chore: cleanup garbage code (#29770)

test/manual/chunked_prefill/test_scripted_regression.py

@@ -112,7 +112,7 @@ class TestRegressionBasic(ScriptedTestCase):
         "dbdcdde245 mamba_pool_idx cleanup-skip is mamba-architecture-specific. "
         "The shared test fixture does not configure a mamba model; running this "
         "regression against the default transformer model would not exercise the "
-        "mamba NO_TOKEN cleanup path, so the body would be a pure smoke test "
+        "mamba NO_TOKEN cleanup path, so the body would be a check "
         "with no real protection. Re-enable when a mamba fixture is wired in."
     )
     def test_mamba_chunked_resume_no_token(self):

test/manual/dsv4/_common.py

@@ -8,7 +8,7 @@ and contains one ``CustomTestCase`` subclass per recipe
 Each subclass launches the server with the cookbook's exact flags and
 runs two sgl-eval evaluations (https://github.com/sgl-project/sgl-eval):
 - ``test_smoke_gsm8k`` — short, cheap GSM8K pass to verify the server
-  can produce coherent math answers at all (smoke gate).
+  can produce coherent math answers at all (sanity gate).
 - ``test_aime25`` — full AIME25 accuracy run (heavy; 16 repeats default).
 
 Cookbook reference:
@@ -19,7 +19,7 @@ These are MANUAL tests (not CI). ``sgl-eval`` must be on PATH.
 Per-variant defaults (set on the Flash/Pro intermediate base classes):
     Flash recipes -> AIME25 score threshold 0.93
     Pro   recipes -> AIME25 score threshold 0.95
-GSM8K smoke threshold (0.93) is shared across Flash and Pro.
+GSM8K sanity threshold (0.93) is shared across Flash and Pro.
 
 AIME25 knobs (env vars):
     DSV4_AIME25_NUM_REPEATS       (default 16    -> --n-repeats)
@@ -30,7 +30,7 @@ AIME25 knobs (env vars):
     DSV4_AIME25_SCORE_METRIC      (default "score"; sgl-eval JSON key under "aggregate")
     DSV4_AIME25_SCORE_THRESHOLD   (default 0; >0 overrides per-variant default)
 
-GSM8K smoke knobs (env vars):
+GSM8K sanity knobs (env vars):
     DSV4_GSM8K_NUM_EXAMPLES       (default 50    -> --num-examples)
     DSV4_GSM8K_N_REPEATS          (default 1     -> --n-repeats)
     DSV4_GSM8K_TEMPERATURE        (default 0.6   -> --temperature)

test/registered/8-gpu-models/test_ling_2_6_flash.py

@@ -20,7 +20,7 @@ class TestLing26Flash(GSM8KMixin, DefaultServerBase):
 
     # Native 128K context (no YaRN) — avoids the
     # SGLANG_ALLOW_OVERWRITE_LONGER_CONTEXT_LEN env-var dance and keeps the
-    # smoke test focused on the dispatcher / hybrid-attention path.
+    # coverage focused on the dispatcher / hybrid-attention path.
     other_args = [
         "--tp-size",
         "4",

test/registered/amd/disaggregation/test_mori_transfer_engine_e2e.py

@@ -39,7 +39,7 @@ class MoriTransferEngineBase(PDDisaggregationServerBase):
                 raise unittest.SkipTest("torch.cuda is not available.")
             if torch.cuda.device_count() < cls.required_gpus:
                 raise unittest.SkipTest(
-                    f"MORI PD smoke test requires >= {cls.required_gpus} visible GPUs."
+                    f"MORI PD check requires >= {cls.required_gpus} visible GPUs."
                 )
         except Exception as e:
             raise unittest.SkipTest(f"torch is not available/usable: {e}")

test/registered/amd/disaggregation/test_nixl_transfer_engine_e2e.py

@@ -46,7 +46,7 @@ class NixlTransferEngineBase(PDDisaggregationServerBase):
                 raise unittest.SkipTest("torch.cuda is not available.")
             if torch.cuda.device_count() < cls.required_gpus:
                 raise unittest.SkipTest(
-                    f"NIXL PD smoke test requires >= {cls.required_gpus} visible GPUs."
+                    f"NIXL PD check requires >= {cls.required_gpus} visible GPUs."
                 )
         except unittest.SkipTest:
             raise

test/registered/core/test_basic_sanity.py

@@ -1,4 +1,4 @@
-"""Stage-a basic sanity: small-but-broad server smoke that downstream
+"""Stage-a basic sanity: small-but-broad server coverage that downstream
 stages depend on. Multiple sanity-kit mixins driving one shared server,
 covering protocol, decode correctness, scheduler stress, occupancy, and
 hellaswag accuracy."""

