crosshair-tool 0.0.97__cp314-cp314-macosx_11_0_arm64.whl

crosshair/libimpl/builtinslib_test.py

@@ -0,0 +1,3705 @@
+# -*- coding: utf-8 -*-
+
+import collections.abc
+import copy
+import dataclasses
+import enum
+import math
+import operator
+import re
+import sys
+from abc import ABC, abstractmethod
+from array import array
+from numbers import Integral
+from typing import (
+    Callable,
+    Dict,
+    FrozenSet,
+    Hashable,
+    Iterable,
+    List,
+    Mapping,
+    MutableMapping,
+    Optional,
+    Sequence,
+    Set,
+    SupportsAbs,
+    SupportsBytes,
+    SupportsFloat,
+    SupportsInt,
+    SupportsRound,
+    Tuple,
+    Type,
+    TypeVar,
+    Union,
+    get_type_hints,
+)
+
+import pytest
+import z3  # type: ignore
+
+from crosshair import type_repo
+from crosshair.core import (
+    analyze_function,
+    deep_realize,
+    proxy_for_type,
+    realize,
+    standalone_statespace,
+)
+from crosshair.core_and_libs import run_checkables
+from crosshair.dynamic_typing import origin_of
+from crosshair.libimpl.builtinslib import (
+    LazyIntSymbolicStr,
+    ModelingDirector,
+    PreciseIeeeSymbolicFloat,
+    RealBasedSymbolicFloat,
+    SymbolicArrayBasedUniformTuple,
+    SymbolicBool,
+    SymbolicByteArray,
+    SymbolicBytes,
+    SymbolicInt,
+    SymbolicList,
+    SymbolicObject,
+    SymbolicRange,
+    SymbolicType,
+    crosshair_types_for_python_type,
+)
+from crosshair.options import AnalysisOptionSet
+from crosshair.statespace import (
+    CANNOT_CONFIRM,
+    CONFIRMED,
+    EXEC_ERR,
+    POST_FAIL,
+    MessageType,
+    StateSpace,
+)
+from crosshair.test_util import (
+    check_exec_err,
+    check_messages,
+    check_states,
+    summarize_execution,
+)
+from crosshair.tracers import NoTracing, ResumedTracing
+from crosshair.util import (
+    CrossHairInternal,
+    CrossHairValue,
+    IgnoreAttempt,
+    UnknownSatisfiability,
+)
+
+
+class Cat:
+    def size(self) -> int:
+        return 1
+
+
+class BiggerCat(Cat):
+    def size(self) -> int:
+        return 2
+
+
+class AbstractBase(ABC):
+    @abstractmethod
+    def do(self):
+        pass
+
+
+class ConcreteSubclass(AbstractBase):
+    def do(self):
+        pass
+
+
+# Type repo doesn't load crosshair classes by default; load manually:
+type_repo._add_class(AbstractBase)
+type_repo._add_class(ConcreteSubclass)
+
+
+class Color(enum.Enum):
+    RED = 0
+    BLUE = 1
+    GREEN = 2
+
+
+@dataclasses.dataclass
+class SmokeDetector:
+    """inv: not (self._is_plugged_in and self._in_original_packaging)"""
+
+    _in_original_packaging: bool
+    _is_plugged_in: bool
+
+    def signaling_alarm(self, air_samples: List[str]) -> bool:
+        """
+        pre: self._is_plugged_in
+        post: implies('smoke' in air_samples, _ == True)
+        """
+        return "smoke" in air_samples
+
+
+if sys.version_info >= (3, 9):
+    from typing import TypedDict
+
+    class Movie(TypedDict):
+        name: str
+        year: int
+
+
+INF = float("inf")
+NAN = float("nan")
+
+
+def test_crosshair_types_for_python_type() -> None:
+    assert crosshair_types_for_python_type(int) == ((SymbolicInt, 1.0),)
+    assert crosshair_types_for_python_type(SmokeDetector) == ()
+
+
+def test_isinstance():
+    with standalone_statespace:
+        with NoTracing():
+            f = RealBasedSymbolicFloat("f")
+        assert isinstance(f, float)
+        assert not isinstance(f, int)
+
+
+def test_smtfloat_like_a_float():
+    with standalone_statespace:
+        with NoTracing():
+            f1 = RealBasedSymbolicFloat("f")
+        f2 = type(f1)(12)
+        with NoTracing():
+            assert isinstance(f2, float)
+            assert f2 == 12.0
+
+
+def test_bool_simple_conditional_fail() -> None:
+    def f(a: bool, b: bool) -> bool:
+        """post: _ == a"""
+        return True if a else b
+
+    check_states(f, POST_FAIL)
+
+
+def test_bool_simple_conditional_ok() -> None:
+    def f(a: bool, b: bool) -> bool:
+        """post: _ == a or b"""
+        return True if a else b
+
+    check_states(f, CONFIRMED)
+
+
+def test_bool_ors_fail() -> None:
+    def f(a: bool, b: bool, c: bool, d: bool) -> bool:
+        """post: _ == (a ^ b) or (c ^ d)"""
+        return a or b or c or d
+
+    check_states(f, POST_FAIL)
+
+
+def test_bool_ors() -> None:
+    def f(a: bool, b: bool, c: bool, d: bool) -> bool:
+        """
+        pre: (not a) and (not d)
+        post: _ == (a ^ b) or (c ^ d)
+        """
+        return a or b or c or d
+
+    check_states(f, CONFIRMED)
+
+
+def test_bool_as_numbers() -> None:
+    def f(a: bool, b: bool) -> int:
+        """post: _ in (1, 2)"""
+        return (a * b) + True
+
+    check_states(f, CONFIRMED)
+
+
+def test_int___floordiv___ok() -> None:
+    def f(n: int, d: int) -> Tuple[int, int]:
+        """
+        pre: n in (5, -5)
+        pre: d in (5, 3, -3, -5)
+        post: _[0] == _[1]
+        """
+        return ((n // d), (int(n) // int(d)))
+
+    check_states(f, CONFIRMED)
+
+
+def test_number_simple_compare_ok() -> None:
+    def f(i: List[int]) -> bool:
+        """
+        pre: 10 < len(i)
+        post: _
+        """
+        return 9 < len(i[1:])
+
+    check_states(f, CONFIRMED)
+
+
+def test_number_promotion_compare_ok() -> None:
+    def f(i: int, f: float) -> bool:
+        """
+        pre: i == 7
+        pre: f == 7.0
+        post: _
+        """
+        return i == f and f >= i and i >= f
+
+    check_states(f, CONFIRMED)
+
+
+def test_numeric_promotions() -> None:
+    def f(b: bool, i: int) -> Tuple[int, float, float]:
+        """
+        post: _ != (101, 4.14, 13.14)
+        """
+        return ((b + 100), (b + 3.14), (i + 3.14))
+
+    check_states(f, POST_FAIL)
+
+
+def test_float_as_bool() -> None:
+    def f(x: float, y: float):
+        """
+        pre: math.isfinite(x) and math.isfinite(y)
+        post: _ == x or _ == y
+        """
+        return x or y
+
+    check_states(f, CONFIRMED)
+
+
+def test_int_reverse_operators() -> None:
+    def f(i: int) -> float:
+        """
+        pre: i != 0
+        post: _ != 1
+        """
+        return (1 + i) + (1 - i) + (1 / i)
+
+    check_states(f, POST_FAIL)
+
+
+@pytest.mark.demo
+def test_int___add___method():
+    def f(a: int, b: int) -> int:
+        """
+        Can the sum of two consecutive integers be 37?
+
+        pre: a + 1 == b
+        post: _ != 37
+        """
+        return a + b
+
+    check_states(f, POST_FAIL)
+
+
+@pytest.mark.demo
+def test_int___mod___method():
+    def f(a: int) -> int:
+        """
+        Can the last digit of a given large number be 3?
+
+        pre: a > 1234
+        post: _ != 3
+        """
+        return a % 10
+
+    check_states(f, POST_FAIL)
+
+
+@pytest.mark.demo
+def test_int___mul___method():
+    def f(a: int, b: int) -> int:
+        """
+        Can we multiply two integers and return 42?
+
+        NOTE: Although this example works, nonlinear integer arithmetic can not
+        always be effectively reasoned about.
+
+        pre: a > b > 1
+        post: _ != 42
+        """
+        return a * b
+
+    check_states(f, POST_FAIL)
+
+
+@pytest.mark.demo("yellow")
+def test_int___pow___method():
+    def f(a: int) -> int:
+        """
+        Can the given integer, cubed, equal 343?
+
+        NOTE: Although this example works, nonlinear integer arithmetic can not
+        always be effectively reasoned about. This is particularly true when
+        the exponent is symbolic.
+
+
+        post: _ != 343
+        """
+        return a**3
+
+    check_states(f, POST_FAIL)
+
+
+def test_int___pow___to_ieee_float():
+    with standalone_statespace as space:
+        with NoTracing():
+            space.extra(ModelingDirector).global_representations[
+                float
+            ] = PreciseIeeeSymbolicFloat
+            a = SymbolicInt("a")
+        with pytest.raises(UnknownSatisfiability):
+            sqrt_a = a**0.5
+
+
+def test_int___pow___to_real_based_float():
+    with standalone_statespace as space:
+        with NoTracing():
+            space.extra(ModelingDirector).global_representations[
+                float
+            ] = RealBasedSymbolicFloat
+            a = SymbolicInt("a")
+        sqrt_a = a**0.5
+        with pytest.raises(UnknownSatisfiability):
+            realize(sqrt_a == 3)
+
+
+@pytest.mark.demo
+def test_int___sub___method():
+    def f(a: int) -> int:
+        """
+        Can we subtract from 42 and get something larger?
+
+        post: _ <= 42
+        """
+        return 42 - a
+
+    check_states(f, POST_FAIL)
+
+
+def test_int___rsub__() -> None:
+    def f(i: int) -> float:
+        """
+        post: _ != 42
+        """
+        return 1 - i
+
+    check_states(f, POST_FAIL)
+
+
+@pytest.mark.demo
+def test_int___floordiv___method() -> None:
+    def f(a: int, b: int) -> int:
+        """
+        Can the average of two integers equal either one?
+
+        pre: a < b
+        post: a < _ < b
+        """
+        return (a + b) // 2
+
+    check_states(f, POST_FAIL)
+
+
+def test_int___floordiv___bounds() -> None:
+    def f(a: int, b: int) -> int:
+        """
+        pre: a < b
+        post: a <= _ < b
+        """
+        return (a + b) // 2
+
+    check_states(f, CONFIRMED)
+
+
+def test_int_bitwise_fail() -> None:
+    def f(a: int, b: int) -> int:
+        """
+        pre: 0 <= a <= 3
+        pre: 0 <= b <= 3
+        post: _ < 7
+        """
+        return (a << 1) ^ b
+
+    check_states(f, POST_FAIL)
+
+
+def test_int_bitwise_ok() -> None:
+    def f(a: int, b: int) -> int:
+        """
+        pre: 0 <= a <= 3
+        pre: 0 <= b <= 3
+        post: _ <= 7
+        """
+        return (a << 1) ^ b
+
+    check_states(f, CONFIRMED)
+
+
+@pytest.mark.demo
+def test_int___truediv___method() -> None:
+    def f(a: int, b: int) -> float:
+        """
+        Can we find an integer that is half as large as another?
+
+        pre: b != 0
+        post: _ != 0.5
+        """
+        return a / b
+
+    check_states(f, POST_FAIL)
+
+
+def test_trunc_fail() -> None:
+    def f(n: float) -> int:
+        """
+        pre: math.isfinite(n)
+        pre: n > 100
+        post: _ < n
+        """
+        return math.trunc(n)
+
+    check_states(f, POST_FAIL)
+
+
+@pytest.mark.demo
+def test_range() -> None:
+    def f(n: int) -> Sequence[int]:
+        """
+        Can the length of range() be different than the integer we pass to it?
+
+        post: len(_) == n
+        """
+        return range(n)
+
+    check_states(f, POST_FAIL)
+
+
+def test_round_fail() -> None:
+    def f(n1: int, n2: int) -> Tuple[int, int]:
+        """
+        pre: n1 < n2
+        post: _[0] < _[1] # because we round towards even
+        """
+        return (round(n1 + 0.5), round(n2 + 0.5))
+
+    check_states(f, POST_FAIL)
+
+
+def test_round_unknown() -> None:
+    def f(num: float, ndigits: Optional[int]) -> float:
+        """
+        pre: math.isfinite(num)
+        post: isinstance(_, int) == (ndigits is None)
+        """
+        return round(num, ndigits)
+
+    check_states(f, CANNOT_CONFIRM)
+
+
+def test_float_isinstance() -> None:
+    def f(x: float) -> float:
+        """post: isinstance(_, float)"""
+        return x
+
+    check_states(f, CONFIRMED)
+
+
+def test_mismatched_types() -> None:
+    def f(x: float, y: list) -> float:
+        """
+        pre: x == 1.0 and y == []
+        post: _ == 1
+        """
+        return x + y  # type: ignore
+
+    (actual, expected) = check_exec_err(f, "TypeError: unsupported operand type")
+    assert actual == expected
+
+
+def test_bool_bitwise_negation() -> None:
+    def f(x: bool) -> float:
+        """
+        pre: x == True
+        post: _ == -2
+        """
+        return ~x
+
+    check_states(f, CONFIRMED)
+
+
+def test_float_from_hex() -> None:
+    def f(s: str) -> float:
+        """
+        pre: s == '0x3.a7p10'
+        post: _ == 3740.0
+        """
+        return float.fromhex(s)
+
+    check_states(f, CONFIRMED)
+
+
+def test_int_from_byte_iterator(space) -> None:
+    byts = proxy_for_type(bytes, "byts")
+    with ResumedTracing():
+        space.add(len(byts) == 2)
+        number = int.from_bytes(iter(byts), byteorder="little")
+        assert space.is_possible(number == 5)
+
+
+def test_int_from_bytes(space) -> None:
+    byts = proxy_for_type(bytes, "byts")
+    with ResumedTracing():
+        space.add(len(byts) == 2)
+        number = int.from_bytes(byts, byteorder="little")
+        assert space.is_possible(number == 5)
+
+
+def test_int_nonlinear() -> None:
+    def make_bigger(x: int, e: int) -> float:
+        """
+        pre: e > 1
+        post: __return__ !=  592704
+        """
+        # Expenentation is not SMT-solvable. (z3 gives unsat for this)
+        # But CrossHair gracefully falls back to realized values, yielding
+        # the counterexample of: 84 ** 3
+        return x**e
+
+    check_states(make_bigger, POST_FAIL)
+
+
+@pytest.mark.demo
+def test_int___str___method() -> None:
+    def f(x: int) -> str:
+        """
+        Can any input make this function return the string "321"?
