• mjr@infosec.pub
    link
    fedilink
    arrow-up
    1
    ·
    7 hours ago

    Thank for elaborating my comment, but I never said never, only that it’s usually better to avoid it.

    And yven if you think it’s provably impossible to get an Error back now, someone or something may change an underlying function behaviour on you in the future and invalidate your proof. There are ways to limit that with version control and pinning and so on, but it’s easy for an assumption to be overlooked when merging in new versions of things.

    So yes, I agree, better to use ? at least here, but like all guidelines, there may be times where you break it, accepting the risks.