[python] Dataclass kernel argument with a list[int] field raises std::bad_cast when the list contains a negative value

[wsttiger]

Required prerequisites

• Consult the security policy. If reporting a security vulnerability, do not report the bug using this form. Use the process described in the policy to report the issue.
• Make sure you've read the documentation. Your issue may be addressed there.
• Search the issue tracker to verify that this hasn't already been reported. +1 or comment there if it has.

Describe the bug

A @dataclass used as a kernel argument works when its list[int] fields
contain only non-negative values, but launching raises
RuntimeError: std::bad_cast as soon as any list element is negative. The
same list passed as a bare list[int] argument works fine with negative
values, so the defect is specific to the struct-member marshaling path (it
behaves as if elements are cast through an unsigned type during packing).

Observed matrix:

Case	Result
dataclass, scalar fields only	OK
dataclass, list[int]/list[float] fields, non-negative values	OK (tested 1-5 list fields, mixed lengths)
dataclass, any list[int] field containing a negative value	std::bad_cast
bare list[int] argument containing negative values	OK
captured dataclass with negative list element	std::bad_cast

Steps to reproduce the bug

from dataclasses import dataclass
import cudaq

cudaq.set_target("qpp-cpu")

@dataclass(slots=True)
class Args:
    values: list[int]

@cudaq.kernel
def kernel(args: Args):
    q = cudaq.qvector(1)
    if args.values[0] < 0:
        x(q[0])

cudaq.sample(kernel, Args([1]))    # OK
cudaq.sample(kernel, Args([-1]))   # RuntimeError: std::bad_cast

@cudaq.kernel
def bare(values: list[int]):
    q = cudaq.qvector(1)
    if values[0] < 0:
        x(q[0])

cudaq.sample(bare, [-1])           # OK -> '1'

Expected behavior

Negative integers in a dataclass list[int] field marshal identically to
negative integers in a bare list[int] argument.

Is this a regression? If it is, put the last known working version (or commit) here.

Not a regression

Environment

• CUDA-Q built from source at commit 0be565550f4c23affdcbed9e4eaec38d2d0915e6
• Python 3.11.x, target qpp-cpu (library mode)
• AlmaLinux 8.10, x86_64

All reproducers below were executed and verified against this build.