Python random tests are no longer susceptible to input mutation.

I've reproduced those tests in the sample tests section and my code pass them.

Also after a closer look at the random tests it would seem that all the tests with a Money value other than 0 expects a result of 0 so the issue might be somewhere else in the random tests implementation.

I ran the "oracle" in your two examples, and the expected output is 1 for both, not 0. You might be misreading the failure message.

Hi, the oracle for the Python random tests seems broken.

In the random tests I've seen the following results:

That's because there are not sufficient tests, duplicate issue.

The expression ROMAN[k] has an implicit any type since TypeScript currently always treats the variable in a for...in as a string. This is problematic and means you do not have type protection when accessing ROMAN with random string values. The TypeScript issue https://github.com/Microsoft/TypeScript/issues/3500#issuecomment-579360564 is attempting to address this.

This has not been correct since 2015: https://2ality.com/2015/10/property-traversal-order-es6.html
The insertion order for the specific object they created can be relied upon.

The object ROMAN is not guaranteed to be iterated through in the order it was declared according to JS specification, you should instead rely on an array, or just convert using Object.keys(ROMAN).map(k => [k, ROMAN[k]]).sort((a, b) => b[1] - a[1]).