pschanely
Crash when integer has (way) too many digits
This is a horrible regression test for a case where Hypothesis wasn't handling large integers (https://github.com/HypothesisWorks/hypothesis/issues/3874), but I presume you'll want to do something which isn't crashing. And it looks like we have two meaningfully distinct issues in the same traceback! Relevant stdlib docs here; converting really big ints to strings is expensive if the base is not a power of two.
@given(st.integers(min_value=0, max_value=1 << 25_000)) # I'm so sorry about this...
def test_overrun_during_datatree_simulation_3874(n):
pass
Traceback (most recent call last):
File ".../hypothesis/internal/conjecture/data.py", line 2450, in draw
return strategy.do_draw(self)
File ".../hypothesis/strategies/_internal/lazy.py", line 167, in do_draw
return data.draw(self.wrapped_strategy)
File ".../hypothesis/internal/conjecture/data.py", line 2444, in draw
return strategy.do_draw(self)
File ".../hypothesis/strategies/_internal/numbers.py", line 85, in do_draw
return data.draw_integer(
File ".../hypothesis/internal/conjecture/data.py", line 2113, in draw_integer
value = self.provider.draw_integer(
File ".../hypothesis_crosshair_provider/crosshair_provider.py", line 195, in draw_integer
conditions.append(symbolic <= max_value)
File ".../crosshair/libimpl/builtinslib.py", line 910, in __le__
return numeric_binop(ops.le, self, other)
File ".../crosshair/libimpl/builtinslib.py", line 478, in numeric_binop
return numeric_binop_internal(op, a, b)
File ".../crosshair/libimpl/builtinslib.py", line 498, in numeric_binop_internal
return binfn(op, a, b)
File ".../crosshair/libimpl/builtinslib.py", line 741, in _
return SymbolicBool(apply_smt(op, a.var, z3IntVal(b)))
File ".../crosshair/z3util.py", line 43, in z3IntVal
return IntNumRef(Z3_mk_numeral(ctx_ref, x.__index__().__str__(), int_sort_ast), ctx)
ValueError: Exceeds the limit (4300) for integer string conversion; use sys.set_int_max_str_digits() to increase the limit
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
File ".../hypothesis/internal/conjecture/data.py", line 2457, in draw
add_note(err, f"while generating {key[9:]!r} from {strategy!r}")
File ".../crosshair/opcode_intercept.py", line 141, in __repr__
self.formatted = repr(self.value)
File ".../crosshair/libimpl/builtinslib.py", line 4521, in _repr
return invoke_dunder(obj, "__repr__")
File ".../crosshair/libimpl/builtinslib.py", line 254, in invoke_dunder
return method(obj, *args, **kwargs)
File ".../hypothesis/strategies/_internal/lazy.py", line 158, in __repr__
self.__representation = repr_call(
File ".../hypothesis/internal/reflection.py", line 496, in repr_call
bits.append(f"{p.name}={nicerepr(kwargs.pop(p.name))}")
File ".../hypothesis/internal/reflection.py", line 484, in nicerepr
return re.sub(r"(\[)~([A-Z][a-z]*\])", r"\g<1>\g<2>", pretty(v))
File ".../hypothesis/vendor/pretty.py", line 101, in pretty
printer.pretty(obj)
File ".../hypothesis/vendor/pretty.py", line 197, in pretty
return self.type_pprinters[cls](obj, self, cycle)
File ".../hypothesis/vendor/pretty.py", line 722, in _repr_pprint
output = repr(obj)
File ".../crosshair/libimpl/builtinslib.py", line 4521, in _repr
return invoke_dunder(obj, "__repr__")
File ".../crosshair/libimpl/builtinslib.py", line 254, in invoke_dunder
return method(obj, *args, **kwargs)
File ".../crosshair/core.py", line 333, in with_checked_self
return native_method(self, *a, **kw)
ValueError: Exceeds the limit (4300) for integer string conversion; use sys.set_int_max_str_digits() to increase the limit
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
File ".../hypothesis/internal/conjecture/data.py", line -1, in draw
File ".../crosshair/opcode_intercept.py", line 262, in post_op
frame_stack_write(frame, -1, wrapper.formatted)
AttributeError: 'FormatStashingValue' object has no attribute 'formatted'
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
File ".../crosshair/opcode_intercept.py", line 262, in post_op
frame_stack_write(frame, -1, wrapper.formatted)
AttributeError: 'FormatStashingValue' object has no attribute 'formatted'
(via https://github.com/HypothesisWorks/hypothesis/pull/4034)
Oh wow. OOC, from a product perspective, what does hypothesis do about this? st.integers is by default unbounded, and stringification is common (logging etc). But most users probably don't care to think about this kind of failure case. I assume it takes hypothesis so long to attempt such an integer that it doesn't matter?
Also, does hypothesis do something to avoid hitting this limit internally? Someone might want to make a test to check how this limitation affects their code, but hypothesis still needs to display the counterexample.
Currently we don't do anything particular to handle this case; unbounded integers only go up to 256-bits by default and that regression test would fail if we happened to convert to a string internally.
Having been reminded that this new error exists, I'm planning to upgrade the pretty-printer to display too-large integers in hexadecimal instead. https://github.com/HypothesisWorks/hypothesis/issues/4058 / https://github.com/HypothesisWorks/hypothesis/pull/4065/files#diff-cdb64915e1b0b4db52f343734ea143e51ccf52a88bfa4175eb1038363444ad2cR742-R751