What we are discussing here is the `self-alignment-interface`. Not
all grobs have this interface.
> Saying "associated with an object" is a bit ambiguous to me. Does an
> object "own" or define its own alignment point (like an object
> property)?
Yes, via the `self-alignment-{X,Y}` properties.
> Does a child redefine a parent's alignment point by specifying
> parent-alignment
Yes, via the `parent-alignment-{X,Y}` properties. However, there is
no 'redefinition' involved at all. A grob simply specifies with these
properties which 'point' of itself should be aligned to which 'point'
of the parent, based on the extents of itself and the extents of the
parent (which a child has access to). I put the word 'point' into
quotes since this is something virtual.
> (an externally modifiable property of the parent)? Would other
> children of the same parent see the effects of one sibling modifying
> a parent's alignment point?
Nope, `parent-alignment-{X,Y}` are not properties of the parent but of
the grob itself.
> Or does a child's override of the parent-alignment point affect only
> that child's means of specifying its desired alignment relative to
> some aspect/measurement/extent/dimension of that parent, without
> affecting any properties owned by the parent
Yes.
> or seen by other siblings?
Well, a property changed with `\override Foo.bla-bla` affects all
`Foo` grobs. You can restrict the effects by using `\once \override`
or `\tweak` to a specific grob.
>> A key information is missing IMHO: LilyPond moves a grob so that
>> its alignment point is at the same position as the parent's
>> alignment point.
>
> Agree that would be helpful if true. But who owns, or can modify,
> "the parent's alignment point"?
The grob itself, see above.
>> This means that -1, 0, and 1 represent the left/lower edge, the
>> middle, and the right/upper edge of a bounding box, respectively.
>>
>>
>> -1 0 1
>> +----------------------+----------------------+ -1
>> | |
>> | |
>> + + 0
>> | |
>> | |
>> +----------------------+----------------------+ 1
>
> The vertical -1 and 1 labels in the diagram seem to contradict the
> explanation above. Shouldn't -1 be in the lower right?
Oops! You are right.
Werner
