Use wrapping sub to prevent underflow

[casey]

Fix possible underflow when first diff is an addition, see #3.

[casey]

Is wrapping sub the right behavior here? Would it be preferable to skip this test and always do format_add if this is the first diff?

[CAD97]

This patch is easier, as a wrapping_sub will "always" be out of bounds for the get. We could treat the first diff differently, or check for the i=1 case explicitly, but I think treating every case the same is better.

[casey]

I think that's totally reasonable. An alternative is to do:

if let Some(Rem(removed)) = i.checked_sub(1).map(|i| &diffs[i]) {

Which is weird, but maybe slightly more explicit.

[casey]

What do you think about my proposal to use checked_sub? I think it's fine as written too, so please feel free to merge that if you prefer it.

[CAD97]

Actually, yes, checked_sub is the correct way to do it. If you update the patch to use that I can probably push a patch update today or tomorrow.

[casey]

Great, just pushed a diff w/checked_sub instead of wrapping_sub

[casey]

Sweet!