+
+        post: _ != "321"
+        """
+        return str(x)
+
+    check_states(f, POST_FAIL)
+
+
+@pytest.mark.demo
+def test_int___repr___method() -> None:
+    def f(x: int) -> str:
+        """
+        Can any input make this function return the string "321"?
+
+        post: _ != '321'
+        """
+        return repr(x)
+
+    check_states(f, POST_FAIL)
+
+
+@pytest.mark.parametrize("b", (False, 1, -2.0, NAN, INF, -INF))
+@pytest.mark.parametrize("op", (operator.lt, operator.eq, operator.add, operator.mul))
+def test_bool_ops(b, op):
+    with standalone_statespace as space:
+        with NoTracing():
+            a = SymbolicBool("a")
+        space.add(a)
+        symbolic_ret = summarize_execution(lambda: op(a, b))
+        concrete_ret = summarize_execution(lambda: op(realize(a), b), detach_path=False)
+        assert symbolic_ret == concrete_ret
+
+
+@pytest.mark.parametrize("b", (False, 1, -2.0, NAN, INF, -INF))
+@pytest.mark.parametrize(
+    "op",
+    (operator.lt, operator.eq, operator.add, operator.mul, operator.eq, operator.ne),
+)
+def test_float_ops(b, op):
+    with standalone_statespace as space:
+        with NoTracing():
+            a = space.extra(ModelingDirector).choose(float)("a")
+        space.add(a < 0)
+        symbolic_ret = summarize_execution(lambda: op(a, b))
+        concrete_ret = summarize_execution(lambda: op(realize(a), b), detach_path=False)
+        assert symbolic_ret == concrete_ret
+
+
+def test_int_from_str():
+    def f(a: str) -> int:
+        """
+        post: _ != 7
+        raises: ValueError
+        """
+        return int(a)
+
+    check_states(f, POST_FAIL)
+
+
+def test_int_from_str_with_bases(space):
+    i = proxy_for_type(int, "i")
+    s_11 = proxy_for_type(str, "s_11")
+    s_a = proxy_for_type(str, "s_a")
+    with ResumedTracing():
+        space.add(len(s_11) == 2)
+        space.add(s_11 == "11")
+        space.add(len(s_a) == 1)
+        space.add(s_a == "a")
+        assert int(s_11, 4) == 5
+        assert int(s_11, 16) == 17
+        assert int(s_11, 36) == 37
+        assert int(s_11, 0) == 11
+        assert int(s_a, 16) == 10
+        with pytest.raises(TypeError):
+            assert int(s_a, base=2.5)
+        with pytest.raises(TypeError):
+            int(i, base="foo")
+
+
+def test_easy_float_from_str():
+    def f(a: str) -> float:
+        """
+        post: _ != 0.0
+        raises: ValueError
+        """
+        return float(a)
+
+    check_states(
+        f,
+        MessageType.POST_FAIL,
+        AnalysisOptionSet(max_iterations=100),
+    )
+
+
+def test_float_from_three_digit_str():
+    with standalone_statespace as space:
+        with NoTracing():
+            codepoints = [
+                proxy_for_type(int, "xat0"),
+                proxy_for_type(int, "xat1"),
+                proxy_for_type(int, "xat2"),
+            ]
+            x = LazyIntSymbolicStr(codepoints)
+        for point in codepoints:
+            space.add(point >= ord("0"))
+            space.add(point <= ord("9"))
+        asfloat = float(x)
+        assert space.is_possible(asfloat <= 999)
+        assert not space.is_possible(asfloat > 999)
+        assert space.is_possible(asfloat == 0)  # (because "000" is a valid float)
+        assert not space.is_possible(asfloat == 500.5)
+
+
+@pytest.mark.demo("yellow")
+def test_float___pow___operator():
+    def f(a: float) -> float:
+        """
+        Can the given float, cubed, equal 0.125?
+
+        NOTE: Although this example works, nonlinear arithmetic is quite difficult
+        to reason about in most cases.
+
+        post: _ != 0.125
+        """
+        return a**3
+
+    check_states(f, POST_FAIL)
+
+
+def test_int_bitwise_find_negative_input():
+    def f(x: int) -> int:
+        """
+        pre: x < 0
+        post: _ != 7
+        """
+        return x & 255
+
+    check_states(f, POST_FAIL)
+
+
+@pytest.mark.parametrize("val", [-256, 2**16] + list(range(-4, 9, 2)))
+def test_int_bit_length(val):
+    with standalone_statespace as space:
+        x = proxy_for_type(int, "x")
+        space.add(x == val)
+        assert realize(x.bit_length()) == val.bit_length()
+
+
+@pytest.mark.parametrize(
+    "val", [-256, -(2**15), 2**9, 2**15 - 1] + list(range(-4, 9, 3))
+)
+def test_int_to_bytes(val):
+    with standalone_statespace as space:
+        x = proxy_for_type(int, "x")
+        space.add(x == val)
+        assert realize(x.to_bytes(2, "big", signed=True)) == val.to_bytes(
+            2, "big", signed=True
+        )
+
+
+def test_int_format():
+    with standalone_statespace as space:
+        with NoTracing():
+            x = SymbolicInt("x")
+        space.add(x == 42)
+        assert x.__format__("") == "42"
+        # TODO this fails:
+        # assert x.__format__("f") == "42.000000"
+
+
+def test_class_format():
+    with standalone_statespace as space:
+        with NoTracing():
+            t = SymbolicType("t", Type[int])
+            space.add(t.var == SymbolicType._coerce_to_smt_sort(int))
+        assert "a{}b".format(t) == "a<class 'int'>b"
+
+
+@pytest.mark.demo
+def test_sorted() -> None:
+    def f(lst: List[int]) -> List[int]:
+        """
+        Can sorting shift the number 4002 to the front?
+
+        pre: len(lst) >= 3
+        pre: lst[0] > 4002
+        post: _[0] != 4002
+        """
+        return list(sorted(lst))
+
+    check_states(f, POST_FAIL)
+
+
+def test_str___bool__() -> None:
+    def f(a: str) -> str:
+        """post: a"""
+        return a
+
+    check_states(f, POST_FAIL)
+
+
+@pytest.mark.demo
+def test_str___mul___method() -> None:
+    def f(a: str) -> str:
+        """
+        Can this string-trippling-function produce a 6-character string?
+
+        post: len(_) != 6
+        """
+        return 3 * a
+
+    check_states(f, POST_FAIL)
+
+
+def test_str___mul___ok() -> None:
+    def f(a: str) -> str:
+        """
+        pre: len(a) == 2
+        post: len(_) == 10
+        """
+        return a * 3 + 2 * a
+
+    check_states(f, CONFIRMED)
+
+
+def test_str___mul___by_symbolic_fail() -> None:
+    def f(i: int) -> str:
+        """post: len(_) != 6"""
+        return "a\x00b" * i
+
+    check_states(f, POST_FAIL)
+
+
+def test_str___mul___full_symbolic_multiply_unknown() -> None:
+    def f(s: str, i: int) -> str:
+        """
+        pre: s and i > 0
+        post: _[0] == s[0]
+        """
+        return s * i
+
+    check_states(f, CANNOT_CONFIRM)
+
+
+def test_str___add___prefixing_fail() -> None:
+    def f(a: str, indent: bool) -> str:
+        """post: len(_) == len(a) + indent"""
+        return ("  " if indent else "") + a
+
+    check_states(f, POST_FAIL)
+
+
+def test_str___add___prefixing_ok() -> None:
+    def f(a: str, indent: bool) -> str:
+        """post: len(_) == len(a) + (2 if indent else 0)"""
+        return ("  " if indent else "") + a
+
+    check_states(f, CONFIRMED)
+
+
+@pytest.mark.demo
+def test_str___contains___method() -> None:
+    def f(a: str) -> str:
+        """
+        What 3 characters can we append to "purple" so that the result contains "please"?
+
+        pre: len(a) == 3
+        post: "please" not in _
+        """
+        return "purple" + a
+
+    check_states(f, POST_FAIL)
+
+
+def test_str_find_does_not_realize_string_length() -> None:
+    def f(a: str) -> str:
+        """post: len(_) != 100"""
+        a.find("abc")
+        return a
+
+    check_states(f, POST_FAIL)
+
+
+def test_str_find_with_limits_ok() -> None:
+    def f(a: str) -> int:
+        """post: _ == -1"""
+        return a.find("abc", 1, 3)
+
+    check_states(f, CONFIRMED)
+
+
+def test_str_find_with_negative_limits_fail() -> None:
+    def f(a: str) -> int:
+        """post: _ == -1"""
+        return a.find("ab", -2, 3)
+
+    check_states(f, POST_FAIL)
+
+
+def test_str_ljust_fail() -> None:
+    def f(s: str) -> str:
+        """post: len(_) == len(s)"""
+        return s.ljust(3, " ")
+
+    check_states(f, POST_FAIL)
+
+
+def test_str_rfind_with_limits_ok() -> None:
+    def f(a: str) -> int:
+        """post: _ == -1"""
+        return a.rfind("abc", 1, 3)
+
+    check_states(f, CONFIRMED)
+
+
+def test_str_rfind_with_negative_limits_fail() -> None:
+    def f(a: str) -> int:
+        """post: _ == -1"""
+        return a.rfind("ab", -2, 3)
+
+    check_states(f, POST_FAIL)
+
+
+def test_str_rindex_fail() -> None:
+    def f(a: str) -> int:
+        """post: _ != 2"""
+        try:
+            return a.rindex("abc")
+        except ValueError:
+            return 0
+
+    check_states(f, POST_FAIL)
+
+
+def test_str_rindex_err() -> None:
+    def f(a: str) -> int:
+        """post: True"""
+        return a.rindex("abc", 1, 3)
+
+    check_states(f, EXEC_ERR)
+
+
+def test_str_rjust_fail() -> None:
+    def f(s: str) -> str:
+        """post: len(_) == len(s)"""
+        return s.rjust(3, "x")
+
+    check_states(f, POST_FAIL)
+
+
+@pytest.mark.demo
+def test_str_replace_method() -> None:
+    def f(a: str) -> str:
+        """
+        Can this function return a changed string?
+
+        post: _ == a
+        """
+        return a.replace("abc", "x", 1)
+
+    check_states(f, POST_FAIL)
+
+
+def test_str_startswith(space) -> None:
+    symbolic_char = proxy_for_type(str, "x")
+    symbolic_empty = proxy_for_type(str, "y")
+    with ResumedTracing():
+        space.add(len(symbolic_char) == 1)
+        space.add(len(symbolic_empty) == 0)
+        assert symbolic_char.startswith(symbolic_empty)
+        assert symbolic_char.startswith(symbolic_char)
+        assert symbolic_char.startswith(("foo", symbolic_empty))
+        assert not symbolic_char.startswith(("foo", "bar"))
+        assert symbolic_char.startswith(("", "bar"))
+        assert symbolic_char.startswith("")
+        assert symbolic_char.startswith(symbolic_empty, 1)
+        assert symbolic_char.startswith(symbolic_empty, 1, 1)
+        assert str.startswith(symbolic_char, symbolic_empty)
+        assert "foo".startswith(symbolic_empty)
+        assert not "".startswith(symbolic_char)
+
+        # Yes, the empty string is findable off the left side but not the right
+        assert "x".startswith("", -10, -9)
+        assert symbolic_char.startswith(symbolic_empty, -10, -9)
+        assert not "x".startswith("", 9, 10)
+        assert not symbolic_char.startswith(symbolic_empty, 9, 10)
+
+
+@pytest.mark.demo
+def test_str_index_method() -> None:
+    def f(a: str) -> int:
+        """
+        Can we find "abc" at position 2?
