diff --git a/mypyc/test-data/driver/driver.py b/mypyc/test-data/driver/driver.py
index c9d179224a30..1ec1c48dfb75 100644
--- a/mypyc/test-data/driver/driver.py
+++ b/mypyc/test-data/driver/driver.py
@@ -11,10 +11,12 @@
 import native
 
 failures = []
+tests_run = 0
 
 for name in dir(native):
     if name.startswith('test_'):
         test_func = getattr(native, name)
+        tests_run += 1
         try:
             test_func()
         except Exception as e:
@@ -46,3 +48,5 @@ def extract_line(tb):
     print(f'<< {failures[-1][0]} >>')
     sys.stdout.flush()
     raise failures[-1][1][1]
+
+assert tests_run > 0, 'Default test driver did not find any functions prefixed "test_" to run.'
diff --git a/mypyc/test-data/fixtures/ir.py b/mypyc/test-data/fixtures/ir.py
index 532cbbc06177..3776a3dcc79a 100644
--- a/mypyc/test-data/fixtures/ir.py
+++ b/mypyc/test-data/fixtures/ir.py
@@ -350,7 +350,12 @@ class GeneratorExit(BaseException): pass
 
 def any(i: Iterable[_T]) -> bool: pass
 def all(i: Iterable[_T]) -> bool: pass
-def sum(i: Iterable[_T]) -> int: pass
+@overload
+def sum(i: Iterable[bool]) -> int: pass
+@overload
+def sum(i: Iterable[_T]) -> _T: pass
+@overload
+def sum(i: Iterable[_T], start: _T) -> _T: pass
 def reversed(object: Sequence[_T]) -> Iterator[_T]: ...
 def id(o: object) -> int: pass
 # This type is obviously wrong but the test stubs don't have Sized anymore
diff --git a/mypyc/test-data/run-bools.test b/mypyc/test-data/run-bools.test
index a0b8ea31ebc0..3409665bfb37 100644
--- a/mypyc/test-data/run-bools.test
+++ b/mypyc/test-data/run-bools.test
@@ -223,7 +223,8 @@ def test_mixed_comparisons_i64() -> None:
             assert gt_mixed_i64(n, x) == (n > int(x))
 
 [case testBoolMixInt]
-y = False
-print((y or 0) and True)
+def test_mix() -> None:
+    y = False
+    print((y or 0) and True)
 [out]
 0
diff --git a/mypyc/test-data/run-bytes.test b/mypyc/test-data/run-bytes.test
index fa63c46a6798..bee6b6fe9f76 100644
--- a/mypyc/test-data/run-bytes.test
+++ b/mypyc/test-data/run-bytes.test
@@ -277,7 +277,6 @@ class bytes_subclass(bytes):
         return b'spook'
 
 [case testBytesFormatting]
-[typing fixtures/typing-full.pyi]
 from testutil import assertRaises
 
 # https://www.python.org/dev/peps/pep-0461/
@@ -314,6 +313,7 @@ def test_bytes_formatting_2() -> None:
     aa = b'\xe4\xbd\xa0\xe5\xa5\xbd%b' % b'\xe4\xbd\xa0\xe5\xa5\xbd'
     assert aa == b'\xe4\xbd\xa0\xe5\xa5\xbd\xe4\xbd\xa0\xe5\xa5\xbd'
     assert aa.decode() == '你好你好'
+[typing fixtures/typing-full.pyi]
 
 
 class A:
diff --git a/mypyc/test-data/run-classes.test b/mypyc/test-data/run-classes.test
index fd486980ef16..54f5343bc7bb 100644
--- a/mypyc/test-data/run-classes.test
+++ b/mypyc/test-data/run-classes.test
@@ -1278,9 +1278,10 @@ class Bar(Foo):
     def f(self, *args: int, **kwargs: int) -> None:
         print("stuff", args, kwargs)
 
