From 2b8afb8c10367a166f2f727f244b73a13ab96ed6 Mon Sep 17 00:00:00 2001 From: Alex Date: Thu, 16 Jul 2026 09:45:18 +0100 Subject: [PATCH] fix: cleanup comments --- application/core/settings.py | 10 ---------- tests/llm/test_openai_responses.py | 2 -- 2 files changed, 12 deletions(-) diff --git a/application/core/settings.py b/application/core/settings.py index 080ba53c..a10f90df 100644 --- a/application/core/settings.py +++ b/application/core/settings.py @@ -248,12 +248,6 @@ class Settings(BaseSettings): # reasoning is carried across the in-turn tool loop via encrypted # reasoning items instead. OPENAI_RESPONSES_STORE: bool = False - - # Reasoning-summary verbosity requested on the Responses path when a - # model has a reasoning_effort configured ("auto" | "concise" | - # "detailed"). Some models/gateways reject "detailed", so the default - # stays "auto"; stream liveness is handled by the wire-level SSE - # keepalive (see SSE_KEEPALIVE_SECONDS) rather than summary verbosity. OPENAI_REASONING_SUMMARY: str = "auto" # OpenAI-compatible clients can identify a logical chat with session @@ -294,10 +288,6 @@ class Settings(BaseSettings): # Used by gunicorn_worker.BoundedDrainUvicornWorker. GRACEFUL_SHUTDOWN_TIMEOUT_SECONDS: int = 30 WSGI_THREADPOOL_WORKERS: int = 96 - # SSE keepalive comment cadence. Must sit under Cloudflare's 100s idle - # close and iOS Safari's ~60s — 15s gives generous headroom. ge=1 - # because a zero/negative interval would busy-loop or crash the - # keepalive wrapper's queue timeout. SSE_KEEPALIVE_SECONDS: int = Field(default=15, ge=1) # Cap on simultaneous SSE connections per user. Each connection holds # one WSGI thread (32 per gunicorn worker) and one Redis pub/sub diff --git a/tests/llm/test_openai_responses.py b/tests/llm/test_openai_responses.py index a8fbadd9..967393a8 100644 --- a/tests/llm/test_openai_responses.py +++ b/tests/llm/test_openai_responses.py @@ -389,8 +389,6 @@ def test_build_responses_params_stateless(monkeypatch): assert params["model"] == "gpt-5.5" assert params["stream"] is True assert params["max_output_tokens"] == 256 - # Summary verbosity comes from settings.OPENAI_REASONING_SUMMARY - # ("auto" unless overridden). assert params["reasoning"] == {"effort": "high", "summary": "auto"} assert params["store"] is False assert params["include"] == ["reasoning.encrypted_content"]