+
+        raises: ValueError
+        post: _ != 2
+        """
+        return a.rindex("abc")
+
+    check_states(f, POST_FAIL)
+
+
+def test_str_index_err() -> None:
+    def f(s1: str, s2: str) -> int:
+        """
+        pre: s1 == 'aba'
+        pre: 'ab' in s2
+        post: True
+        """
+        return s1.index(s2)
+
+    # index() raises ValueError when a match isn't found:
+    (actual, expected) = check_exec_err(f, "ValueError")
+    assert actual == expected
+
+
+def test_str_negative_index_slicing() -> None:
+    def f(s: str) -> Tuple[str, str]:
+        """post: sum(map(len, _)) == len(s) - 1"""
+        idx = s.find(":")
+        return (s[:idx], s[idx + 1 :])
+
+    check_states(f, POST_FAIL)  # (fails when idx == -1)
+
+
+def test_str_starts_and_ends_ok() -> None:
+    def f(s: str) -> str:
+        """
+        pre: s == 'aba'
+        post: s.startswith('ab')
+        post: s.endswith('ba')
+        """
+        return s
+
+    check_states(f, CONFIRMED)
+
+
+@pytest.mark.demo
+def test_str_count_method() -> None:
+    def f(s: str) -> int:
+        """
+        Can this function find two "a" characters?
+
+        post: _ != 2
+        """
+        return s.count("a")
+
+    check_states(f, POST_FAIL)
+
+
+@pytest.mark.demo
+def test_str_split_method() -> None:
+    def f(s: str) -> list:
+        """
+        Does any string comma-split into "a" and "b"?
+
+        post: _ != ['a', 'b']
+        """
+        return s.split(",")
+
+    check_states(f, POST_FAIL)
+
+
+def test_str_rsplit_fail() -> None:
+    def f(s: str) -> list:
+        """post: __return__ != ['a', 'b']"""
+        return s.rsplit(":", 1)
+
+    check_states(f, POST_FAIL)
+
+
+def test_str_partition_ok() -> None:
+    def f(s: str) -> tuple:
+        """
+        pre: len(s) == 3
+        post: len(_) == 3
+        """
+        return s.partition(":")
+
+    check_states(f, CONFIRMED)
+
+
+@pytest.mark.smoke
+@pytest.mark.demo
+def test_str_partition_method() -> None:
+    def f(s: str) -> tuple:
+        """
+        Does any input to this partitioning yield ("a", "bc", "d")?
+
+        post: _ != ("a", "bc", "d")
+        """
+        return s.partition("bc")
+
+    check_states(f, POST_FAIL)
+
+
+def test_str_rpartition_ok() -> None:
+    def f(s: str) -> tuple:
+        """
+        pre: len(s) == 2
+        post: len(_) == 3
+        """
+        return s.rpartition(":")
+
+    check_states(f, CONFIRMED)
+
+
+def test_str_rpartition_fail() -> None:
+    def f(s: str) -> tuple:
+        """
+        pre: len(s) == 4
+        post: _ != ("abb", "b", "")
+        """
+        return s.rpartition("b")
+
+    check_states(f, POST_FAIL)
+
+
+def test_str___ge___fail() -> None:
+    def f(s1: str, s2: str) -> bool:
+        """post: _"""
+        return s1 >= s2
+
+    check_states(f, POST_FAIL)
+
+
+@pytest.mark.demo
+def test_str___le___method() -> None:
+    def f(a: str, b: str) -> bool:
+        """
+        Can a be greater than b, even though its first charater is not?
+
+        pre: a[0] <= b[0]
+        post: _
+        """
+        return a <= b
+
+    check_states(f, POST_FAIL)
+
+
+def test_str_realized_compare() -> None:
+    def f(a: str, b: str) -> bool:
+        """
+        post: implies(_, a == b)
+        """
+        return realize(a) == b
+
+    check_states(f, CANNOT_CONFIRM)
+
+
+def test_str_int_comparison_fail() -> None:
+    def f(a: int, b: str) -> Tuple[bool, bool]:
+        """post: (not _[0]) or (not _[1])"""
+        return (a != b, b != a)
+
+    check_states(f, POST_FAIL)
+
+
+def test_str_int_comparison_ok() -> None:
+    def f(a: int, b: str) -> bool:
+        """post: _ == False"""
+        return a == b or b == a
+
+    check_states(f, CONFIRMED)
+
+
+def test_str_formatting_wrong_key() -> None:
+    def f(o: object) -> str:
+        """post: True"""
+        return "object of type {typ} with repr {zzzzz}".format(  # type: ignore
+            typ=type(o), rep=repr(o)
+        )
+
+    check_states(f, EXEC_ERR)
+
+
+def test_str_format_symbolic_format() -> None:
+    def f(fmt: str) -> str:
+        """
+        pre: '{}' in fmt
+        post: True
+        """
+        return fmt.format(ver=sys.version, platform=sys.platform)
+
+    check_states(f, EXEC_ERR)
+
+
+def test_str_format_percent_unknown() -> None:
+    def f(fmt: str) -> str:
+        """
+        pre: '%' not in fmt
+        post: True
+        """
+        return fmt % ()
+
+    check_states(f, CANNOT_CONFIRM)
+
+
+@pytest.mark.demo
+def test_str_join_method() -> None:
+    def f(items: List[str]) -> str:
+        """
+        Any inputs that produce a 5-character string?
+
+        pre: len(items) == 2
+        post: len(_) != 5
+        """
+        return "and".join(items)
+
+    check_states(f, POST_FAIL)
+
+
+def test_str_upper_fail() -> None:
+    def f(s: str) -> str:
+        """
+        Does any character uppercase to "F"?
+
+        pre: len(s) == 1
+        pre: s != "F"
+        post: __return__ != "F"
+        """
+        return s.upper()
+
+    # TODO: make this use case more efficient.
+    options = AnalysisOptionSet(per_path_timeout=20.0)
+    check_states(f, POST_FAIL, options)
+
+
+def test_csv_example() -> None:
+    def f(lines: List[str]) -> List[str]:
+        """
+        pre: all(',' in line for line in lines)
+        post: __return__ == [line.split(',')[0] for line in lines]
+        """
+        return [line[: line.index(",")] for line in lines]
+
+    check_states(f, CANNOT_CONFIRM)
+
+
+@pytest.mark.demo
+def test_str_zfill_method() -> None:
+    def f(s: str) -> str:
+        """
+        Can zero-filling a two-character string produce "0ab"?
+
+        pre: len(s) == 2
+        post: _ != "0ab"
+        """
+        return s.zfill(3)
+
+    check_states(f, POST_FAIL)
+
+
+@pytest.mark.demo("yellow")
+def test_str_format_method() -> None:
+    def f(s: str) -> str:
+        """
+        Does any substitution produce the string "abcdef"?
+
+        NOTE: CrossHair will not be effective with a symbolic template string;
+        e.g. trying to solve s.format("cd") is much more difficult.
+
+        post: _ != "abcdef"
+        """
+        return "ab{}ef".format(s)
+
+    check_states(f, POST_FAIL)
+
+
+def test_str_constructor() -> None:
+    with standalone_statespace as space:
+        with NoTracing():
+            x = LazyIntSymbolicStr("x")
+        assert str(x) is x
+
+
+def test_str_str() -> None:
+    with standalone_statespace:
+        with NoTracing():
+            x = LazyIntSymbolicStr("x")
+        strx = x.__str__()
+        with NoTracing():
+            assert isinstance(strx, str)
+
+
+def test_str_center():
+    with standalone_statespace as space:
+        with NoTracing():
+            string = LazyIntSymbolicStr("string")
+            fillch = LazyIntSymbolicStr("fillch")
+            sz = SymbolicInt("sz")
+            sz6 = SymbolicInt("sz6")
+        space.add(string.__len__() == 3)
+        space.add(fillch.__len__() == 1)
+        space.add(sz > 5)
+        space.add(sz6 == 6)
+        assert "boo".center(sz6) == " boo  "
+        symbolic_centered = "boo".center(sz, fillch)
+        starts_with_nonfill = ord(symbolic_centered[0]) != ord(fillch)
+        assert not space.is_possible(starts_with_nonfill)
+
+
+def test_str_map_chars() -> None:
+    with standalone_statespace:
+        with NoTracing():
+            string = LazyIntSymbolicStr(list(map(ord, "ab")))
+        codepoints = list(map(ord, string))
+
+
+@pytest.mark.demo
+def test_str___add___method() -> None:
+    def f(s: str) -> str:
+        """
+        Can any input make this function return "Hello World"?
+
+        post: _ != "Hello World"
+        """
+        return s + "World"
+
+    check_states(f, POST_FAIL)
+
+
+def test_str_bool():
+    with standalone_statespace as space, NoTracing():
+        a = LazyIntSymbolicStr("a")
+        with ResumedTracing():
+            space.add(a.__len__() > 0)
+            assert bool(a)
+            # Can we retain our symbolic state after forcing a positive truthiness?:
+            assert space.is_possible(a == "this")
+            assert space.is_possible(a == "that")
+
+
+def test_str_eq():
+    with standalone_statespace, NoTracing():
+        assert LazyIntSymbolicStr([]) == ""
+
+
+def test_str_getitem():
+    with standalone_statespace, NoTracing():
+        assert LazyIntSymbolicStr(list(map(ord, "abc")))[0] == "a"
+        assert LazyIntSymbolicStr(list(map(ord, "abc")))[-1] == "c"
+        assert LazyIntSymbolicStr(list(map(ord, "abc")))[-5:2] == "ab"
+
+
+def test_str_filter():
+    with standalone_statespace, NoTracing():
+        string = LazyIntSymbolicStr(list(map(ord, "  ")))
+        with ResumedTracing():
+            ret = list(filter(str.isspace, [string]))
+        assert ret == [string]
+
+
+def test_str_filter_with_none():
+    with standalone_statespace as space, NoTracing():
+        string = LazyIntSymbolicStr([ord("a")])
+        truthyint = proxy_for_type(int, "truthyint")
+        falseyint = proxy_for_type(int, "falseyint")
+        with ResumedTracing():
+            space.add(truthyint == 10)
+            space.add(falseyint == 0)
+            ret = deep_realize(list(filter(None, [falseyint, 42, 0, truthyint])))
+        assert ret == [42, 10]
+
+
+def test_str_find() -> None:
+    with standalone_statespace, NoTracing():
+        string = LazyIntSymbolicStr(list(map(ord, "aabc")))
+        assert string.find("ab") == 1
+
+
+def test_str_find_symbolic() -> None:
+    def f(s: str) -> int:
+        """
+        pre: len(s) == 3
+        post: _ == -1
+        """
+        return "haystack".find(s)
+
+    check_states(f, POST_FAIL)
+
+
+def test_str_find_notfound() -> None:
+    with standalone_statespace, NoTracing():
+        string = LazyIntSymbolicStr([])
+        assert string.find("abc", 1, 3) == -1
+
+
+def test_str_format_basic():
+    with standalone_statespace as space:
+        with NoTracing():
+            s = LazyIntSymbolicStr("s")
+        space.add(s.__len__() == 1)
+        assert space.is_possible(s == "z")
+        assert space.is_possible(ord("a{0}c".format(s)[1]) == ord("b"))
+
+
+def test_str_format_map():
+    with standalone_statespace as space:
+        with NoTracing():
+            s = LazyIntSymbolicStr("s")
+        space.add(s.__len__() == 1)
+        assert space.is_possible(s == "z")
+        assert space.is_possible(ord("a{foo}c".format_map({"foo": s})[1]) == ord("b"))
+
+
+def test_str_rfind() -> None:
+    with standalone_statespace, NoTracing():
+        string = LazyIntSymbolicStr(list(map(ord, "ababb")))
+        assert string.rfind("ab") == 2
+
+
+def test_str_rfind_notfound() -> None:
+    with standalone_statespace, NoTracing():
+        string = LazyIntSymbolicStr(list(map(ord, "ababb")))
+        assert string.rfind("ab") == 2
+
+
+def test_str_split_limits():
+    with standalone_statespace, NoTracing():
+        string = LazyIntSymbolicStr(list(map(ord, "a:b:c")))
+        parts = realize(string.split(":", 1))
+        assert parts == ["a", "b:c"]
+        parts = realize(string.split(":"))
+        assert parts == ["a", "b", "c"]
+
+
+def test_str_rsplit():
+    with standalone_statespace, NoTracing():
+        string = LazyIntSymbolicStr(list(map(ord, "a:b:c")))
+        parts = realize(string.rsplit(":", 1))
+        assert parts == ["a:b", "c"]
+
+
+def test_str_contains():
+    with standalone_statespace:
+        with NoTracing():
+            small = LazyIntSymbolicStr([ord("b"), ord("c")])
+            big = LazyIntSymbolicStr([ord("a"), ord("b"), ord("c"), ord("d")])
+        assert small in big
+        assert big not in small
+        assert small in "bc"
+        assert small not in "b"
+        assert "c" in small
+        assert "cd" not in small
+
+
+def test_str_deep_realize():
+    with standalone_statespace, NoTracing():
+        a = LazyIntSymbolicStr("a")
+        tupl = (a, (a,))
+        realized = deep_realize(tupl)
+    assert list(map(type, realized)) == [str, tuple]
+    assert list(map(type, realized[1])) == [str]
+    assert realized[0] is realized[1][0]
+
+
+def test_str_strip():
+    with standalone_statespace:
+        with NoTracing():
+            x = LazyIntSymbolicStr(list(map(ord, "  A b\n")))
+        assert x.strip() == "A b"
+
+
+def test_str_lower():
+    chr_Idot = "\u0130"  # Capital I with dot above
+    # (it's the only unicde char that lower()s to 2 characters)
+    with standalone_statespace:
+        with NoTracing():
+            x = LazyIntSymbolicStr(list(map(ord, "Ab" + chr_Idot)))
+        assert x.lower() == "abi\u0307"
+
+
+def test_str_title():
+    chr_lj = "\u01c9"  # "lj"
+    chr_Lj = "\u01c8"  # "Lj" (different from "LJ", "\u01c7")
+    with standalone_statespace:
+        with NoTracing():
+            lj = LazyIntSymbolicStr(list(map(ord, chr_lj)))
+            lja_b = LazyIntSymbolicStr(list(map(ord, chr_lj + "a_b")))
+        assert lja_b.title() == chr_Lj + "a_B"
+
+
+def test_object_deep_realize():
+    @dataclasses.dataclass
+    class Container:
+        contents: int
+
+    with standalone_statespace as space, NoTracing():
+        a = SymbolicObject("a", Container)
+        shallow = realize(a)
+        assert type(shallow) is Container
+        assert type(shallow.contents) is not int
+        deep = deep_realize(a)
+        assert type(deep) is Container
+        assert type(deep.contents) is int
+
+
+def test_seq_string_deep_realize():
+    with standalone_statespace as space, NoTracing():
+        tupl = SymbolicArrayBasedUniformTuple("s", List[str])
+        space.add(tupl._len() == 2)
+        realized = deep_realize(tupl)
+    assert list(map(type, realized)) == [str, str]
+
+
+@pytest.mark.demo
+def test_tuple___add___method():
+    def f(a: Tuple[int, ...]):
+        """
+        Can we get this function to return (1, 2, 3, 4)?
