Documentation of .[ is wrong.

[orlp]

For example it says:

On <any><int><seq> pad left.

Which is just flat out wrong.

[isaacg1]

Fixed by #151

[orlp]

No, no, no, no!

I just realized that Jakob fixed the entirely wrong thing. The reference implementation was (almost) good, the documentation was wrong!

Consider the current (wrong) documentation:

On <seq><any><int> pad right. On <int><seq><any> pad left and right. On <any><int><seq> pad left.

That is incredibly, incredibly inconsistent. The order of <seq> and <any> is inconsistent, and the place of the integer does not match where the padding will occur. I propose the following:

On <seq><any><int> pad right. On <seq><int><any> pad left and right. On <int><seq><any> pad left.

[jakobkogler]

@orlp With your notation you can't distinguish pad right from pad left and right, when <any> is also from type <int>.

My main idea was the following:

• <seq> on the left means, that <seq> stays on the left, so padding happens on the right.
• <seq> in the middle means, that it stays in the middle => padding left and right.
• <seq> on the right, stay on the right => padding happens on the left.

And the other two parameter rotate with it.

[jakobkogler]

And if you don't rotate the parameters, you get problems with <any> = <seq>.

The main reason I changed the behavior instead of the documentation was to make the documentation easier. The behavior was correct for <any> = <int>, but worked differently for <any> = <string>. The documentation would have been twice as long and more complicated.

[orlp]

I apologize, I totally forgot about distinguishing <any> = <seq>