RPR_NAV_RESTORE,
+
 	/* aggregation related nodes */
 	EEOP_AGG_STRICT_DESERIALIZE,
 	EEOP_AGG_DESERIALIZE,
@@ -695,6 +699,18 @@ typedef struct ExprEvalStep
 			SubPlanState *sstate;
 		}			subplan;
 
+		/* for EEOP_RPR_NAV_SET / EEOP_RPR_NAV_RESTORE */
+		struct
+		{
+			WindowAggState *winstate;
+			RPRNavKind	kind;	/* navigation kind (simple or compound) */
+			Datum	   *offset_value;	/* offset value(s), or NULL */
+			bool	   *offset_isnull;	/* offset null flag(s) */
+			/* For compound nav: offset_value[0] = inner, [1] = outer */
+			int16		resulttyplen;	/* RESTORE: result type length */
+			bool		resulttypbyval; /* RESTORE: result pass-by-value? */
+		}			rpr_nav;
+
 		/* for EEOP_AGG_*DESERIALIZE */
 		struct
 		{
@@ -902,6 +918,10 @@ extern void ExecEvalMergeSupportFunc(ExprState *state, ExprEvalStep *op,
 									 ExprContext *econtext);
 extern void ExecEvalSubPlan(ExprState *state, ExprEvalStep *op,
 							ExprContext *econtext);
+extern void ExecEvalRPRNavSet(ExprState *state, ExprEvalStep *op,
+							  ExprContext *econtext);
+extern void ExecEvalRPRNavRestore(ExprState *state, ExprEvalStep *op,
+								  ExprContext *econtext);
 extern void ExecEvalWholeRowVar(ExprState *state, ExprEvalStep *op,
 								ExprContext *econtext);
 extern void ExecEvalSysVar(ExprState *state, ExprEvalStep *op,
diff --git a/src/include/executor/execRPR.h b/src/include/executor/execRPR.h
new file mode 100644
index 00000000000..7b2b0febb76
--- /dev/null
+++ b/src/include/executor/execRPR.h
@@ -0,0 +1,40 @@
+/*-------------------------------------------------------------------------
+ *
+ * execRPR.h
+ *	  prototypes for execRPR.c (NFA-based Row Pattern Recognition engine)
+ *
+ *
+ * Portions Copyright (c) 1996-2026, PostgreSQL Global Development Group
+ * Portions Copyright (c) 1994, Regents of the University of California
+ *
+ * src/include/executor/execRPR.h
+ *
+ *-------------------------------------------------------------------------
+ */
+#ifndef EXECRPR_H
+#define EXECRPR_H
+
+#include "nodes/execnodes.h"
+#include "windowapi.h"
+
+/* NFA context management */
+extern RPRNFAContext *ExecRPRStartContext(WindowAggState *winstate,
+										  int64 startPos);
+extern RPRNFAContext *ExecRPRGetHeadContext(WindowAggState *winstate,
+											int64 pos);
+extern void ExecRPRFreeContext(WindowAggState *winstate, RPRNFAContext *ctx);
+
+/* NFA processing */
+extern void ExecRPRProcessRow(WindowAggState *winstate, int64 currentPos,
+							  bool hasLimitedFrame, int64 frameOffset);
+extern void ExecRPRCleanupDeadContexts(WindowAggState *winstate,
+									   RPRNFAContext *excludeCtx);
+extern void ExecRPRFinalizeAllContexts(WindowAggState *winstate, int64 lastPos);
+
+/* NFA statistics */
+extern void ExecRPRRecordContextSuccess(WindowAggState *winstate,
+										int64 matchLen);
+extern void ExecRPRRecordContextFailure(WindowAggState *winstate,
+										int64 failedLen);
+
+#endif							/* EXECRPR_H */
diff --git a/src/include/executor/nodeWindowAgg.h b/src/include/executor/nodeWindowAgg.h
index ada4a1c458c..f6f6645131c 100644
--- a/src/include/executor/nodeWindowAgg.h
+++ b/src/include/executor/nodeWindowAgg.h
@@ -20,4 +20,7 @@ extern WindowAggState *ExecInitWindowAgg(WindowAgg *node, EState *estate, int ef
 extern void ExecEndWindowAgg(WindowAggState *node);
 extern void ExecReScanWindowAgg(WindowAggState *node);
 