+
+        post: _ != (1, 2, 3, 4)
+        """
+        return (1,) + a + (4,)
+
+    check_states(f, POST_FAIL)
+
+
+@pytest.mark.demo
+def test_tuple___getitem___method() -> None:
+    def f(t: Tuple[int, ...], idx: int) -> int:
+        """
+        Can we find 42 in the given tuple at the given index?
+
+        pre: idx >= 0 and idx < len(t)
+        post: _ != 42
+        """
+        return t[idx]
+
+    check_states(f, POST_FAIL)
+
+
+@pytest.mark.demo
+def test_tuple___len___method():
+    def f(a: Tuple[int, ...]):
+        """
+        Can we find a tuple of length 8?
+
+        post: _ != 8
+        """
+        return len(a)
+
+    check_states(f, POST_FAIL)
+
+
+def test_tuple___repr__symbolic_in_concrete(space) -> None:
+    x = proxy_for_type(int, "x")
+    with ResumedTracing():
+        space.add(x == 4)  # type: ignore
+        container = (x, x)
+        assert repr(container) == "(4, 4)"
+
+
+def test_tuple___repr__symbolic_in_concrete_namedtuple(space) -> None:
+    NamedTupleClass = collections.namedtuple("NamedTupleClass", ["target"])
+    x = proxy_for_type(int, "x")
+    with ResumedTracing():
+        space.add(x == 4)  # type: ignore
+        container = NamedTupleClass(target=x)
+        assert repr(container) == "NamedTupleClass(target=4)"
+
+
+def test_tuple_range_intersection_fail() -> None:
+    def f(a: Tuple[int, int], b: Tuple[int, int]) -> Optional[Tuple[int, int]]:
+        """
+        pre: a[0] < a[1] and b[0] < b[1]
+        post: _[0] <= _[1]
+        """
+        return (max(a[0], b[0]), min(a[1], b[1]))
+
+    check_states(f, POST_FAIL)
+
+
+def test_tuple_range_intersection_ok() -> None:
+    def f(a: Tuple[int, int], b: Tuple[int, int]) -> Optional[Tuple[int, int]]:
+        """
+        pre: a[0] < a[1] and b[0] < b[1]
+        post: _ is None or _[0] <= _[1]
+        """
+        if a[1] > b[0] and a[0] < b[1]:  # (if the ranges overlap)
+            return (max(a[0], b[0]), min(a[1], b[1]))
+        else:
+            return None
+
+    check_states(f, CONFIRMED)
+
+
+def test_tuple_with_uniform_values_fail() -> None:
+    def f(a: Tuple[int, ...]) -> float:
+        """
+        post: True
+        """
+        return sum(a) / len(a)
+
+    check_states(f, EXEC_ERR)
+
+
+def test_tuple_with_uniform_values_ok() -> None:
+    def f(a: Tuple[int, ...]) -> Tuple[int, ...]:
+        """
+        pre: len(a) < 4
+        post: 0 not in _
+        """
+        return tuple(x for x in a if x)
+
+    check_states(f, CONFIRMED)
+
+
+def test_tuple_runtime_type() -> None:
+    def f(t: Tuple) -> Tuple:
+        """post: t != (1, 2)"""
+        return t
+
+    check_states(f, POST_FAIL)
+
+
+def test_empty_tuple(space) -> None:
+    t = proxy_for_type(Tuple[()], "t")
+    with ResumedTracing():
+        assert type(t) is tuple
+        assert t == ()
+
+
+def test_tuple_isinstance_check() -> None:
+    def f(uniform_tuple: Tuple[List, ...], basic_tuple: tuple) -> Tuple[bool, bool]:
+        """post: _ == (True, True)"""
+        return (isinstance(uniform_tuple, tuple), isinstance(basic_tuple, tuple))
+
+    check_states(f, CONFIRMED)
+
+
+def test_range___len___method() -> None:
+    def f(a: int) -> Iterable[int]:
+        """
+        post: len(_) == a or a < 0
+        """
+        return range(a)
+
+    check_states(f, CONFIRMED)
+
+
+@pytest.mark.demo
+def test_range___reversed___method() -> None:
+    def f(start: int, stop: int, step: int) -> List[int]:
+        """
+        Does some reversed range equal [12, 8]?
+
+        pre: step != 0
+        post: _ != [12, 8]
+        """
+        return list(reversed(range(start, stop, step)))
+
+    check_states(f, POST_FAIL)
+
+
+def test_range_slicing(space) -> None:
+    assert list(range(3, 40, 2)[5::-1]) == [13, 11, 9, 7, 5, 3]
+    # Repeat this with a symbolic range:
+    rng = proxy_for_type(range, "rng")
+    newstart = proxy_for_type(int, "newstart")
+    with ResumedTracing():
+        space.add(rng.start == 3)  # type: ignore
+        space.add(rng.stop == 40)  # type: ignore
+        space.add(rng.step == 2)  # type: ignore
+        space.add(newstart == 5)  # type: ignore
+        assert list(rng[newstart::-1]) == [13, 11, 9, 7, 5, 3]
+
+
+def test_range_realization(space) -> None:
+    rng = proxy_for_type(range, "rng")
+    assert isinstance(rng, SymbolicRange)
+    with ResumedTracing():
+        assert isinstance(rng, range)
+    realized_range = realize(rng)
+    assert isinstance(realized_range, range)
+    with ResumedTracing():
+        assert realized_range == rng
+        assert hash(realized_range) == hash(rng)
+
+
+@pytest.mark.demo
+def test_list___contains___method() -> None:
+    def f(a: int, b: List[int]) -> bool:
+        """
+        Is full containment checking equivalent to checking the first 3 elements?
+
+        post: _ == (a in b[:3])
+        """
+        return a in b
+
+    check_states(f, POST_FAIL)
+
+
+def test_list___contains___ok() -> None:
+    def f(a: int, b: List[int]) -> bool:
+        """
+        pre: 1 == len(b)
+        post: _ == (a == b[0])
+        """
+        return a in b
+
+    check_states(f, CONFIRMED)
+
+
+@pytest.mark.demo
+def test_list___add___method() -> None:
+    def f(a: List[int]) -> List[int]:
+        """
+        Does doubling the list always make it longer?
+
+        post: len(_) > len(a)
+        """
+        return a + a
+
+    check_states(f, POST_FAIL)
+
+
+def test_list___repr___symbolic_in_concrete(space) -> None:
+    x = proxy_for_type(int, "x")
+    with ResumedTracing():
+        space.add(x == 4)  # type: ignore
+        continer = [x]
+        assert f"{continer=}" == "continer=[4]"
+
+
+def test_list_doubling_ok() -> None:
+    def f(a: List[int]) -> List[int]:
+        """
+        post: len(_) > len(a) or not a
+        """
+        return a + a
+
+    check_states(f, CONFIRMED)
+
+
+def test_list_multiply_ok() -> None:
+    def f(a: List[int]) -> List[int]:
+        """post: len(_) == len(a) * 5"""
+        return a * 3 + 2 * a
+
+    check_states(f, CONFIRMED)
+
+
+def test_list_average() -> None:
+    def average(numbers: List[float]) -> float:
+        """
+        pre: len(numbers) > 0
+        post: min(numbers) <= _ <= max(numbers)
+        """
+        return sum(numbers) / len(numbers)
+
+    check_states(average, CANNOT_CONFIRM)
+
+
+def test_list_mixed_symbolic_and_literal_concat_ok() -> None:
+    def f(ls: List[int], i: int) -> List[int]:
+        """
+        pre: i >= 0
+        post: len(_) == len(ls) + 1
+        """
+        return (
+            ls[:i]
+            + [
+                42,
+            ]
+            + ls[i:]
+        )
+
+    check_states(f, CONFIRMED)
+
+
+def test_list_range_fail() -> None:
+    def f(ls: List[int]) -> List[int]:
+        """
+        pre: len(ls) == 3
+        post: len(_) > len(ls)
+        """
+        n: List[int] = []
+        for i in range(len(ls)):
+            n.append(ls[i] + 1)
+        return n
+
+    check_states(f, POST_FAIL)
+
+
+def test_list_range_ok() -> None:
+    def f(ls: List[int]) -> List[int]:
+        """
+        pre: ls and len(ls) < 10  # (max is to cap runtime)
+        post: _[0] == ls[0] + 1
+        """
+        n: List[int] = []
+        for i in range(len(ls)):
+            n.append(ls[i] + 1)
+        return n
+
+    check_states(f, CONFIRMED)
+
+
+def test_list_equality() -> None:
+    def f(ls: List[int]) -> List[int]:
+        """
+        pre: len(ls) > 0
+        post: _ != ls
+        """
+        # extra check for positive equality:
+        assert ls == [x for x in ls], "list does not equal itself"
+        nl = ls[:]
+        nl[0] = 42
+        return nl
+
+    check_states(f, POST_FAIL)
+
+
+def test_list_extend_literal_unknown() -> None:
+    def f(ls: List[int]) -> List[int]:
+        """
+        post: _[:2] == [1, 2]
+        """
+        r = [1, 2, 3]
+        r.extend(ls)
+        return r
+
+    check_states(f, CANNOT_CONFIRM)
+
+
+@pytest.mark.demo
+def test_list___getitem___method() -> None:
+    def f(ls: List[int], idx: int) -> int:
+        """
+        Can we find 42 in the given list at the given index?
