From 5283a6e47e7048054c70aee58528323b082f80f0 Mon Sep 17 00:00:00 2001
From: "copilot-swe-agent[bot]" <198982749+Copilot@users.noreply.github.com>
Date: Wed, 15 Oct 2025 19:29:32 +0000
Subject: [PATCH 1/2] Initial plan


From 160361e42c54d8fc1cb4e326def669f470ce0e81 Mon Sep 17 00:00:00 2001
From: "copilot-swe-agent[bot]" <198982749+Copilot@users.noreply.github.com>
Date: Wed, 15 Oct 2025 19:37:50 +0000
Subject: [PATCH 2/2] Update memory.rst to correctly document max_queue in
 threading implementation

Co-authored-by: aaugustin <788910+aaugustin@users.noreply.github.com>
---
 docs/topics/memory.rst | 14 +++++++++++---
 1 file changed, 11 insertions(+), 3 deletions(-)

diff --git a/docs/topics/memory.rst b/docs/topics/memory.rst
index 61b1113e2..d39277540 100644
--- a/docs/topics/memory.rst
+++ b/docs/topics/memory.rst
@@ -152,6 +152,14 @@ and :func:`~asyncio.server.serve` varies across implementations.
   use up to 4 bytes of memory per character, the queue may use up to ``4 *
   max_size * max_queue`` bytes of memory. By default, this is 128 MiB.
 
-* In the :mod:`threading` implementation, there is no queue of incoming
-  messages. The ``max_queue`` argument doesn't exist. The connection keeps at
-  most one message in memory at a time.
+* In the :mod:`threading` implementation, ``max_queue`` has the same semantics
+  as in the new :mod:`asyncio` implementation. It is the high-water mark of a
+  queue of incoming frames. It defaults to 16 frames. If the queue grows larger,
+  the connection stops reading from the network until the application consumes
+  messages and the queue goes below the low-water mark. This creates backpressure
+  on the TCP connection.
+
+  Each item in the queue is a frame. A frame can be a message or a message
+  fragment. Either way, it must be smaller than ``max_size``, the maximum size
+  of a message. The queue may use up to ``max_size * max_queue`` bytes of
+  memory. By default, this is 16 MiB.