test/registered/model_loading/test_load_weights_from_remote_instance.py

@@ -302,7 +302,7 @@ def test_load_weights_from_remote_instance(
         try:
             key, value = param_queue.get(timeout=5)
             results[key] = value
-        except Exception as e:
+        except Exception:
             if all(not p.is_alive() for p in context.processes):
                 break
 

test/registered/model_loading/test_load_weights_from_remote_instance_npu.py

@@ -310,7 +310,7 @@ def test_load_weights_from_remote_instance(
         try:
             key, value = param_queue.get(timeout=5)
             results[key] = value
-        except Exception as e:
+        except Exception:
             if all(not p.is_alive() for p in context.processes):
                 break
 

test/registered/musa/test_llm_server_smoke_musa.py

@@ -18,7 +18,7 @@ _REQUEST_TIMEOUT = 60
     "MUSA device not available",
 )
 class TestMusaDeepSeekV2LiteChatServerSmoke(DefaultServerBase):
-    """MUSA LLM server smoke test: launch, health check, and non-empty generation."""
+    """MUSA LLM server sanity check: launch, health check, and non-empty generation."""
 
     model = os.getenv("SGLANG_MUSA_LLM_MODEL", "deepseek-ai/DeepSeek-V2-Lite-Chat")
     served_model_name = "deepseek-v2-lite-chat"

test/registered/radix_cache/unified_radix_tree/test_unified_radix_cache_kl_dsv4.py

@@ -112,7 +112,7 @@ class TestUnifiedDeepSeekV4FlashHiCache(UnifiedRadixTreeTestMixin, CustomTestCas
 class TestUnifiedDeepSeekV4FlashHiCachePageFirstDirect(
     TestUnifiedDeepSeekV4FlashHiCache
 ):
-    """DeepSeek V4 Flash HiCache layout smoke: page_first_direct + direct."""
+    """DeepSeek V4 Flash HiCache layout check: page_first_direct + direct."""
 
     hicache_io_backend = "kernel"
     hicache_mem_layout = "layer_first"

test/registered/rl/test_update_weights_from_distributed.py

@@ -396,7 +396,8 @@ def init_process_sgl(
             return response.json()
 
         with ThreadPoolExecutor(32) as executor:
-            futures = [executor.submit(run_decode, 1000) for _ in range(32)]
+            for _ in range(32):
+                executor.submit(run_decode, 1000)
             time.sleep(2)
 
     # The last parameter is lm_head.weight, which is tied
@@ -573,7 +574,7 @@ def test_update_weights_from_distributed(
         try:
             key, value = param_queue.get(timeout=5)
             results[key] = value
-        except Exception as e:
+        except Exception:
             if all(not p.is_alive() for p in context.processes):
                 break
 

test/registered/scripted_runtime/test_scripted_runtime_core.py

@@ -736,9 +736,7 @@ class TestScriptedRuntimeCore(ScriptedTestCase):
 
     @staticmethod
     def _script_empty_return(t: ScriptedContext):
-        if False:
-            yield
-        return
+        yield from ()
 
     def test_failing_script_surfaces_and_session_survives(self):
         with self.assertRaises(AssertionError) as ctx:
@@ -753,7 +751,7 @@ class TestScriptedRuntimeCore(ScriptedTestCase):
 
     @staticmethod
     def _script_minimal_ok(t: ScriptedContext):
-        r = t.start_req(prompt_len=_SHORT_PROMPT_LEN, max_new_tokens=2)
+        t.start_req(prompt_len=_SHORT_PROMPT_LEN, max_new_tokens=2)
         yield
         yield
 

test/registered/unit/entrypoints/openai/test_serving_responses.py

@@ -280,7 +280,7 @@ class FullResponseUsageTestCase(unittest.TestCase):
         metadata = RequestResponseMetadata(request_id=request.request_id)
 
         async def empty_generator():
-            if False:
+            for _ in ():
                 yield None
 
         response = asyncio.run(

test/registered/unit/hardware_backend/mlx/test_quantization.py

@@ -3,7 +3,7 @@
 Covers:
   - mlx_q4 / mlx_q8 quantize fp16 weights to QuantizedLinear in-place
   - active-memory drops after quantization
-  - smoke /generate still works post-quantize
+  - /generate still works post-quantize
   - pre-quantized HF repos still load (regression guard for mlx_lm passthrough)
   - mlx_q4 flag on an already-quantized model is a no-op (skip + log)
 

test/registered/unit/models/test_locate_anything.py

@@ -265,7 +265,7 @@ def _image_item(feature, grid_hws):
 
 
 class TestGetImageFeatureWiring(CustomTestCase):
-    """Forward-shape smoke test for get_image_feature.
+    """Forward-shape coverage for get_image_feature.
 
     Guards the production path (pixel concat -> vision tower -> projector) and
     the precomputed-embedding passthrough so a future change to the wiring or