Each send is numbered, inside the signed bytes

novox/hq 04-ISSUES/107. The controller already held a per-node lock while
it composed and recorded each send; the order existed and was thrown away
at the wire. Each send now takes the next number for its node, one
higher than the last, under that hold and before the body exists — so
the number is inside what the mesh signs, and a replayed older
declaration cannot borrow a newer one's.

Zero is not sent. A host reads absence as "no order claimed", which is
the shape of every declaration before this, so nothing that worked
before changes for a machine sent nothing since numbering existed.

One subtlety, and it is the one that would have read every machine as
behind for ever: the mesh decides a machine is behind by comparing the
digest of what it WOULD send against what it DID send, and a number
changes the bytes. The read-only comparison composes with the number the
machine was LAST sent, not a fresh one, so it is byte for byte what was
sent when nothing else changed.

Hosts went first and every machine runs one that understands the field.

internal/inventory/migrations/0046-each-send-is-numbered.sql

@@ -0,0 +1,11 @@
+-- The order of what a machine was sent, so a host can tell an older declaration from a newer.
+--
+-- novox/hq 04-ISSUES/107. A declaration's only identity was the digest of its bytes. The host could
+-- say "this is not the last one" and could not say "this is older", so a backlog drained out of
+-- order applied a declaration the mesh had already superseded. The controller already holds a
+-- per-node lock while it composes and records each send — the order existed and was thrown away at
+-- the wire.
+--
+-- One counter per node, taken under that hold, one higher per send. Null is a machine sent nothing
+-- since this existed, which is not the same as a machine sent nothing.
+alter table node add column sequence bigint;

internal/inventory/nodes.go

@@ -1005,3 +1005,34 @@ func (i *Inventory) RecordHostVersion(ctx context.Context, id, version string) e
 		`update node set host_version = $2, last_seen = now() where id = $1`, id, version)
 	return err
 }
+
+// NextSequence takes the next number for a declaration to this node, one higher than the last it
+// was sent (novox/hq 04-ISSUES/107).
+//
+// **One statement, so two composers cannot take the same number.** The caller holds the node while it
+// composes and sends, so in practice there is one; the increment is atomic anyway, because a rule
+// that is true only while a lock is held somewhere else is a rule nobody can see from here.
+func (i *Inventory) NextSequence(ctx context.Context, id string) (int64, error) {
+	var n int64
+	err := i.store.Pool().QueryRow(ctx,
+		`update node set sequence = coalesce(sequence, 0) + 1 where id = $1 returning sequence`, id).Scan(&n)
+	if err != nil {
+		return 0, fmt.Errorf("taking the next sequence for %s: %w", id, err)
+	}
+	return n, nil
+}
+
+// Sequence is the number of the last declaration this node was sent, and zero for one sent nothing
+// since sends were numbered. Read, not taken: what the mesh WOULD send is composed with this, so it
+// is byte for byte what it DID send when nothing else changed — a comparison that took a fresh number
+// would read every machine as behind for ever (novox/hq 04-ISSUES/107).
+func (i *Inventory) Sequence(ctx context.Context, id string) (int64, error) {
+	var n *int64
+	if err := i.store.Pool().QueryRow(ctx, `select sequence from node where id = $1`, id).Scan(&n); err != nil {
+		return 0, fmt.Errorf("reading the sequence of %s: %w", id, err)
+	}
+	if n == nil {
+		return 0, nil
+	}
+	return *n, nil
+}