I'm willing to look at that sure, but first I need an idea for a level that can be used to test that. Can't think of anything at the moment.golf wrote:I'm particularly interested to hear for sure exactly how that input buffer works, makes sense. Would you be willing to look at the buffer in the opposite direction, when you jump after running off a surface? I feel like that works a bit differently, where it will still let you jump even when you're not physically on the surface anymore.
The other case is easy to work out from a one tile width chimney. I think it's best explained in the form of a table.
The first row is what I believe to be a perfect chimney. The sequence of a perfect chimney in N++ appears to be jhnjhnjhn... and so on.
The second sequence is a test case where I simply jump the first frame it is possible to do so. So to start off a jnjnj sequence gives a output of three jumps. But then you try to do a fourth jump you find that the game won't accept an input for the fourth jump until frame nine. From this you can conclude that the ninja does not always jump on the same frame as the keyboard input. If it did then you should logically be able to do the fourth jump on frame seven.
I'm not sure about the actual sequence of jumps physically taking place without FBF, but it doesn't seem to match the inputs. The opposing wall can't be magically further away every other jump.