-z: Foo = Bar()
-z.f(1, z=50)
-z.f()
+def test_override() -> None:
+    z: Foo = Bar()
+    z.f(1, z=50)
+    z.f()
 
 [out]
 stuff (1,) {'z': 50}
@@ -1300,18 +1301,19 @@ class Foo:
 def baz_f(self: Any, *args: int, **kwargs: int) -> None:
     print("Baz", args, kwargs)
 
-# Make an "interpreted" subtype of Foo
-type2: Any = type
-Bar = type2('Bar', (Foo,), {})
-Baz = type2('Baz', (Foo,), {'f': baz_f})
+def test_override() -> None:
+    # Make an "interpreted" subtype of Foo
+    type2: Any = type
+    Bar = type2('Bar', (Foo,), {})
+    Baz = type2('Baz', (Foo,), {'f': baz_f})
 
-y: Foo = Bar()
-y.f(1, z=2)
-y.f()
+    y: Foo = Bar()
+    y.f(1, z=2)
+    y.f()
 
-z: Foo = Baz()
-z.f(1, z=2)
-z.f()
+    z: Foo = Baz()
+    z.f(1, z=2)
+    z.f()
 
 [out]
 Foo 1 2
@@ -1330,9 +1332,10 @@ class Bar(Foo):
     def f(self, x: Optional[int]=None) -> None:
         print(x)
 
-z: Foo = Bar()
-z.f(1)
-z.f()
+def test_override() -> None:
+    z: Foo = Bar()
+    z.f(1)
+    z.f()
 
 [out]
 1
@@ -1349,10 +1352,11 @@ class Bar(Foo):
     def f(self, *args: int, **kwargs: int) -> None:
         print("Bar", args, kwargs)
 
-z: Foo = Bar()
-z.f(1, z=2)
-z.f(1, 2, 3)
-# z.f(x=5)  # Not tested because we (knowingly) do the wrong thing and pass it as positional
+def test_override() -> None:
+    z: Foo = Bar()
+    z.f(1, z=2)
+    z.f(1, 2, 3)
+    # z.f(x=5)  # Not tested because we (knowingly) do the wrong thing and pass it as positional
 
 [out]
 Bar (1,) {'z': 2}
@@ -1370,10 +1374,11 @@ class Bar(Foo):
     def f(self, x: int = 10, *args: int, **kwargs: int) -> None:
         print("Bar", x, args, kwargs)
 
-z: Foo = Bar()
-z.f(1, z=2)
-z.f(1, 2, 3)
-z.f()
+def test_override() -> None:
+    z: Foo = Bar()
+    z.f(1, z=2)
+    z.f(1, 2, 3)
+    z.f()
 
 [out]
 Bar 1 () {'z': 2}
@@ -1397,18 +1402,19 @@ class Foo:
 def baz_f(self, a: int=30, y: int=50) -> None:
     print("Baz", a, y)
 
-# Make an "interpreted" subtype of Foo
-type2: Any = type
-Baz = type2('Baz', (Foo,), {'f': baz_f})
+def test_override() -> None:
+    # Make an "interpreted" subtype of Foo
+    type2: Any = type
+    Baz = type2('Baz', (Foo,), {'f': baz_f})
 
-z: Foo = Baz()
-z.f()
-z.f(y=1)
-z.f(1, 2)
-# Not tested because we don't (and probably won't) match cpython here
-# from testutil import assertRaises
-# with assertRaises(TypeError):
-#     z.f(x=7)
+    z: Foo = Baz()
+    z.f()
+    z.f(y=1)
+    z.f(1, 2)
+    # Not tested because we don't (and probably won't) match cpython here
+    # from testutil import assertRaises
+    # with assertRaises(TypeError):
+    #     z.f(x=7)
 
 [out]
 Baz 30 50
@@ -2591,7 +2597,8 @@ class Base:
 class Derived(Base):
     pass
 
-assert Derived()() == 1
+def test_inherited() -> None:
+    assert Derived()() == 1
 