+
+        pre: idx >= 0 and idx < len(ls)
+        post: _ != 42
+        """
+        return ls[idx]
+
+    check_states(f, POST_FAIL)
+
+
+def test_list____getitem___error() -> None:
+    def f(ls: List[int], idx: int) -> int:
+        """
+        pre: idx >= 0 and len(ls) > 2
+        post: True
+        """
+        return ls[idx]
+
+    (actual, expected) = check_exec_err(f, "IndexError")
+    assert actual == expected
+
+
+def test_list____getitem___type_error() -> None:
+    def f(ls: List[int]) -> int:
+        """post: True"""
+        return ls[0.0:]  # type: ignore
+
+    (actual, expected) = check_exec_err(f, "TypeError")
+    assert actual == expected
+
+
+def test_list____getitem___ok() -> None:
+    def f(ls: List[int]) -> bool:
+        """
+        pre: len(ls) <= 3
+        post: _ == (7 in ls)
+        """
+        try:
+            return ls[ls.index(7)] == 7
+            return True
+        except ValueError:
+            return False
+
+    check_states(f, CONFIRMED)
+
+
+def test_list_nested_lists_fail() -> None:
+    def f(ls: List[List[int]]) -> int:
+        """
+        post: _ > 0
+        """
+        total = 0
+        for i in ls:
+            total += len(i)
+        return total
+
+    check_states(f, POST_FAIL)
+
+
+def test_list_nested_lists_ok() -> None:
+    def f(ls: List[List[int]]) -> int:
+        """
+        pre: len(ls) < 4
+        post: _ >= 0
+        """
+        total = 0
+        for i in ls:
+            total += len(i)
+        return total
+
+    check_states(f, CONFIRMED)
+
+
+def test_list_iterable() -> None:
+    def f(a: Iterable[int]) -> int:
+        """
+        pre: a
+        post: _ in a
+        """
+        return next(iter(a))
+
+    check_states(f, CONFIRMED)
+
+
+def test_list_isinstance_check() -> None:
+    def f(ls: List) -> bool:
+        """post: _"""
+        return isinstance(ls, list)
+
+    check_states(f, CONFIRMED)
+
+
+def test_list_slice_outside_range_ok() -> None:
+    def f(ls: List[int], i: int) -> List[int]:
+        """
+        pre: i >= len(ls)
+        post: _ == ls
+        """
+        return ls[:i]
+
+    check_states(f, CONFIRMED)
+
+
+def test_list_slice_amount() -> None:
+    def f(ls: List[int]) -> List[int]:
+        """
+        pre: len(ls) >= 3
+        post: len(_) == 1
+        """
+        return ls[2:3]
+
+    check_states(f, CONFIRMED)
+
+
+def test_list____setitem___ok() -> None:
+    def f(ls: List[int]) -> None:
+        """
+        pre: len(ls) >= 2
+        post[ls]:
+            ls[1] == 42
+            ls[2] == 43
+            len(ls) == 4
+        """
+        ls[1:-1] = [42, 43]
+
+    check_states(f, CONFIRMED)
+
+
+def test_list___setitem___out_of_bounds() -> None:
+    def f(ls: List[int], i: int) -> None:
+        """
+        pre: i != -1
+        post: ls == __old__.ls[:i] + __old__.ls[i+1:]
+        """
+        ls[i : i + 1] = []
+
+    check_states(f, CANNOT_CONFIRM)
+
+
+def test_list_insert_ok() -> None:
+    def f(ls: List[int]) -> None:
+        """
+        pre: len(ls) == 4
+        post[ls]:
+            len(ls) == 5
+            ls[2] == 42
+        """
+        ls.insert(-2, 42)
+
+    check_states(f, CONFIRMED)
+
+
+def test_list_insert_with_conversions() -> None:
+    def f(ls: List[Set[int]], a: bool, b: int) -> None:
+        """
+        # self.insert(a,b) with {'a': True, 'b': 10, 'self': [{0}]}
+        post: True
+        """
+        ls.insert(a, b)  # type: ignore
+
+    check_states(f, CONFIRMED)
+
+
+def test_list_pop_ok() -> None:
+    def f(ls: List[int]) -> None:
+        """
+        pre: ls == [4, 5]
+        post: ls == [4]
+        """
+        ls.pop()
+
+    check_states(f, CONFIRMED)
+
+
+def test_list_count_ok() -> None:
+    def f(ls: List[Dict[int, Dict[int, int]]]) -> int:
+        """
+        pre: ls == [{1: {2: 3}}]
+        post: _ == 1
+        """
+        return ls.count({1: {2: 3}})
+
+    check_states(f, CONFIRMED)
+
+
+@pytest.mark.smoke
+def test_list___setitem___ok() -> None:
+    def f(ls: List[int]) -> None:
+        """
+        pre: len(ls) >= 4
+        post[ls]: ls[3] == 42
+        """
+        ls[3] = 42
+
+    check_states(f, CONFIRMED)
+
+
+@pytest.mark.demo
+def test_list___delitem___method() -> None:
+    def f(ls: List[int]) -> None:
+        """
+        Can we trim the tail two elements and have three left over?
+
+        post[ls]: len(ls) != 3
+        """
+        del ls[-2:]
+
+    check_states(f, POST_FAIL)
+
+
+def test_list___delitem___ok() -> None:
+    def f(ls: List[int]) -> None:
+        """
+        pre: len(ls) == 5
+        post[ls]: len(ls) == 4
+        """
+        del ls[2]
+
+    check_states(f, CONFIRMED)
+
+
+def test_list___delitem___type_error() -> None:
+    def f(ls: List[float]) -> None:
+        """
+        pre: len(ls) == 0
+        post: True
+        """
+        del ls[1.0]  # type: ignore
+
+    (actual, expected) = check_exec_err(f, "TypeError")
+    assert actual == expected
+
+
+def test_list___delitem___out_of_bounds() -> None:
+    def f(ls: List[float]) -> None:
+        """post: True"""
+        del ls[1]
+
+    (actual, expected) = check_exec_err(f, "IndexError")
+    assert actual == expected
+
+
+def test_list_sort_ok() -> None:
+    def f(ls: List[int]) -> None:
+        """
+        pre: len(ls) == 3
+        post[ls]: ls[0] == min(ls)
+        """
+        ls.sort()
+
+    check_states(f, CONFIRMED)
+
+
+def test_list_reverse_ok() -> None:
+    def f(ls: List[int]) -> None:
+        """
+        pre: len(ls) == 2
+        post[ls]: ls[0] == 42
+        """
+        ls.append(42)
+        ls.reverse()
+
+    check_states(f, CONFIRMED)
+
+
+def test_list_comparison_type_error() -> None:
+    def f(a: List[Set], b: str):
+        """post: True"""
+        return a <= b  # type: ignore
+
+    (actual, expected) = check_exec_err(f, "TypeError")
+    assert actual == expected
+
+
+def test_list_shallow_realization():
+    with standalone_statespace as space:
+        nums = proxy_for_type(List[int], "nums")
+        numslen = len(nums)
+        space.add(numslen == 1)
+        with NoTracing():
+            realized = realize(nums)
+            assert type(realized) is list
+            assert len(realized) == 1
+            assert type(realized[0]) is SymbolicInt
+
+
+def test_list_concrete_with_symbolic_slice(space):
+    idx = proxy_for_type(int, "i")
+    with ResumedTracing():
+        space.add(1 <= idx)
+        space.add(idx <= 3)
+        prefix = [0, 1, 2, 3][:idx]
+        prefixlen = len(prefix)
+    assert isinstance(prefix, CrossHairValue)
+    assert isinstance(prefixlen, CrossHairValue)
+    with ResumedTracing():
+        assert space.is_possible(prefixlen == 1)
+        assert space.is_possible(prefixlen == 3)
+
+
+def test_list_copy(space):
+    lst = proxy_for_type(List[int], "lst")
+    with ResumedTracing():
+        # Mostly just ensure the various ways of copying don't explode
+        assert lst[:] is not lst
+        assert lst.copy() is not lst
+        assert copy.deepcopy(lst) is not lst
+        assert copy.copy(lst) is not lst
+
+
+def test_list_copy_compare_without_forking(space):
+    lst = proxy_for_type(List[int], "lst")
+    with ResumedTracing():
+        lst2 = copy.deepcopy(lst)
+    assert lst.inner is not lst2.inner
+    assert type(lst2) is SymbolicList
+    assert lst.inner.var is lst2.inner.var
+    with ResumedTracing():
+        are_same = lst == lst2
+    assert type(are_same) is SymbolicBool
+    assert not space.is_possible(z3.Not(are_same.var))
+
+
+def test_dict___bool___ok() -> None:
+    def f(a: Dict[int, str]) -> bool:
+        """
+        post[a]: _ == True
+        """
+        a[0] = "zero"
+        return bool(a)
+
+    check_states(f, CONFIRMED)
+
+
+def test_dict___iter___fail() -> None:
+    def f(a: Dict[int, str]) -> List[int]:
+        """
+        post[a]: 5 in _
+        """
+        a[10] = "ten"
+        return list(a.__iter__())
+
+    check_states(f, POST_FAIL)
+
+
+def test_dict___iter___ok() -> None:
+    def f(a: Dict[int, str]) -> List[int]:
+        """
+        pre: len(a) < 3
+        post[a]: 10 in _
+        """
+        a[10] = "ten"
+        return list(a.__iter__())
+
+    check_states(f, CONFIRMED)
+
+
+def test_dict___or___method():
+    with standalone_statespace as space:
+        d = proxy_for_type(Dict[int, int], "d")
+        space.add(len(d) == 0)
+        with pytest.raises(TypeError):
+            d | set()
+        if sys.version_info >= (3, 9):
+            assert d | {1: 2} == {1: 2}
+        else:
+            with pytest.raises(TypeError):
+                d | {1: 2}
+
+
+@pytest.mark.demo("yellow")
+def test_dict___delitem___method() -> None:
+    def f(a: Dict[str, int]) -> None:
+        """
+        Can deleting the key "foo" leave an empty dictionary?
+
+        NOTE: Deleting a symbolic key from a concrete dictionary is not effectively
+        reasoned about at present:
+
+            dictionary | key      | effective?
+            -----------+----------+-----------
+            symbolic  | *        | yes
+            *         | concrete | yes
+            concrete  | symbolic | no
+
+
+        raises: KeyError
+        post[a]: len(a) != 0
+        """
+        del a["foo"]
+
+    check_states(f, POST_FAIL)
+
+
+@pytest.mark.demo
+def test_dict___eq___method() -> None:
+    def f(t: dict) -> dict:
+        """
+        Can we find a dictionary that maps 50 to 100?
+
+        post: t != {50: 100}
+        """
+        return t
+
+    check_states(f, POST_FAIL)
+
+
+def test_dict___eq___deep() -> None:
+    def f(a: Dict[bool, set], b: List[Set[float]]) -> object:
+        """
+        pre: a == {True: set()}
+        pre: b == [set(), {1.0}]
+        post: _
+        """
+        if a == {True: set()}:
+            if b == [set(), {1.0}]:
+                return False
+        return True
+
+    check_states(f, POST_FAIL)
+
+
+def test_dict___eq___ok() -> None:
+    def f(d: Dict[int, int]) -> Dict[int, int]:
+        """post: _ == {**_}"""
+        return d
+
+    check_states(f, CANNOT_CONFIRM)
+
+
+@pytest.mark.demo
+def test_dict___getitem___method() -> None:
+    def f(m: Dict[int, int]):
+        """
+        Can we make a path from 0 to 2, by indexing into the dictionary twice?
+
+        pre: len(m) == 2
+        raises: KeyError
+        post: _ != 2
+        """
+        return m[m[0]]
+
+    check_states(f, POST_FAIL)
+
+
+def test_dict___getitem___implicit_conversion_for_keys_fail() -> None:
+    def f(m: Dict[complex, float], b: bool, i: int):
+        """
+        pre: not m
+        post: len(m) != 1
+        """
+        m[b] = 2.0
+        m[i] = 3.0
+
+    check_states(f, POST_FAIL)
+
+
+def test_dict__items__works_with_symbolic_self(space) -> None:
+    x = proxy_for_type(Dict[int, int], "x")
+    with ResumedTracing():
+        x[42] = 42
+        assert (42, 42) in list(dict.items(x))
+
+
+def test_dict___repr___symbolic_in_concrete(space) -> None:
+    x = proxy_for_type(int, "x")
+    with ResumedTracing():
+        space.add(x == 4)  # type: ignore
+        container = {x: x}
+        assert repr(container) == "{4: 4}"
+
+
+@pytest.mark.demo("yellow")
+def test_dict___setitem___method() -> None:
+    def f(a: Dict[int, int], k: int, v: int) -> None:
+        """
+        Can we make a dictionary assignment, and be left with {4: 5, 10: 20}?
+
+        NOTE: CrossHair cannot effectively handle the assignment of a symbolic key on a
+        concrete dictionary, e.g. `d={4:5}; d[k] = 20`
+
+            dictionary | key      | value | effective?
+            -----------+----------+-------+-----------
+            symbolic  | *        | *     | yes
+            *         | concrete | *     | yes
+            concrete  | symbolic | *     | no
+
+
+        post[a]: a != {4: 5, 10: 20}
+        """
+        a[k] = v
+
+    check_states(f, POST_FAIL)
+
+
+def test_dict___setitem___ok() -> None:
+    def f(a: Dict[int, int], k: int, v: int) -> None:
+        """
+        post[a]: a[k] == v
+        """
+        a[k] = v
+
+    check_states(f, CONFIRMED)
+
+
+def test_dict___setitem___on_copy() -> None:
+    def f(d: Dict[int, int]) -> Dict[int, int]:
+        """post: _ != d"""
+        d = d.copy()
+        d[42] = 100
+        return d
+
+    check_states(f, POST_FAIL)
+
+
+def TODO_test_dict___setitem___on_concrete() -> None:
+    # NOTE: This is very challenging to implement: opcode interception could upgrade
+    # the concrete dictionary to a symbolic, but the original dictionary may be aliased.
+    # One investigation: start everything out as symbolic by intercepting BUILD_MAP and
+    # BUILD_SET. This potentially also lets us detect writes to persistent state
+    # (because all pre-existing dicts/sets will be concrete).
+    def f(k: int, v: int) -> Dict[int, int]:
+        """post: _[100] == 200"""
+        d = {100: 200}
+        d[k] = v
+        return d
+
+    check_states(f, POST_FAIL)
+
+
+def test_dict___str__() -> None:
+    def f(a: Dict[int, str]) -> str:
+        """
+        pre: len(a) == 0
+        post: _ == '{}'
+        """
+        return str(a)
+
+    check_states(f, CONFIRMED)
+
+
+@pytest.mark.demo
+def test_dict_get_method():
+    def f(x: int) -> int:
+        """
+        Can we find the key that is mapped to 5?
