Re: behaviour change in cl-subseq

[phillip.lord@newcastle.ac.uk (Phillip Lord)]

Leo Liu <sdl.web@gmail.com> writes:

> On 2015-08-06 17:34 +0800, Phillip Lord wrote:
>> Fortunately, Tassilo has answered this; I've never used CL.
>
> I answered it in a hurry before going to work :(
>
>> I agree that this change is entirely justifiable. I'm generally not a
>> huge fan of silently ignoring inaccurate arguments. If you are happy
>> with the changed interface, then I have no major problem, although it
>> may break packages besides mine.
>
> The old behaviour was odd and inconsistent (try it on []).
>
>> Do you mind if I patch the docstrings, though?
>
> Not at all.


Was working through this, and found this behaviour.

;; error
(seq-subseq '() 10 10)

;; nil
(seq-subseq '() -10 -10)

So, a bounding box error when the index is too large, but it is silently
ignored when too low.

Not convinced this makes sense.

Phil