+/* RPR navigation support for expression evaluation opcodes */
+extern TupleTableSlot *ExecRPRNavGetSlot(WindowAggState *winstate, int64 pos);
+
 #endif							/* NODEWINDOWAGG_H */
diff --git a/src/include/nodes/execnodes.h b/src/include/nodes/execnodes.h
index 13359180d25..0cb01baa949 100644
--- a/src/include/nodes/execnodes.h
+++ b/src/include/nodes/execnodes.h
@@ -38,6 +38,9 @@
 #include "nodes/plannodes.h"
 #include "partitioning/partdefs.h"
 #include "storage/buf.h"
+#include "storage/condition_variable.h"
+#include "utils/hsearch.h"
+#include "utils/queryenvironment.h"
 #include "utils/reltrigger.h"
 #include "utils/typcache.h"
 
@@ -2524,6 +2527,71 @@ typedef enum WindowAggStatus
 									 * tuples during spool */
 } WindowAggStatus;
 
+/* RPR reduced frame states returned by get_reduced_frame_status() */
+#define	RF_NOT_DETERMINED	0	/* not yet processed */
+#define	RF_FRAME_HEAD		1	/* start row of a match */
+#define	RF_SKIPPED			2	/* interior row of a match */
+#define	RF_UNMATCHED		3	/* no match at this row */
+#define	RF_EMPTY_MATCH		4	/* empty match (0 rows); treated as unmatched */
+
+/*
+ * RPRNFAState - single NFA state for pattern matching
+ *
+ * counts[] tracks repetition counts at each nesting depth.
+ *
+ * isAbsorbable tracks if state is in absorbable region (ABSORBABLE_BRANCH).
+ * Monotonic property: once false, stays false (can't re-enter region).
+ */
+typedef struct RPRNFAState
+{
+	struct RPRNFAState *next;	/* next state in linked list */
+	int16		elemIdx;		/* current pattern element index */
+	bool		isAbsorbable;	/* true if state is in absorbable region */
+	int32		counts[FLEXIBLE_ARRAY_MEMBER];	/* repetition counts by depth */
+} RPRNFAState;
+
+/*
+ * RPRNFAContext - context for NFA pattern matching execution
+ *
+ * Two-flag absorption design:
+ *   hasAbsorbableState: can this context absorb others? (>=1 absorbable state)
+ *     - Monotonic: true->false only, cannot recover once false
+ *     - Used to skip absorption attempts once all absorbable states are gone
+ *   allStatesAbsorbable: can this context be absorbed? (ALL states absorbable)
+ *     - Dynamic: can change false->true (when non-absorbable states die)
+ *     - Used to determine if this context is eligible for absorption
+ */
+typedef struct RPRNFAContext
+{
+	struct RPRNFAContext *next; /* next context in linked list */
+	struct RPRNFAContext *prev; /* previous context (for reverse traversal) */
+	RPRNFAState *states;		/* active states (linked list) */
+
+	int64		matchStartRow;	/* row where match started */
+	int64		matchEndRow;	/* row where match ended (-1 = no match) */
+	int64		lastProcessedRow;	/* last row processed (for fail depth) */
+	RPRNFAState *matchedState;	/* FIN state for greedy fallback (cloned) */
+
+	/* Two-flag absorption optimization */
+	bool		hasAbsorbableState; /* can absorb others (>=1 absorbable
+									 * state) */
+	bool		allStatesAbsorbable;	/* can be absorbed (ALL states
+										 * absorbable) */
+} RPRNFAContext;
+
+/*
+ * NFALengthStats
+ *
+ * Statistics for length measurements (min/max/total) used for computing
+ * average lengths in EXPLAIN ANALYZE output.
+ */
+typedef struct NFALengthStats
+{
+	int64		min;			/* minimum length */
+	int64		max;			/* maximum length */
+	int64		total;			/* total length (for computing average) */
+} NFALengthStats;
+
 typedef struct WindowAggState
 {
 	ScanState	ss;				/* its first field is NodeTag */
@@ -2583,6 +2651,49 @@ typedef struct WindowAggState
 	int64		groupheadpos;	/* current row's peer group head position */
 	int64		grouptailpos;	/* " " " " tail position (group end+1) */
 