 [case testClassWithFinalAttribute]
 from typing import Final
diff --git a/mypyc/test-data/run-dunders-special.test b/mypyc/test-data/run-dunders-special.test
index 30c618374f88..2672434e10ef 100644
--- a/mypyc/test-data/run-dunders-special.test
+++ b/mypyc/test-data/run-dunders-special.test
@@ -6,4 +6,5 @@ class UsesNotImplemented:
     def __eq__(self, b: object) -> bool:
         return NotImplemented
 
-assert UsesNotImplemented() != object()
+def test_not_implemented() -> None:
+    assert UsesNotImplemented() != object()
diff --git a/mypyc/test-data/run-functions.test b/mypyc/test-data/run-functions.test
index 46f343fa3798..3d7f1f3cc747 100644
--- a/mypyc/test-data/run-functions.test
+++ b/mypyc/test-data/run-functions.test
@@ -1235,13 +1235,11 @@ from contextlib import contextmanager
 def f() -> Iterator[None]:
     yield
 
-def g() -> None:
+def test_special_case() -> None:
     a = ['']
     with f():
         a.pop()
 
-g()
-
 [case testUnpackKwargsCompiled]
 from typing import TypedDict
 from typing_extensions import Unpack
@@ -1253,8 +1251,9 @@ class Person(TypedDict):
 def foo(**kwargs: Unpack[Person]) -> None:
     print(kwargs["name"])
 
-# This is not really supported yet, just test that we behave reasonably.
-foo(name='Jennifer', age=38)
+def test_unpack() -> None:
+    # This is not really supported yet, just test that we behave reasonably.
+    foo(name='Jennifer', age=38)
 [typing fixtures/typing-full.pyi]
 [out]
 Jennifer
@@ -1269,8 +1268,9 @@ def foo() -> None:
     print(inner.__dict__)  # type: ignore[attr-defined]
     print(inner.x)  # type: ignore[attr-defined]
 
-if sys.version_info >= (3, 12):  # type: ignore
-    foo()
+def test_nested() -> None:
+    if sys.version_info >= (3, 12):  # type: ignore
+        foo()
 [out]
 [out version>=3.12]
 {}
@@ -1285,7 +1285,8 @@ def bar() -> None:
     functools.update_wrapper(inner, bar)  # type: ignore
     print(inner.__dict__)  # type: ignore
 
-bar()
+def test_update() -> None:
+    bar()
 [typing fixtures/typing-full.pyi]
 [out]
 {'__module__': 'native', '__name__': 'bar', '__qualname__': 'bar', '__doc__': None, '__wrapped__': <built-in function bar>}
diff --git a/mypyc/test-data/run-generators.test b/mypyc/test-data/run-generators.test
index a43aff27dd45..3b4581f849e9 100644
--- a/mypyc/test-data/run-generators.test
+++ b/mypyc/test-data/run-generators.test
@@ -617,16 +617,22 @@ else:
 from typing import Iterator
 
 class Foo:
-    flag: bool
+    flag = False
 
 class C:
-    foo: Foo
+    foo = Foo()
 
     def genf(self) -> Iterator[None]:
         self.foo.flag = True
         yield
         self.foo.flag = False
 
+def test_near_yield() -> None:
+    c = C()
+    for x in c.genf():
+        pass
+    assert c.foo.flag == False
+
 [case testGeneratorEarlyReturnWithBorrows]
 from typing import Iterator
 class Bar:
@@ -639,6 +645,12 @@ class Foo:
             return
         yield 0
 
+def test_early_return() -> None:
+    foo = Foo()
+    for x in foo.f():
+        pass
+    assert foo.bar.bar == 1
+
 [case testBorrowingInGeneratorInTupleAssignment]
 from typing import Iterator
 
diff --git a/mypyc/test-data/run-generics.test b/mypyc/test-data/run-generics.test
index bc78a3b8ab86..55e5adbbb4f9 100644
--- a/mypyc/test-data/run-generics.test
+++ b/mypyc/test-data/run-generics.test
@@ -27,22 +27,23 @@ def fn_typeddict(t: T) -> None:
     print([x for x in t.keys()])
     print({k: v for k, v in t.items()})
 
