Re: negative values for \hspace don't work anymore

[David Kastrup]

Eluze <address@hidden> writes:

> David Kastrup wrote
>> Eluze &lt;
>
>> eluzew@
>
>> &gt; writes:
>> 
>>> he had to do this because overriding the circle-padding also increases
>>> the
>>> distance of the following object:
>>>
>>> \markup  \center-column { 
>>>   \override #'(circle-padding . 0) \box  \concat { "xxx" \box \circle 3
>>> "xxx" } 
>>>   \override #'(circle-padding . 12) \box  \concat { "xxx" \box \circle 3
>>> "xxx" } 
>>> }
>>>
>>> hspace.png
>>> &lt;http://lilypond.1069038.n5.nabble.com/file/n136649/hspace.png&gt;  
>>>
>>> is this intended? and why?
>> 
>> Well, as I explained \hspace special-cases negative distances by setting
>> _both_ left and right border of the stencil to a negative value rather
>> than just the right one.  So the padding sees the left border and pushes
>> out.
>
> now looking at \box with \pad-to-box:
>
> using a negative value in the first of the horizontal pair of numbers draws
> a box to the left of the boxed object, thus boxing what is already there,
> and putting extra-space to the right - shouldn't the boxed object be treated
> as a new object which is considered when calculating further spacing?
>
> it seems all very confusing and not practically unusable!
>
> shall I raise an issue to clarify all this?

I have the suspicion that this needs something more like fixing rather
than clarification.  Using intervals for stencil extents here, and most
particularly the emptiness check of intervals seems like a mistake.

-- 
David Kastrup