On 28 Apr 2008, at 18:41, Stephen J. Turnbull wrote:
John Paul Wallington writes:
> On 27 Apr 2008, at 23:59, Stephen J. Turnbull wrote:
>
>> D'oh. It *is* redundant, isn't it. Unless somebody comes up with a
>> good reason not to, I'm just going to get rid of the const nil and
>> parent choice widget, too.
>
> FWIW, although it's strictly redundant I think offering the choice
> makes for a better user experience and would leave Alan's patch as-
> is.
I tend to disagree because I think of this as an interface to a (Lisp)
list, and because it makes customizing this variable slower than it
need be. Can you say why having the "none" option works for you?
I can't really put my finger on why. It seems more explicit to choose
between a labelled none/nil/empty list option versus the repeat
widget. But having played more with Customize I don't think it matters
much.
My point is that consistency is presumably a good thing, and if
having
a choice of "none" is good for this list-that-might-be-empty, wouldn't
it be a better idea to have that built-in to the `repeat' widget?
If we had a clean slate then I think that would be a better idea.
However, there's probably tonnes of existing code with choices much
like the `Manual-switches' variable or variations thereof such that
unconditionally building an explicit nil/empty list choice into the
`repeat' widget would result in an ugly redundant interface whilst
working out when to conditionally display it could be too much effort.
_______________________________________________
XEmacs-Beta mailing list
XEmacs-Beta(a)xemacs.org
http://calypso.tux.org/cgi-bin/mailman/listinfo/xemacs-beta