+
+        post: _ != 5
+        """
+        a = {2: 3, 4: 5, 6: 7}
+        return a.get(x, 9)
+
+    check_states(f, POST_FAIL)
+
+
+def test_dict_get_with_defaults_ok() -> None:
+    def f(a: Dict[int, int]) -> int:
+        """post: (_ == 2) or (_ == a[4])"""
+        return a.get(4, 2)
+
+    check_states(f, CONFIRMED)
+
+
+def test_dict_items_ok() -> None:
+    def f(a: Dict[int, str]) -> Iterable[Tuple[int, str]]:
+        """
+        pre: len(a) < 5
+        post[a]: (10,'ten') in _
+        """
+        a[10] = "ten"
+        return a.items()
+
+    check_states(f, CONFIRMED)
+
+
+def test_dict_setdefault_float_int_comparison() -> None:
+    def f(a: Dict[int, int]):
+        """
+        pre: a == {2: 0}
+        post: _ == 0
+        """
+        return a.setdefault(2.0, {True: "0"})  # type: ignore
+
+    check_states(f, CONFIRMED)
+
+
+def test_dict_construction_from_generator(space):
+    with ResumedTracing():
+        d = dict((k, v) for k, v in [(1, 2), (3, 4)])
+        assert d == {1: 2, 3: 4}
+
+
+def test_dict_construction_copies_dict(space):
+    with ResumedTracing():
+        source = {1: 2, 3: 4}
+        copy = dict(source)
+        source[5] = 6
+        assert copy == {1: 2, 3: 4}
+
+
+def test_dict_construction_with_duplicate_keys(space):
+    with ResumedTracing():
+        d = dict([(1, 2), (3, 4), (1, 10), (5, 6), (3, 10)])
+        assert d == {1: 10, 3: 10, 5: 6}
+
+
+def test_dict_over_objects() -> None:
+    def f(a: Dict[object, object]) -> int:
+        """
+        post: _ >= 0
+        """
+        return len(a)
+
+    check_states(f, CONFIRMED)
+
+
+def test_dict_over_heap_objects() -> None:
+    def f(a: Dict[Tuple[int], int]) -> Optional[int]:
+        """
+        post: _ != 10
+        """
+        return a.get((5,))
+
+    check_states(f, POST_FAIL)
+
+
+def test_dict_complex_contents() -> None:
+    def f(d: Dict[Tuple[int, bool], Tuple[float, int]]) -> int:
+        """
+        post: _ > 0
+        """
+        if (42, True) in d:
+            return d[(42, True)][1]
+        else:
+            return 42
+
+    check_states(f, MessageType.POST_FAIL)
+
+
+def test_dict_isinstance_check() -> None:
+    def f(smtdict: Dict[int, int], heapdict: Dict) -> Tuple[bool, bool]:
+        """post: _ == (True, True)"""
+        return (isinstance(smtdict, dict), isinstance(heapdict, dict))
+
+    check_states(f, CONFIRMED)
+
+
+def test_dict_subtype_lookup() -> None:
+    def f(d: Dict[Tuple[int, str], int]) -> None:
+        """
+        pre: not d
+        post[d]: [(42, 'fourty-two')] == list(d.keys())
+        """
+        d[(42, "fourty-two")] = 1
+
+    check_states(f, CONFIRMED)
+
+
+def test_dict_complex_keys() -> None:
+    def f(dx: Dict[Tuple[int, str], int]) -> None:
+        """
+        pre: not dx
+        post[dx]:
+            len(dx) == 1
+            dx[(42, 'fourty-two')] == 1
+        """
+        dx[(42, "fourty-two")] = 1
+
+    check_states(f, CONFIRMED)
+
+
+def test_dict_has_unique_keys() -> None:
+    def f(d: Dict[Tuple[int, str], int]) -> None:
+        """
+        pre: len(d) == 2 and (1, 'one') in d
+        post[d]: (1, 'one') not in d
+        """
+        del d[(1, "one")]
+
+    check_states(f, CONFIRMED)
+
+
+def test_dict_wrong_key_type() -> None:
+    def f(d: Dict[int, int], s: str, i: int) -> bool:
+        """
+        post: _
+        raises: KeyError
+        """
+        if i == 0:
+            del d[s]  # type: ignore
+        elif i < 0:
+            d[s] = 7  # type: ignore
+        else:
+            _val = d[s]  # type: ignore
+        return True
+
+    check_states(f, CANNOT_CONFIRM)
+
+
+def test_dict_key_type_union() -> None:
+    def f(d: Dict[Union[int, str], int]) -> Dict:
+        """
+        pre: len(d) == 2
+        post: not (42 in d and '42' in d)
+        """
+        return d
+
+    check_states(f, POST_FAIL)
+
+
+if sys.version_info >= (3, 10):
+
+    def test_dict_type_union_operator() -> None:
+        def f(a: int | str, b: int | str) -> Tuple[int | str, int | str]:
+            """post: _ != (42, "hi")"""
+            return (a, b)
+
+        check_states(f, POST_FAIL)
+
+
+@pytest.mark.smoke
+def test_dict_nonuniform_dict_key_types() -> None:
+    def f(a: Dict[Hashable, int]) -> Dict[Hashable, int]:
+        """
+        pre: len(a) == 1
+        post: _[0] == 100
+        """
+        b: Dict[Hashable, int] = {0: 100}
+        b.update(a)
+        return b
+
+    check_states(f, POST_FAIL)
+
+
+def test_dict_inside_lists() -> None:
+    def f(dicts: List[Dict[int, int]]) -> Dict[int, int]:
+        """
+        pre: len(dicts) <= 1  # to narrow search space (would love to make this larger)
+        post: len(_) <= len(dicts)
+        """
+        ret = {}
+        for d in dicts:
+            ret.update(d)
+        return ret
+
+    check_states(f, POST_FAIL)
+
+
+@pytest.mark.skip(
+    reason="flakey; seems to get stuck sometimes despite timeout extensions"
+)
+def test_dict_inside_lists_with_identity() -> None:
+    def f(dicts: List[Dict[int, int]]):
+        """
+        Removes duplicate keys.
+        pre: len(dicts) == 2
+        pre:  len(dicts[0]) == 1
+        post: len(dicts[0]) == 1
+        """
+        # crosshair: max_uninteresting_iterations=50
+        seen: Set[int] = set()
+        for d in dicts:
+            for k in d.keys():
+                if k in seen:
+                    del d[k]
+                else:
+                    seen.add(k)
+
+    check_states(f, POST_FAIL)
+
+
+def test_dict_consistent_ordering() -> None:
+    def f(symbolic: Dict[int, int]) -> Tuple[List[int], List[int]]:
+        """post: _[0] == _[1]"""
+        return (list(symbolic.keys()), list(symbolic.keys()))
+
+    check_states(f, CANNOT_CONFIRM)
+
+
+def test_dict_ordering_after_mutations() -> None:
+    def f(d: Dict[int, int]) -> Tuple[Tuple[int, int], Tuple[int, int]]:
+        """
+        pre: len(d) == 3
+        post[d]: _[0] == _[1]
+        """
+        o1, middle, o2 = d.keys()
+        d[o1] = 42
+        d[o2] = 42
+        del d[middle]
+        n1, n2 = d.keys()
+        return ((o1, o2), (n1, n2))
+
+    check_states(f, CONFIRMED)
+
+
+def test_dict_alternate_mapping_types() -> None:
+    def f(m1: Mapping[int, int], m2: MutableMapping[int, int]) -> int:
+        """
+        pre: 1 in m1 and 2 in m2
+        post: _ != 10
+        """
+        return m1[1] + m2[2]
+
+    check_states(f, POST_FAIL)
+
+
+def test_dict_untyped_access():
+    def f(d: dict, k: int) -> dict:
+        """
+        pre: len(d) == 2  # (just to bound the search space a little)
+        pre: 42 in d
+        post: 42 in __return__
+        raises: KeyError
+        """
+        del d[k]
+        return d
+
+    check_states(f, MessageType.POST_FAIL)
+
+
+# NOTE: TypedDict appeared earlier than 3.9, but was not runtime-detectable until then
+if sys.version_info >= (3, 9):
+
+    def test_TypedDict_fail() -> None:
+        def f(td: Movie):
+            '''post: _['year'] != 2020 or _['name'] != "hi"'''
+            return td
+
+        check_states(f, POST_FAIL)
+
+    def test_TypedDict_in_container_fail() -> None:
+        def f(tdlist: List[Movie]):
+            """post: _[1]['year'] != 2020"""
+            return tdlist
+
+        check_states(f, POST_FAIL)
+
+
+@pytest.mark.smoke
+def test_set_basic_fail() -> None:
+    def f(a: Set[int], k: int) -> None:
+        """
+        pre: len(a) <= 2
+        post[a]: k+1 in a
+        """
+        a.add(k)
+
+    check_states(f, POST_FAIL)
+
+
+def test_set_basic_ok() -> None:
+    def f(a: Set[int], k: int) -> None:
+        """
+        post[a]: k in a
+        """
+        a.add(k)
+
+    check_states(f, CONFIRMED)
+
+
+def test_set_union_fail() -> None:
+    def f(a: Set[str], b: Set[str]) -> Set[str]:
+        """
+        pre: len(a) == len(b) == 1  # (just for test performance)
+        post: all(((i in a) and (i in b)) for i in _)
+        """
+        return a | b
+
+    check_states(f, POST_FAIL)
+
+
+def test_set_union_ok() -> None:
+    def f(a: Set[str], b: Set[str]) -> Set[str]:
+        """
+        post: all(((i in a) or (i in b)) for i in _)
+        """
+        return a | b
+
+    check_states(f, CANNOT_CONFIRM)
+
+
+def test_set_contains_different_but_equivalent() -> None:
+    def f(s: Set[Union[int, str]]) -> str:
+        """
+        pre: "foobar" in s
+        post: (_ + "bar") in s
+        """
+        return "foo"
+
+    check_states(f, CANNOT_CONFIRM)
+
+
+# The heaprefs + deferred set assumptions make this too expensive.
+# TODO: Optimize & re-enable
+def TODO_set_test_subtype_union() -> None:
+    def f(s: Set[Union[int, str]]) -> Set[Union[int, str]]:
+        """post: not ((42 in s) and ('42' in s))"""
+        return s
+
+    check_states(f, MessageType.POST_FAIL)
+
+
+def test_set_subset_compare_ok() -> None:
+    # a >= b with {'a': {0.0, 1.0}, 'b': {2.0}}
+    def f(s1: Set[int], s2: Set[int]) -> bool:
+        """
+        pre: s1 == {0, 1}
+        pre: s2 == {2}
+        post: not _
+        """
+        return s1 >= s2
+
+    check_states(f, CONFIRMED)
+
+
+def test_set_numeric_promotion() -> None:
+    def f(b: bool, s: Set[int]) -> bool:
+        """
+        pre: b == True
+        pre: s == {1}
+        post: _
+        """
+        return b in s
+
+    check_states(f, CONFIRMED)
+
+
+def test_set_runtime_type_ok() -> None:
+    def f(s: set) -> bool:
+        """post: _"""
+        return True
+
+    check_states(f, CONFIRMED)
+
+
+def test_set_isinstance_check() -> None:
+    def f(s: Set[object]) -> bool:
+        """post: _"""
+        return isinstance(s, set)
+
+    check_states(f, CONFIRMED)
+
+
+def test_frozenset___eq__(space):
+    fs = proxy_for_type(Set[FrozenSet[int]], "fs")
+    with ResumedTracing():
+        space.add(len(fs) == 1)
+        space.add(len(next(iter(fs))) == 0)
+        linearset = set([frozenset()])
+        iseq = realize(fs == linearset)
+    assert iseq is True
+
+
+def test_set___eq__() -> None:
+    def f(a: Set[FrozenSet[int]]) -> object:
+        """
+        pre: a == {frozenset({7}), frozenset({42})}
+        post: _ in ('{frozenset({7}), frozenset({42})}', '{frozenset({42}), frozenset({7})}')
+        """
+        return repr(a)
+
+    check_states(
+        f,
+        MessageType.CONFIRMED,
+        AnalysisOptionSet(per_path_timeout=10, per_condition_timeout=10),
+    )
+
+
+def test_frozenset___repr__symbolic_in_concrete(space) -> None:
+    x = proxy_for_type(int, "x")
+    with ResumedTracing():
+        space.add(x == 4)  # type: ignore
+        container = frozenset([x])
+        assert repr(container) == "frozenset({4})"
+
+
+def test_set___repr__symbolic_in_concrete(space) -> None:
+    x = proxy_for_type(int, "x")
+    with ResumedTracing():
+        space.add(x == 4)  # type: ignore
+        container = {x}
+        assert repr(container) == "{4}"
+
+
+def test_set_copy(space):
+    x = proxy_for_type(Set[int], "x")
+    with ResumedTracing():
+        space.add(len(x) == 1)
+        (y, z) = copy.deepcopy((x, x))
+        assert not space.is_possible(len(y) != 1)
+        assert not space.is_possible(list(x)[0] != list(y)[0])
+
+
+def test_set_no_duplicates() -> None:
+    def f(s: Set[int]) -> int:
+        """
+        pre: len(s) == 2
+        post: _
+        """
+        i = iter(s)
+        x = next(i)
+        y = next(i)
+        return x != y
+
+    check_states(f, CONFIRMED)
+
+
+def test_set_additions_to_concrete() -> None:
+    class F:
+        set_type = set
+
+    def f(self: F, x: int) -> Set[int]:
+        """
+        post: _ != set([1234])
+        """
+        return self.set_type([x, x])
+
+    check_states(f, POST_FAIL)
+
+
+def test_frozenset_realize():
+    with standalone_statespace as space:
+        with NoTracing():
+            x = proxy_for_type(FrozenSet[int], "x")
+            y = realize(x)
+            assert type(y) is frozenset
+        assert type(x) is frozenset
+
+
+def test_frozenset_covariance():
+    with standalone_statespace as space:
+        frozen_set = proxy_for_type(FrozenSet[AbstractBase], "x")
+        space.add(frozen_set.__len__() == 1)
+        with NoTracing():
+            assert isinstance(next(iter(frozen_set)), ConcreteSubclass)
+
+
+def test_set_invariance():
+    # `set` is invariant, but that doesn't mean it cannot contain subclasses -
+    # it just means that something *typed* as a subclass container cannot be
+    # assigned to a superclass container.