-fn_mapping({})
-print("=====")
-fn_mapping({"a": 1, "b": 2})
-print("=====")
+def test_mapping() -> None:
+    fn_mapping({})
+    print("=====")
+    fn_mapping({"a": 1, "b": 2})
+    print("=====")
 
-fn_union({"a": 1, "b": 2})
-print("=====")
-fn_union({"a": "1", "b": "2"})
-print("=====")
+    fn_union({"a": 1, "b": 2})
+    print("=====")
+    fn_union({"a": "1", "b": "2"})
+    print("=====")
 
-orig: Union[Dict[str, int], Dict[str, str]] = {"a": 1, "b": 2}
-fn_union(orig)
-print("=====")
+    orig: Union[Dict[str, int], Dict[str, str]] = {"a": 1, "b": 2}
+    fn_union(orig)
+    print("=====")
 
-td: TD = {"foo": 1}
-fn_typeddict(td)
+    td: TD = {"foo": 1}
+    fn_typeddict(td)
 [typing fixtures/typing-full.pyi]
 [out]
 \[]
@@ -96,8 +97,9 @@ def deco(func: Callable[P, int]) -> Callable[P, int]:
 def f(x: int, y: str) -> int:
     return x
 
-assert f(1, 'a') == 1
-assert f(2, y='b') == 2
+def test_usable() -> None:
+    assert f(1, 'a') == 1
+    assert f(2, y='b') == 2
 [out]
 \[1, 'a']
 {}
diff --git a/mypyc/test-data/run-imports.test b/mypyc/test-data/run-imports.test
index c5839d57820e..ce83a882e2de 100644
--- a/mypyc/test-data/run-imports.test
+++ b/mypyc/test-data/run-imports.test
@@ -212,9 +212,10 @@ import shared
 def do_import() -> None:
     import a
 
-assert shared.counter == 0
-do_import()
-assert shared.counter == 1
+def test_lazy() -> None:
+    assert shared.counter == 0
+    do_import()
+    assert shared.counter == 1
 
 [file a.py]
 import shared
@@ -224,9 +225,10 @@ shared.counter += 1
 counter = 0
 
 [case testDelayedImport]
-import a
-print("inbetween")
-import b
+def test_delayed() -> None:
+    import a
+    print("inbetween")
+    import b
 
 [file a.py]
 print("first")
@@ -240,19 +242,21 @@ inbetween
 last
 
 [case testImportErrorLineNumber]
-try:
-    import enum
-    import dataclasses, missing  # type: ignore[import]
-except ImportError as e:
-    line = e.__traceback__.tb_lineno # type: ignore[attr-defined]
-    assert line == 3, f"traceback's line number is {line}, expected 3"
+def test_error() -> None:
+    try:
+        import enum
+        import dataclasses, missing  # type: ignore[import]
+    except ImportError as e:
+        line = e.__traceback__.tb_lineno # type: ignore[attr-defined]
+        assert line == 4, f"traceback's line number is {line}, expected 4"
 
 [case testImportGroupIsolation]
 def func() -> None:
     import second
 
-import first
-func()
+def test_isolation() -> None:
+    import first
+    func()
 
 [file first.py]
 print("first")
diff --git a/mypyc/test-data/run-misc.test b/mypyc/test-data/run-misc.test
index f6a1c744cade..129946a4c330 100644
--- a/mypyc/test-data/run-misc.test
+++ b/mypyc/test-data/run-misc.test
@@ -109,10 +109,11 @@ def gen(b: bool) -> Generator[Any, None, None]:
         y = None
     yield y
 
