Quoting Trevor Ba?a <
address@hidden>:
> At the risk of belaboring the point, I'm having trouble understanding *why*
> this works. Here's an absolutely minimal example:
>
>
> %%% TURNING ON Y-EXTENT %%%
>
> \version "2.11.39"
>
> \layout { ragged-right = ##t }
>
> \new Staff {
> \override DynamicLineSpanner #'staff-padding = #4
> c'2 \p
> c'2 \f
> }
>
> \new Staff {
> \override DynamicLineSpanner #'staff-padding = #4
> \override DynamicLineSpanner #'Y-extent = #'(0 . 0)
> c'2 \p
> c'2 \f
> }
>
>
> %%% END %%%
>
>
> In the first case we get usual (top-aligned) behavior; in the second case we
> get this excellent (center-aligned) behavior. The only difference is that
> the second example overrides DynamicLineSpanner #'Y-extent. But notice that
> the actual pair of values passed to DynamicLineSpanner #'Y-extent doesn't
> seem to matter because #'(0 . 0) and #'(1 . -1) and in fact #'(10 . -10) all
> provide this nifty center alignment.
>
> So what's going on here? It seems that merely "turning on"
> DynamicLineSpanner #'Y-extent is enough to induce vertical center alignment,
> regardless of the actual value. Is this a good way to conceptualize what's
> happening here? If so, that's quite an implicit (hidden, even) principle ...
> ie, that turning on Y-extent switches the origin of alignment from top to
> center.