+	/* these fields are used in Row pattern recognition: */
+	RPSkipTo	rpSkipTo;		/* Row Pattern Skip To type */
+	struct RPRPattern *rpPattern;	/* compiled pattern for NFA execution */
+	List	   *defineVariableList; /* list of row pattern definition
+									 * variables (list of String) */
+	List	   *defineClauseList;	/* expression for row pattern definition
+									 * search conditions ExprState list */
+	RPRNFAContext *nfaContext;	/* active matching contexts (head) */
+	RPRNFAContext *nfaContextTail;	/* tail of active contexts (for reverse
+									 * traversal) */
+	RPRNFAContext *nfaContextFree;	/* recycled NFA context nodes */
+	RPRNFAState *nfaStateFree;	/* recycled NFA state nodes */
+	Size		nfaStateSize;	/* pre-calculated RPRNFAState size */
+	bool	   *nfaVarMatched;	/* per-row cache: varMatched[varId] for varId
+								 * < numDefines */
+	Bitmapset  *defineMatchStartDependent;	/* DEFINE vars needing per-context
+											 * evaluation
+											 * (match_start-dependent) */
+	bitmapword *nfaVisitedElems;	/* elemIdx visited bitmap for cycle
+									 * detection */
+	int			nfaVisitedNWords;	/* number of bitmapwords in
+									 * nfaVisitedElems */
+	int64		nfaLastProcessedRow;	/* last row processed by NFA (-1 =
+										 * none) */
+
+	/* NFA statistics for EXPLAIN ANALYZE */
+	int64		nfaStatesActive;	/* current active states (internal) */
+	int64		nfaStatesMax;	/* peak active states */
+	int64		nfaStatesTotalCreated;	/* total states allocated */
+	int64		nfaStatesMerged;	/* states merged (deduplicated) */
+	int64		nfaContextsActive;	/* current active contexts (internal) */
+	int64		nfaContextsMax; /* peak active contexts */
+	int64		nfaContextsTotalCreated;	/* total contexts allocated */
+	int64		nfaContextsAbsorbed;	/* contexts absorbed (optimization) */
+	int64		nfaContextsSkipped; /* contexts skipped (SKIP PAST LAST ROW) */
+	int64		nfaContextsPruned;	/* contexts pruned on first row */
+	int64		nfaMatchesSucceeded;	/* successful pattern matches */
+	int64		nfaMatchesFailed;	/* failed pattern matches */
+	NFALengthStats nfaMatchLen; /* successful match length stats */
+	NFALengthStats nfaFailLen;	/* mismatch length stats */
+	NFALengthStats nfaAbsorbedLen;	/* absorbed context length stats */
+	NFALengthStats nfaSkippedLen;	/* skipped context length stats */
+
 	MemoryContext partcontext;	/* context for partition-lifespan data */
 	MemoryContext aggcontext;	/* shared context for aggregate working data */
 	MemoryContext curaggcontext;	/* current aggregate's working data */
@@ -2610,6 +2721,25 @@ typedef struct WindowAggState
 	TupleTableSlot *agg_row_slot;
 	TupleTableSlot *temp_slot_1;
 	TupleTableSlot *temp_slot_2;
+
+	/* RPR navigation */
+	RPRNavOffsetKind navMaxOffsetKind;	/* status of navMaxOffset */
+	int64		navMaxOffset;	/* max backward nav offset (when FIXED) */
+	bool		hasFirstNav;	/* FIRST() present in DEFINE */
+	RPRNavOffsetKind navFirstOffsetKind;	/* status of navFirstOffset */
+	int64		navFirstOffset; /* min FIRST() offset (when FIXED) */
+	struct WindowObjectData *nav_winobj;	/* winobj for RPR nav fetch */
+	int64		nav_slot_pos;	/* position cached in nav_slot, or -1 */
+	TupleTableSlot *nav_slot;	/* slot for PREV/NEXT/FIRST/LAST target row */
+	TupleTableSlot *nav_saved_outertuple;	/* saved slot during nav swap */
+	TupleTableSlot *nav_null_slot;	/* all NULL slot */
+	int64		nav_match_start;	/* match_start for FIRST/LAST nav */
+
+	/* RPR current match result */
+	bool		rpr_match_valid;	/* true if a match result is set */
+	bool		rpr_match_matched;	/* true if the result was a match */
+	int64		rpr_match_start;	/* start position of the match result */
+	int64		rpr_match_length;	/* number of rows matched (0 = empty) */
 } WindowAggState;
 
 /* ----------------
-- 
2.43.0