Skip to content

Qwen3.8-Flash-Next on any consumer hardware: one-click install for Windows / Linux. Strata inference engine, OpenAI/Anthropic API on localhost, optional image input.

C++

Fix #4209: a saved preset's lower-case model tag reaches setup as its key

吴佳浩Alben2f9c6fc35cdc
diff --git a/strata_cli/launch/runner.py b/strata_cli/launch/runner.py
index cd323caa6..a79c78ef3 100644
--- a/strata_cli/launch/runner.py
+++ b/strata_cli/launch/runner.py
@@ -880,6 +880,10 @@ class Runner:
def setup_args(self) -> list[str]:
cat, s = self.catalog, self.s
fam, model = s.family, s.model
+ try: # a saved preset or choices file keeps the lower-case tag ("iq3_xxs", #4209), while setup's
+ fam, model = cat.resolve(str(model), None if fam is None else str(fam)) # --model takes the table's own key
+ except Exception: # noqa: BLE001 a name the list does not hold, or a list that cannot be read here, is setup's
+ pass # to report: passing the values on unchanged keeps the launch behaving as it did before
args = ["--yes", "--no-start", "--family", str(fam), "--model", str(model)]
ctx = s.context
if ctx:
diff --git a/tests/terminal/test_launch.py b/tests/terminal/test_launch.py
index a4b8c6c6c..e5965feb7 100644
--- a/tests/terminal/test_launch.py
+++ b/tests/terminal/test_launch.py
@@ -346,6 +346,38 @@ class SetupArgs(unittest.TestCase):
s = S.LaunchSettings(family="qwen", model="IQ3_XXS", context=32768, kv="q4_0", **kw)
return Runner(s, catalog=cat).setup_args()
+ def args_for(self, **kw):
+ values = dict(family="qwen", model="IQ3_XXS", context=32768, kv="q4_0", vision="off")
+ values.update(kw)
+ with Home():
+ return Runner(S.LaunchSettings(**values), catalog=catalog(HW_12)).setup_args()
+
+ def test_a_lower_case_tag_from_a_saved_preset_reaches_setup_as_its_key(self):
+ """#4209: a preset keeps "model": "iq3_xxs", and setup's --model choices are the table's own keys
+ ("IQ3_XXS"). Starting such a preset with any changed choice reruns setup, which then refused it."""
+ args = self.args_for(model="iq3_xxs")
+ self.assertEqual(args[args.index("--model") + 1], "IQ3_XXS")
+ self.assertEqual(args[args.index("--family") + 1], "qwen")
+ for fam, model, want in (("coder", "iq1_m", "IQ1_M"), ("gyro", "gyro-s", "Gyro-S"), ("qwen", "IQ3_S", "IQ3_S")):
+ args = self.args_for(family=fam, model=model)
+ self.assertEqual(args[args.index("--family") + 1], fam)
+ self.assertEqual(args[args.index("--model") + 1], want)
+
+ def test_an_unknown_model_is_still_setup_to_report(self):
+ args = self.args_for(model="not-a-model")
+ self.assertEqual(args[args.index("--model") + 1], "not-a-model")
+
+ def test_a_model_list_that_cannot_be_read_leaves_the_values_alone(self):
+ """The resolver is an improvement, never a new way for the launch to fail: when it cannot answer, the
+ preset's own values go to setup as they did before #4209."""
+ with Home():
+ cat = catalog(HW_12)
+ r = Runner(S.LaunchSettings(family="qwen", model="iq3_xxs", context=32768, vision="off"), catalog=cat)
+ with mock.patch.object(cat, "resolve", side_effect=RuntimeError("the model list cannot be read")):
+ args = r.setup_args()
+ self.assertEqual(args[args.index("--model") + 1], "iq3_xxs")
+ self.assertEqual(args[args.index("--family") + 1], "qwen")
+
def test_pictures_default_on_the_cpu_reader_under_24_gb_and_the_gpu_reader_from_24_gb_on_every_make(self):
from tests.terminal.launch_support import HW_12 as NV
card = lambda name, gb, vendor: {**NV, "gpus": [{"index": 0, "name": name, "vram_gb": gb, "usable": True, "vendor": vendor}]} # noqa: E731