+    # Therefore, CrossHair should happily create set contents using subclasses:
+    with standalone_statespace as space:
+        mutable_set = proxy_for_type(Set[AbstractBase], "x")
+        space.add(mutable_set.__len__() == 1)
+        with NoTracing():
+            assert isinstance(next(iter(mutable_set)), ConcreteSubclass)
+
+
+def test_set_realize():
+    with standalone_statespace as space:
+        with NoTracing():
+            x = proxy_for_type(Set[str], "x")
+            assert type(x) is not set
+            y = realize(x)
+            assert type(y) is set
+        assert type(x) is set
+
+
+def test_set_iter_partial():
+    with standalone_statespace as space:
+        with NoTracing():
+            x = proxy_for_type(Set[int], "x")
+        space.add(x.__len__() == 2)
+        itr = iter(x)
+        first = next(itr)
+        # leave the iterator incomplete; looking for generator + context mgr problems
+
+
+def test_set_containment_check_without_iteration():
+    with standalone_statespace as space:
+        x = proxy_for_type(FrozenSet[int], "x")
+        with ResumedTracing():
+            space.add(len(x) == 1)
+            assert space.is_possible(x._is_consistent())
+            space.add(x.__contains__(42))
+            assert space.is_possible(x._is_consistent())
+            space.add(x.__contains__(10))
+            assert not space.is_possible(x._is_consistent())
+
+
+def test_set_independence(space):
+    with ResumedTracing():
+        s = set()
+        copy = set(s)
+        s |= s
+        s != copy
+
+
+def test_frozenset___or__(space):
+    x = proxy_for_type(int, "x")
+    y = proxy_for_type(int, "y")
+    with ResumedTracing():
+        space.add(x != y)
+        s1 = frozenset([x])
+        s2 = frozenset([y])
+        assert len(s1 | s2) == 2
+
+
+class TestProtocols:
+    # TODO: move most of this into a collectionslib_test.py file
+    def test_hashable_values_fail(self) -> None:
+        def f(b: bool, i: int, t: Tuple[str, ...]) -> int:
+            """post: _ % 5 != 0"""
+            return hash((b, i, t))
+
+        check_states(f, POST_FAIL)
+
+    def test_hashable_values_ok(self) -> None:
+        def f(a: Tuple[str, int, float, bool], b: Tuple[str, int, float, bool]) -> int:
+            """post: _ or not (a == b)"""
+            return hash(a) == hash(b)
+
+        check_states(f, CANNOT_CONFIRM)
+
+    def test_symbolic_hashable(self) -> None:
+        def f(a: Hashable) -> int:
+            """post[]: 0 <= _ <= 1"""
+            return hash(a) % 2
+
+        check_states(f, CONFIRMED)
+
+    def test_symbolic_supports(self) -> None:
+        def f(
+            a: SupportsAbs,
+            f: SupportsFloat,
+            i: SupportsInt,
+            r: SupportsRound,
+            # c: SupportsComplex,  # TODO: symbolic complex not yet really working
+            b: SupportsBytes,
+        ) -> float:
+            """
+            pre: math.isfinite(f) and math.isfinite(r)
+            post: _.real <= 0
+            """
+            return abs(a) + float(f) + int(i) + round(r) + len(bytes(b))
+            # + complex(c)
+
+        check_states(f, POST_FAIL)
+
+    def test_iterable(self) -> None:
+        T = TypeVar("T")
+
+        def f(a: Iterable[T]) -> T:
+            """
+            pre: a
+            post: _ in a
+            """
+            return next(iter(a))
+
+        check_states(f, CANNOT_CONFIRM)
+
+    def test_bare_type(self) -> None:
+        def f(a: List) -> bool:
+            """
+            pre: a
+            post: _
+            """
+            return bool(a)
+
+        check_states(f, CONFIRMED)
+
+
+def test_enum_identity_matches_equality() -> None:
+    def f(color1: Color, color2: Color) -> bool:
+        """post: _ == (color1 is color2)"""
+        return color1 == color2
+
+    check_states(f, CONFIRMED)
+
+
+def test_enum_in_container() -> None:
+    def f(colors: List[Color]) -> bool:
+        """post: not _"""
+        return Color.RED in colors and Color.BLUE in colors
+
+    check_states(f, POST_FAIL)
+
+
+def test_type_issubclass_ok() -> None:
+    def f(typ: Type[SmokeDetector]):
+        """post: _"""
+        return issubclass(typ, SmokeDetector)
+
+    check_states(f, CONFIRMED)
+
+
+def test_type_can_be_a_subclass() -> None:
+    def f(typ: Type[Cat]):
+        """post: _ == "<class '__main__.Cat'>" """
+        return str(typ)
+
+    # False when the type is instantiated as "BiggerCat":
+    check_states(f, POST_FAIL)
+
+
+def test_type_issubclass_fail() -> None:
+    def f(typ: Type):
+        """post: _"""
+        return issubclass(typ, str)
+
+    check_states(f, POST_FAIL)
+
+
+def test_type_symbolics_without_literal_types() -> None:
+    def f(typ1: Type, typ2: Type[bool], typ3: Type):
+        """post: implies(_, issubclass(typ1, typ3))"""
+        # The counterexample we expect: typ1==str typ2==bool typ3==int
+        return issubclass(typ2, typ3) and typ2 != typ3
+
+    check_states(
+        f,
+        POST_FAIL,
+    )
+
+
+def test_type_instance_creation() -> None:
+    def f(t: Type[Cat]):
+        """post: _.size() > 0"""
+        return t()
+
+    check_states(f, CONFIRMED)
+
+
+def test_type_comparison() -> None:
+    def f(t: Type) -> bool:
+        """post: _"""
+        return t == int
+
+    check_states(f, POST_FAIL)
+
+
+def test_type_as_bool() -> None:
+    def f(t: Type) -> bool:
+        """post: _"""
+        return bool(t)
+
+    check_states(f, CONFIRMED)
+
+
+def test_type_generic_object_and_type() -> None:
+    def f(thing: object, detector_kind: Type[SmokeDetector]):
+        """post: True"""
+        if isinstance(thing, detector_kind):
+            return thing._is_plugged_in
+        return False
+
+    check_states(f, CANNOT_CONFIRM)
+
+
+def test_object___eq__() -> None:
+    def f(thing: object, i: int):
+        """post: not _"""
+        return thing == i
+
+    check_states(f, POST_FAIL)
+
+
+def test_object_get_type_hints(space: StateSpace) -> None:
+    typ = proxy_for_type(type, "typ")
+    with ResumedTracing():
+        get_type_hints(typ)
+
+
+def test_issubclass_abc():
+    with standalone_statespace as space:
+        with NoTracing():
+            dict_subtype = SymbolicType("dict_subtype", Type[dict])
+        issub = issubclass(dict_subtype, collections.abc.Mapping)
+        with NoTracing():
+            # `issub` is lazily determined:
+            assert type(issub) is SymbolicBool
+            assert space.is_possible(issub.var)
+            assert space.is_possible(z3.Not(issub.var))
+            # We can artificially assert that this dict type is somehow not a Mapping:
+            space.add(z3.Not(issub.var))
+            # And CrossHair will give up when it comes time to find some such a type:
+            with pytest.raises(IgnoreAttempt):
+                realize(dict_subtype)
+
+
+def test_object_with_comparison():
+    def f(obj):
+        """post: _"""
+        return obj != b"abc"
+
+    check_states(f, POST_FAIL)
+
+
+def test_object_addition():
+    def f(a, b):
+        """post: True"""
+        return a + b
+
+    actual, expected = check_messages(
+        analyze_function(f),
+        state=MessageType.EXEC_ERR,
+        # We check the message here, because we used to produce an
+        # (incorrect) counterexample of f('', '')
+        # See https://github.com/pschanely/CrossHair/issues/235
+        message="TypeError:  when calling f('', 0)",
+    )
+    assert actual == expected
+
+
+def test_invoke_integer():
+    # https://github.com/pschanely/CrossHair/issues/236
+    def f(a: int):
+        """post: True"""
+        return a(1)  # type: ignore
+
+    check_states(f, EXEC_ERR)
+
+
+def test_callable_zero_args() -> None:
+    def f(size: int, initializer: Callable[[], int]) -> Tuple[int, ...]:
+        """
+        pre: size >= 1
+        post: _[0] != 707
+        """
+        return tuple(initializer() for _ in range(size))
+
+    check_states(f, POST_FAIL)
+
+
+@pytest.mark.smoke
+def test_callable_one_arg() -> None:
+    def f(size: int, mapfn: Callable[[int], int]) -> Tuple[int, ...]:
+        """
+        pre: size >= 1
+        post: _[0] != 707
+        """
+        return tuple(mapfn(i) for i in range(size))
+
+    check_states(f, POST_FAIL)
+
+
+def test_callable_two_args() -> None:
+    def f(i: int, c: Callable[[int, int], int]) -> int:
+        """post: _ != i"""
+        return c(i, i)
+
+    check_states(f, POST_FAIL)
+
+
+def test_callable_as_bool() -> None:
+    def f(fn: Callable[[int], int]) -> bool:
+        """post: _"""
+        return bool(fn)
+
+    check_states(f, CONFIRMED)
+
+
+def test_callable_can_return_different_values(space) -> None:
+    fn = proxy_for_type(Callable[[], int], "fn")
+    with ResumedTracing():
+        first_return = fn()
+        second_return = fn()
+        returns_are_equal = first_return == second_return
+        returns_are_not_equal = first_return != second_return
+    assert space.is_possible(returns_are_equal)
+    assert space.is_possible(returns_are_not_equal)
+
+
+@pytest.mark.smoke
+def test_callable_repr() -> None:
+    def f(f1: Callable[[int], int]) -> int:
+        """post: _ != 1234"""
+        return f1(4)
+
+    messages = run_checkables(analyze_function(f))
+    assert len(messages) == 1
+    assert re.compile(
+        r"false when calling f\(.*lambda.*\(which returns 1234\)"
+    ).fullmatch(messages[0].message)
+
+
+def test_callable_with_typevar_in_args() -> None:
+    T = TypeVar("T")
+
+    def f(a: Callable[[T], int], x: T) -> int:
+        """post: _ != 42"""
+        return a(x)
+
+    check_states(f, POST_FAIL)
+
+
+def test_callable_with_typevar_in_return() -> None:
+    T = TypeVar("T")
+
+    def f(a: Callable[[int], T], x: int) -> T:
+        """post: _"""
+        return a(x)
+
+    check_states(f, POST_FAIL)
+
+
+def TODO_test_callable_with_typevars() -> None:
+    # Right now, this incorrectly reports a counterexample like:
+    # a=`lambda x : 42` and k=''
+    # (the type vars preclude such a counterexample)
+    # Note also a related issue: https://github.com/pschanely/CrossHair/issues/85
+    T = TypeVar("T")
+
+    def f(a: Callable[[T], T], k: T) -> T:
+        """post: _ != 42"""
+        if isinstance(k, int):
+            return 0  # type: ignore
+        return a(k)
+
+    check_states(f, CANNOT_CONFIRM)  # or maybe CONFIRMED?
+
+
+def test_all():
+    class ExplodingBool(BaseException):
+        def __bool__(self):
+            raise self
+
+    with standalone_statespace as space:
+        true_bool = proxy_for_type(bool, "true_bool")
+        space.add(true_bool)
+        nonempty_list = proxy_for_type(List[object], "nonempty_list")
+        space.add(len(nonempty_list) == 1)
+        arbitrary_bool = proxy_for_type(bool, "arbitrary_bool")
+        boom = ExplodingBool()
+
+        assert all([true_bool, True, nonempty_list, 42]) is True
+
+        sym_false = all([True, true_bool, nonempty_list, arbitrary_bool])
+        with NoTracing():
+            assert isinstance(sym_false, SymbolicBool)
+            assert space.is_possible(z3.Not(sym_false.var))
+        assert space.is_possible(sym_false)
+
+        with pytest.raises(ExplodingBool):
+            all([true_bool, True, nonempty_list, boom])
+
+        assert all([true_bool, nonempty_list, False, boom]) is False
+
+
+def test_any():
+    class ExplodingBool(BaseException):
+        def __bool__(self):
+            raise self
+
+    with standalone_statespace as space:
+        false_bool = proxy_for_type(bool, "false_bool")
+        space.add(z3.Not(false_bool.var))
+        empty_list = proxy_for_type(List[object], "empty_list")
+        space.add(len(empty_list) == 0)
+        arbitrary_bool = proxy_for_type(bool, "arbitrary_bool")
+        # arbitrary_list = proxy_for_type(List[object], "arbitrary_list")
+        # TODO: in theory, we should be able to keep `bool(arbitrary_list)`
+        # symbolic. However, not enough of the system is resilient to symbolic
+        # booleans. (at the very least, the UNARY_NOT opcode)
+        boom = ExplodingBool()
+
+        assert any([false_bool, False, empty_list, 0]) is False
+
+        sym_false = any([False, false_bool, empty_list, arbitrary_bool])
+        with NoTracing():
+            assert isinstance(sym_false, SymbolicBool)
+            assert space.is_possible(z3.Not(sym_false.var))
+        assert space.is_possible(sym_false)
+
+        with pytest.raises(ExplodingBool):
+            any([false_bool, False, empty_list, boom])
+
+        assert any([false_bool, empty_list, True, boom]) is True
+
+
+def test_hash() -> None:
+    def f(s: int) -> int:
+        """post: True"""
+        return hash(s)
+
+    check_states(f, CONFIRMED)
+
+
+@pytest.mark.demo
+def test_getattr() -> None:
+    def f(s: str) -> int:
+        """
+        Can this function return 42?