-assert f(False) == ((1, None), (None, 1))
-assert f(True) == ((None, 1), (1, None))
-assert next(gen(False)) is None
-assert next(gen(True)) == 1
+def test_inferred() -> None:
+    assert f(False) == ((1, None), (None, 1))
+    assert f(True) == ((None, 1), (1, None))
+    assert next(gen(False)) is None
+    assert next(gen(True)) == 1
 
 [case testWith]
 from typing import Any
@@ -829,23 +830,23 @@ assert call_any_nested([[1, 1, 1], [1, 1], []]) == 1
 assert call_any_nested([[1, 1, 1], [0, 1], []]) == 0
 
 [case testSum]
-[typing fixtures/typing-full.pyi]
-from typing import Any, List
+from typing import List
 
+empty: List[int] = []
 def test_sum_of_numbers() -> None:
     assert sum(x for x in [1, 2, 3]) == 6
-    assert sum(x for x in [0.0, 1.2, 2]) == 6.2
+    assert sum(x for x in [0.0, 1.2, 2]) == 3.2
     assert sum(x for x in [1, 1j]) == 1 + 1j
 
 def test_sum_callables() -> None:
-    assert sum((lambda x: x == 0)(x) for x in []) == 0
+    assert sum((lambda x: x == 0)(x) for x in empty) == 0
     assert sum((lambda x: x == 0)(x) for x in [0]) == 1
     assert sum((lambda x: x == 0)(x) for x in [0, 0, 0]) == 3
     assert sum((lambda x: x == 0)(x) for x in [0, 1, 0]) == 2
     assert sum((lambda x: x % 2 == 0)(x) for x in range(2**10)) == 2**9
 
 def test_sum_comparisons() -> None:
-    assert sum(x == 0 for x in []) == 0
+    assert sum(x == 0 for x in empty) == 0
     assert sum(x == 0 for x in [0]) == 1
     assert sum(x == 0 for x in [0, 0, 0]) == 3
     assert sum(x == 0 for x in [0, 1, 0]) == 2
@@ -865,13 +866,14 @@ def test_sum_misc() -> None:
 def test_sum_start_given() -> None:
     a = 1
     assert sum((x == 0 for x in [0, 1]), a) == 2
-    assert sum(((lambda x: x == 0)(x) for x in []), 1) == 1
+    assert sum(((lambda x: x == 0)(x) for x in empty), 1) == 1
     assert sum(((lambda x: x == 0)(x) for x in [0]), 1) == 2
     assert sum(((lambda x: x == 0)(x) for x in [0, 0, 0]), 1) == 4
     assert sum(((lambda x: x == 0)(x) for x in [0, 1, 0]), 1) == 3
     assert sum(((lambda x: x % 2 == 0)(x) for x in range(2**10)), 1) == 2**9 + 1
     assert sum((x for x in [1, 1j]), 2j) == 1 + 3j
     assert sum((c == 'd' for c in 'abcdd'), 1) == 3
+[typing fixtures/typing-full.pyi]
 
 [case testNoneStuff]
 from typing import Optional
@@ -1090,19 +1092,20 @@ def test_complex() -> None:
 from typing import cast
 import sys
 
-A = sys.platform == 'x' and foobar
-B = sys.platform == 'x' and sys.foobar
-C = sys.platform == 'x' and f(a, -b, 'y') > [c + e, g(y=2)]
-C = sys.platform == 'x' and cast(a, b[c])
-C = sys.platform == 'x' and (lambda x: y + x)
-C = sys.platform == 'x' and (x for y in z)
-C = sys.platform == 'x' and [x for y in z]
-C = sys.platform == 'x' and {x: x for y in z}
-C = sys.platform == 'x' and {x for y in z}
-
-assert not A
-assert not B
-assert not C
+def test_unreachable() -> None:
+    A = sys.platform == 'x' and foobar
+    B = sys.platform == 'x' and sys.foobar
+    C = sys.platform == 'x' and f(a, -b, 'y') > [c + e, g(y=2)]
+    C = sys.platform == 'x' and cast(a, b[c])
+    C = sys.platform == 'x' and (lambda x: y + x)
+    C = sys.platform == 'x' and (x for y in z)
+    C = sys.platform == 'x' and [x for y in z]
+    C = sys.platform == 'x' and {x: x for y in z}
+    C = sys.platform == 'x' and {x for y in z}
+
+    assert not A
+    assert not B
+    assert not C
 
