Resource limits

Monty enforces hard limits on memory, time, allocations, and recursion to keep untrusted code bounded. When a limit is exceeded, execution terminates with a ResourceError (visible to the host, not catchable inside the sandbox).

Memory / size limits

• Allocation tracking is global; the host sets the bytes budget when constructing the VM.
• Operations whose result is bounded by simple arithmetic on input sizes are pre-checked before allocating: integer multiplication, left shift, integer power, sequence repeat ('x' * n), padding (str.ljust, str.center, str.zfill, bytes.ljust, …), and f-string formatting (both dynamic width f"{v:>{w}}" and dynamic precision on float formats f"{v:.{p}f}" / e / %). The pre-check threshold is 100 KB — anything that would estimate above that is rejected with ResourceError rather than attempting the allocation.
• bigint.pow(base, exp) estimates result size as bits(base) * exp with a 4× safety multiplier to cover repeated-squaring intermediate values.

Integer-specific caps

• pow(base, exp) / base ** exp with an exponent larger than u32::MAX (≈ 4.3 × 10⁹) raises OverflowError: "exponent too large".
• pow(base, exp, mod) requires all integer arguments and rejects negative exponents (ValueError).
• int(str) for very long decimal strings is rejected before the O(n²) BigInt parse runs; the cut-off matches CPython's sys.int_info.str_digits_check_threshold.

Recursion

• Python-level call depth is hardcoded at 1000 frames. The 1001st nested call raises RecursionError.
• There is no sys.getrecursionlimit() / setrecursionlimit() — the limit cannot be changed by sandboxed code.
• Async stacks count toward the limit but each await boundary is treated as one frame, so await-chains do not amplify depth.

Time

• The host can set a wall-clock budget; if exceeded the VM stops on the next bytecode boundary with ResourceError.
• There is no in-sandbox way to observe the budget or remaining time.

JSON

• json.loads rejects input nested deeper than 200 levels with json.JSONDecodeError (independent of the Python recursion limit).

After a ResourceError

When a resource limit fires, no guarantees are made about heap state or reference counts. The host should discard the VM rather than try to recover and continue running code in it.