+
+        post: _ != 42
+        """
+
+        class Otter:
+            def do_things(self) -> int:
+                return 42
+
+        try:
+            return getattr(Otter(), s)()
+        except Exception:
+            return 0
+
+    check_states(f, POST_FAIL)
+
+
+def TODO_test_print_ok() -> None:
+    def f(x: int) -> bool:
+        """
+        post: _ == True
+        """
+        print(x)
+        return True
+
+    check_states(f, CONFIRMED)
+
+
+def test_repr_ok():
+    def f(x: int) -> str:
+        """post: len(_) == 0 or len(_) > 0"""
+        return repr(x)
+
+    check_states(f, CONFIRMED)
+
+
+@pytest.mark.demo
+def test_map() -> None:
+    def f(ls: List[int]) -> List[int]:
+        """
+        Can an incremented list equal [4, 9, 0]?
+
+        post: _ != [4, 9, 0]
+        """
+        return list(map(lambda x: x + 1, ls))
+
+    check_states(f, POST_FAIL)
+
+
+def test_max_fail() -> None:
+    def f(ls: List[int]) -> int:
+        """
+        post: _ in ls
+        """
+        return max(ls)
+
+    check_states(f, EXEC_ERR)
+
+
+def test_max_ok() -> None:
+    def f(ls: List[int]) -> int:
+        """
+        pre: bool(ls)
+        post[]: _ in ls
+        """
+        return max(ls)
+
+    check_states(f, CANNOT_CONFIRM)
+
+
+def test_min_ok() -> None:
+    def f(ls: List[float]) -> float:
+        """
+        pre: bool(ls)
+        post[]: _ in ls
+        """
+        return min(ls)
+
+    check_states(f, CANNOT_CONFIRM)
+
+
+def test_list_index_on_concrete() -> None:
+    def f(i: int) -> int:
+        """post: True"""
+        return [0, 1, 2].index(i)
+
+    (actual, expected) = check_exec_err(f, "ValueError:")
+    assert actual == expected
+
+
+def test_eval_namespaces() -> None:
+    def f(i: int) -> int:
+        """post: _ == i + 1"""
+        return eval("i + Color.BLUE.value")
+
+    check_states(f, CONFIRMED)
+
+
+def test_bytes_specific_length() -> None:
+    def f(b: bytes) -> int:
+        """post: _ != 5"""
+        return len(b)
+
+    check_states(f, POST_FAIL)
+
+
+def test_bytes_out_of_range_byte() -> None:
+    def f(b: bytes) -> bytes:
+        """
+        pre: len(b) == 1
+        post: _[0] != 256
+        """
+        return b
+
+    check_states(f, CONFIRMED)
+
+
+def test_bytes_roundtrip_array_as_symbolic():
+    with standalone_statespace as space:
+        orig_bytes = proxy_for_type(bytes, "origbytes")
+        as_array = bytearray(orig_bytes)
+        new_bytes = bytes(as_array)
+        with NoTracing():
+            assert type(as_array) is SymbolicByteArray
+            assert type(new_bytes) is SymbolicBytes
+            assert new_bytes.inner is orig_bytes.inner
+
+
+@pytest.mark.skipif(
+    sys.version_info >= (3, 13),
+    reason="Need to intercept UnicodeDecodeError.str in 3.13+",
+)
+@pytest.mark.demo
+def test_bytes_decode_method():
+    def f(b: bytes) -> str:
+        """
+        Does any 2-byte sequence represent the character "ε" in UTF-8?
+
+        NOTE: The process of decoding involves a lot of branching;
+        most problems will require minutes of processing or more.
+
+        pre: len(b) == 2
+        post: _ != "ε"
+        raises: UnicodeDecodeError
+        """
+        return b.decode("utf8")
+
+    check_states(f, POST_FAIL)
+
+
+@pytest.mark.demo("red")
+def test_bytes___str___method():
+    def f(b: bytes):
+        """
+        Is the string form of any byte array equal to b''?
+
+        NOTE: This conversion does not have symbolic support (yet). We are able to find
+        the enpty string, but nearly any other bytes string cannot be found.
+
+        post: _ != "b''"
+        """
+        return str(b)
+
+    check_states(f, POST_FAIL)
+
+
+@pytest.mark.demo
+def test_bytearray___add___method():
+    def f(a: bytes, count: int) -> bytearray:
+        """
+        Can some repetitions of a 3-character byte string produce "abcabc"?
+        pre: len(a) == 3
+        post: _ != b"abcabc"
+        """
+        ba = bytearray()
+        for _ in range(count):
+            ba += a
+        return ba
+
+    check_states(f, POST_FAIL)
+
+
+def test_extend_concrete_bytearray():
+    with standalone_statespace as space:
+        b = bytearray(b"abc")
+        xyz = proxy_for_type(bytearray, "xyz")
+        b.extend(xyz)
+        assert not space.is_possible(b[0] != ord("a"))
+        assert space.is_possible(len(b) > 3)
+
+
+def test_bytearray_slice():
+    with standalone_statespace as space:
+        xyz = proxy_for_type(bytearray, "xyz")
+        space.add(xyz.__len__() == 3)
+        assert type(xyz[1:]) is bytearray
+
+
+def test_memoryview_compare():
+    with standalone_statespace as space:
+        mv1 = proxy_for_type(memoryview, "mv1")
+        mv2 = proxy_for_type(memoryview, "mv2")
+        len1, len2 = len(mv1), len(mv2)
+        space.add(len1 == 0)
+        space.add(len2 == 0)
+        views_equal = mv1 == mv2
+        with NoTracing():
+            assert views_equal is True
+
+
+def test_memoryview_cast():
+    """post: _"""
+    with standalone_statespace as space:
+        val = proxy_for_type(int, "val")
+        space.add(val == 254)
+        mv = memoryview(bytearray([val]))
+        assert mv.cast("b")[0] == -2
+
+
+def test_memoryview_toreadonly():
+    """post: _"""
+    with standalone_statespace as space:
+        mv = proxy_for_type(memoryview, "mv")
+        space.add(mv.__len__() == 1)
+        mv2 = mv.toreadonly()
+        mv[0] = 12
+        assert mv2[0] == 12
+        with pytest.raises(TypeError):
+            mv2[0] = 24
+
+
+def test_memoryview_properties():
+    """post: _"""
+    with standalone_statespace as space:
+        symbolic_mv = proxy_for_type(memoryview, "symbolic_mv")
+        space.add(symbolic_mv.__len__() == 1)
+        concrete_mv = memoryview(bytearray(b"a"))
+        assert symbolic_mv.contiguous == concrete_mv.contiguous
+        assert symbolic_mv.c_contiguous == concrete_mv.c_contiguous
+        assert symbolic_mv.f_contiguous == concrete_mv.f_contiguous
+        assert symbolic_mv.readonly == concrete_mv.readonly
+        assert symbolic_mv.format == concrete_mv.format
+        assert symbolic_mv.itemsize == concrete_mv.itemsize
+        assert symbolic_mv.nbytes == concrete_mv.nbytes
+        assert symbolic_mv.ndim == concrete_mv.ndim
+        assert symbolic_mv.shape == concrete_mv.shape
+        assert symbolic_mv.strides == concrete_mv.strides
+        assert symbolic_mv.suboffsets == concrete_mv.suboffsets
+
+
+def test_chr(space):
+    i = proxy_for_type(int, "i")
+    with ResumedTracing():
+        space.add(10 <= i)
+        space.add(i < 256)
+        c = chr(i)
+        assert space.is_possible(c._codepoints[0] == ord("a"))
+        assert not space.is_possible(c._codepoints[0] == 0)
+
+
+def test_ord(space):
+    s = proxy_for_type(str, "s")
+    with ResumedTracing():
+        space.add(len(s) == 1)
+        i = ord(s)
+        assert space.is_possible(i == 42)
+        assert not space.is_possible(i > sys.maxunicode)
+
+
+def test_unicode_support(space):
+    s = proxy_for_type(str, "s")
+    with ResumedTracing():
+        space.add(len(s) == 1)
+        assert space.is_possible(s == "a")
+        assert space.is_possible(s == "\u1234")
+
+
+@pytest.mark.parametrize("concrete_x", (25, 15, 6, -4, -15, -25))
+def test_int_round(concrete_x, space):
+    concrete_ret = round(concrete_x, -1)
+    x = proxy_for_type(int, "x")
+    d = proxy_for_type(int, "d")
+    with ResumedTracing():
+        space.add(x == concrete_x)
+        space.add(d == -1)
+        assert not space.is_possible((round(x, d) != concrete_ret))
+
+
+class ExplodingValue:
+    def __getattribute__(self, name):
+        raise CrossHairInternal
+
+
+@dataclasses.dataclass
+class ClassWithSpecialMemberNames:
+    typ: str
+    var: int
+
+
+def test_class_with_special_member_names():
+    exploding_value = ExplodingValue()
+
+    def f(obj1: ClassWithSpecialMemberNames):
+        """post: True"""
+        assert isinstance(obj1.typ, str)
+        assert isinstance(obj1.var, int)
+        # Ensure run-time creation doesn't explode either:
+        ClassWithSpecialMemberNames(typ=exploding_value, var=exploding_value)  # type: ignore
+
+    check_states(f, CONFIRMED)
+
+
+@pytest.mark.parametrize("type2", (list, set, frozenset, tuple))
+@pytest.mark.parametrize(
+    "type1",
+    (
+        list,
+        set,
+        frozenset,
+        tuple,
+        array,
+    ),
+)
+def test_container_crosstype_equality(space, type1, type2):
+    sym_obj1 = proxy_for_type(type1, "obj1")
+    obj2 = type2()
+    with ResumedTracing():
+        space.add(len(sym_obj1) == 0)
+        obj1 = deep_realize(sym_obj1)
+        sym_eq_on_left = realize(sym_obj1 == obj2)
+        sym_eq_on_right = realize(obj2 == sym_obj1)
+    actually_equal = obj1 == obj2
+    assert actually_equal == sym_eq_on_left
+    assert actually_equal == sym_eq_on_right
+
+
+@pytest.mark.parametrize("type2", (str, bytes, tuple))
+@pytest.mark.parametrize("type1", (str, bytes, tuple, memoryview))
+def test_stringlike_crosstype_equality(space, type1, type2):
+    sym_obj1 = proxy_for_type(type1, "obj1")
+    obj2 = type2()
+    with ResumedTracing():
+        space.add(len(sym_obj1) == 0)
+        obj1 = deep_realize(sym_obj1)
+        sym_eq_on_left = realize(sym_obj1 == obj2)
+        sym_eq_on_right = realize(obj2 == sym_obj1)
+    actually_equal = obj1 == obj2
+    assert actually_equal == sym_eq_on_left
+    assert actually_equal == sym_eq_on_right
+
+
+@pytest.mark.parametrize("typ", (List[int],))
+def test_deep_realization(space, typ):
+    symbolic = proxy_for_type(typ, "symbolic")
+    with ResumedTracing():
+        space.add(len(symbolic) == 3)
+    origin = origin_of(typ)
+    assert type(symbolic) != origin
+    concrete = deep_realize(symbolic)
+    assert type(concrete) == origin
+    with ResumedTracing():
+        assert concrete == symbolic
+
+
+def test_float_round_to_zero(space):
+    space.extra(ModelingDirector).global_representations[
+        float
+    ] = PreciseIeeeSymbolicFloat
+    n = proxy_for_type(float, "n")
+    d = proxy_for_type(float, "d")
+    with ResumedTracing():
+        space.add(d != 0.0)
+        # This is possible for floats, but not reals:
+        assert space.is_possible(n / d != 0.0)
+        assert space.is_possible(n / d == 0.0)
+
+
+def test_float_neg_zero_is_falsey(space):
+    space.extra(ModelingDirector).global_representations[
+        float
+    ] = PreciseIeeeSymbolicFloat
+    x = proxy_for_type(float, "x")
+    with ResumedTracing():
+        space.add(x == -10.0)
+        negzero = x / math.inf
+        assert not space.is_possible(negzero != 0.0)
+        assert bool(negzero) is False
+
+
+def TODO_test_int_mod_float():
+    # (errors at the Z3 level presently)
+    with standalone_statespace as space:
+        x = proxy_for_type(int, "x")
+        y = proxy_for_type(float, "y")
+        modval = x % y
+        with NoTracing():
+            assert type(modval) == RealBasedSymbolicFloat
+        assert space.is_possible(modval == 12.12)
+
+
+def TODO_test_can_generate_integral():
+    with standalone_statespace as space:
+        proxy_for_type(Integral, "x")
+
+
+def TODO_test_deepcopy_independence():
+    # (snapshotting should take care of this for us(?), but it doesn't seem to)
+    with standalone_statespace as space:
+        ls = proxy_for_type(List[List[int]], "ls")
+        lscopy = copy.deepcopy(ls)
+        with NoTracing():
+            assert ls[0] is not lscopy[0]
+        # Next try mutation on one and test the other...