 [case testDoesntSegfaultWhenTopLevelFails]
 # make the initial import fail
@@ -1126,6 +1129,10 @@ class B(A):
     def _(arg): pass
     def _(arg): pass
 
+def test_underscore() -> None:
+    A()
+    B()
+
 [case testGlobalRedefinition_toplevel]
 # mypy: allow-redefinition
 i = 0
diff --git a/mypyc/test-data/run-python38.test b/mypyc/test-data/run-python38.test
index 7de43907cb86..cf7c7d7dea52 100644
--- a/mypyc/test-data/run-python38.test
+++ b/mypyc/test-data/run-python38.test
@@ -75,11 +75,12 @@ class Bar(Foo):
     def f(self, *args: int, **kwargs: int) -> None:
         print("stuff", args, kwargs)
 
-z: Foo = Bar()
-z.f(1, z=50)
-z.f()
-z.f(1)
-z.f(z=50)
+def test_pos_only() -> None:
+    z: Foo = Bar()
+    z.f(1, z=50)
+    z.f()
+    z.f(1)
+    z.f(z=50)
 
 [out]
 stuff (1,) {'z': 50}
diff --git a/mypyc/test-data/run-singledispatch.test b/mypyc/test-data/run-singledispatch.test
index 61e4897c96d6..a119c325984a 100644
--- a/mypyc/test-data/run-singledispatch.test
+++ b/mypyc/test-data/run-singledispatch.test
@@ -152,12 +152,14 @@ from functools import singledispatch
 def fun(arg) -> bool:
     return False
 
-try:
-    @fun.register
-    def fun_specialized(arg: None) -> bool:
-        return True
-except TypeError:
-    pass
+def test_argument() -> None:
+    try:
+        @fun.register
+        def fun_specialized(arg: None) -> bool:
+            return True
+        assert False, "expected to raise an exception"
+    except TypeError:
+        pass
 
 [case testRegisteringTheSameFunctionSeveralTimes]
 from functools import singledispatch
@@ -598,9 +600,11 @@ assert f(1) == 'default'
 def _(arg: B) -> str:
     return 'b'
 
-assert f(A()) == 'a'
-assert f(B()) == 'b'
-assert f(1) == 'default'
+# TODO: Move whole testcase to a function when mypyc#1118 is fixed.
+def test_final() -> None:
+    assert f(A()) == 'a'
+    assert f(B()) == 'b'
+    assert f(1) == 'default'
 
 
 [case testDynamicallyRegisteringFunctionFromInterpretedCode]
diff --git a/mypyc/test-data/run-strings.test b/mypyc/test-data/run-strings.test
index c726c4c70896..6551d9c352df 100644
--- a/mypyc/test-data/run-strings.test
+++ b/mypyc/test-data/run-strings.test
@@ -363,7 +363,6 @@ def test_str_min_max() -> None:
     assert max(x, z) == 'aaa'
 
 [case testStringFormattingCStyle]
-[typing fixtures/typing-full.pyi]
 from typing import Tuple
 
 var = 'mypyc'
@@ -408,6 +407,7 @@ def test_basics() -> None:
     inf_num = float('inf')
     assert '%s, %s' % (nan_num, inf_num) == 'nan, inf'
     assert '%f, %f' % (nan_num, inf_num) == 'nan, inf'
+[typing fixtures/typing-full.pyi]
 
 [case testFStrings]
 import decimal