BUTLAST counts conses before copying whole list

[Johan Andersson]

BUTLAST counts the conses of the argument list before checking if n is zero, in which case the entire list will be copied. The zerop check can be done before counting the conses, as in NBUTLAST. Also, the similar cases of BUTLAST and NBUTLAST can be made clearer.

Regression tests already exist in tests/list.pure.lisp

[Paul Khuong (pvk)]

I'll commit this after the freeze.

Thank you!

[Johan Andersson (nilsjohanandersson)]

You might want to consider this patch instead. It does BUTLAST and NBUTLAST for (< 0 n) without counting the conses beforehand, by looping with two variables down the list N conses apart.

I'm not sure about the style, the names of the variables or if it's worth the fuss. But there you are.

[Paul Khuong (pvk)]

I tried to preserve the logic, but eliminated the counter variable I. It's applied in 498ec57 ([N]BUTLAST perform a single pass over the list).

I also yanked the docstrings out: we try not to duplicate the CLHS.

Thank you!

[Johan Andersson (nilsjohanandersson)]

That's neat! Thanks.