you are viewing a single comment's thread
view the rest of the comments
[–] 10 points 3 weeks ago* (8 children)

(Might be something everyone knows) The Nuclear Ghandi thing was a unsigned integer issue. His "war mongering" value reached zero, and lowering once more made it wrap around to max value and turn him into a lunatic.

  • source
  • parent
  • hideshow 8 child comments
  • [–] 7 points 3 weeks ago* (last edited 3 weeks ago) (4 children)

    This is probably one of the most shared pieces game lore, but it isn't actually true! Sid Meier himself explained in his memoirs.

    Civ 1 only has three aggresion levels, and Gandhi always stayed at the lowest one, like many others leaders. A programmer might have to correct me here, but the language the game was written in (C) doesn't even have unsigned integers by default.

    The myth probably spread out because it's a fun story, and being nuked by Gandhi is just a funny memorable thing that stuck with people. I haven't verified this, but he also has the Scientific trait, which means he's likely to reach the nuclear tech before other Civs.

  • source
  • parent
  • hideshow 4 child comments
  • [–] 2 points 3 weeks ago

    Seem that you are indeed right! Thanks for clearing that up!

    To quote the relevant parts:

    ... It’s the fact that none of it is true. The overflow error never happened at all. It is true that Gandhi would—eventually—use nukes when India was at war, just like any civilization in the game, and at the time this did strike a lot of players as odd. [...] Since all leaders used the same basic diplomacy script [...] But at no point did a democratic score change, or any value approaching 255, come into it. That kind of bug comes from something called unsigned characters, which are [...] not something I used for the leader traits. Brian Reynolds wrote Civ II in C++, and he didn’t use them, either. [...] Gandhi’s military aggressiveness score remained at 1 throughout the game. [...]

  • source
  • parent
  • [–] 1 point 3 weeks ago (2 children)

    Pretty sure C has had unsigned ints from the start, but who knows what data type they actually used. It was the 90s so using small data types where possible was common, aggression could've been a short, a char, etc. Also I think C programmers like using enums and that's a valid use case for enum (low, med, high)

  • source
  • parent
  • hideshow 2 child comments
  • [–] 3 points 3 weeks ago (2 children)
  • [–] 5 points 3 weeks ago (1 child)

    This phenomenon is called integer overflow.

    Underflow, but close enough

  • source
  • parent
  • hideshow 1 child comment
  • [–] 5 points 3 weeks ago*

    I don't think it's wrong to call it integer overflow, as that covers wrap around in either direction. I don't think there is any mention of "integer underflow" in any IEEE standard.

    There is however mention of arithmetic underflow, but that relates to a precision issue with floating point numbers, and not negative wrap around. (See IEEE 754).

    It's still understood what one might mean by integer underflow. But if anything, I'd say it's a further specification of a type of integer overflow, and certainly not a necessary distinction.

  • source
  • parent