Add chat-ui-htmx module: streaming chat page on Spring AI 2.0.2 with htmx and its SSE extension (vendored), MessageChatMemoryAdvisor keyed by the HTTP session, a Stop button, and the traps measured: EventSource strips one leading space per data line, model text must be HTML-escaped, a cancelled stream left the upstream connection open in 7 of 8 runs until an OkHttp interceptor calls Call.cancel (8 of 8 closed after), and the SDK push source ignores back-pressure (limitRate changes only the demand; a bounded buffer works). The model is a stub OpenAI-compatible SSE server written for the module; no real provider was run.

Co-Authored-By: Claude Sonnet 5.5 <[email protected]>
Claude-Session: https://claude.ai/code/session_017ic4b6Kg9RWSTJzyz6KP9X

chat-ui-htmx/scripts/backpressure.sh

@@ -0,0 +1,25 @@
+#!/usr/bin/env bash
+# A slow client against a fast model: curl reads 300 KB/s of a 600 x 32 KB stream (19.6 MB) while the
+# stub writes with no delay. Samples how far the stub got and how much heap the app holds.
+# Usage: backpressure.sh <none|limitRate|slow> [extra app args]
+source "$(dirname "$0")/lib.sh"; set +e
+MODE=${1:-none}; shift
+stop_all; start_stub 0; start_app --chat.backpressure="$MODE" --chat.slow-millis=5 "$@"
+echo "== chat.backpressure=$MODE  (stub delay 0 ms, 600 chunks x 32 KiB, client reads 64 KiB every 0.2 s)"
+echo 0 > "$RUN/recv.bytes"
+python3 scripts/slow_client.py "http://localhost:$APP_PORT/chat/stream?message=big%20600&turn=1" "$RUN/recv.bytes" &
+CURL=$!
+T0=$(date +%s.%N)
+for pause in 1 1 2 3 4; do
+  sleep "$pause"
+  st=$(curl -s "localhost:$STUB_PORT/stub/stats" | sed 's/requests=[0-9]* //')
+  recv=$(cat "$RUN/recv.bytes")
+  now=$(date +%s.%N)
+  jcmd "$(cat $RUN/app.pid)" GC.run >/dev/null 2>&1
+  heap=$(jcmd "$(cat $RUN/app.pid)" GC.heap_info 2>/dev/null | grep -o "used [0-9]*K" | head -1)
+  printf "t=%4.1fs  client got %5d KiB   stub: %s   app heap after GC: %s\n" "$(echo "$now - $T0" | bc)" $((recv/1024)) "$st" "$heap"
+done
+wait $CURL 2>/dev/null
+echo "-- app log (demand + outcome):"
+grep "demand\|CANCELLED\|complete\|error" $RUN/app.log | sed "s/turn 1 //" | sort | uniq -c | sort -rn | head -6
+stop_all