Re: Big Measures in Large Ensemble

[Joe Neeman]

On Sun, 2009-08-02 at 10:39 -0700, Mark Polesky wrote:
> Werner LEMBERG wrote:
> 
> > > There is a problem, however -- a
> > > slight extra space appears where the invisible fake-barlines
> > > (possible break-points) are. And the obvious way of correcting
> > > that didn't work. I'll ask the other developers if there's a way
> > > to fix that.
> > 
> > I've reported this already:
> > 
> >   http://code.google.com/p/lilypond/issues/detail?id=462
> > 
> > Maybe Joe can now `fiddle' with it, as announced :-)
> 
> Actually, I think the solution may be trivial. I *think* it's simply
> a case of updating some values in the 'space-alist property of
> BarLine for the special case of \bar "". I had been using 'X-extent
> which would add space but setting it to '(0 . 0) didn't remove the
> space. I'm looking into it.

Look into it by all means, but I don't think you will find a value for
space-alist that will solve the problem for all durations and all levels
of horizontal stretching. The problem is that the \bar "" causes
lilypond to insert an extraneous spring into the horizontal spacing
problem: we end up with one spring before the \bar "" and one spring
after the \bar "". Furthermore, the methods we use for determining the
space between notes and non-musical columns is inherently different from
the methods for determining the space between two notes (for example,
optical stem spacing correction only happens in the second case). So you
can tweak the barline-to-note spacing all you want, but I don't think
you'll get the right result because you don't have good enough control
over the note-to-barline spacing.

The semantically valid solution (IMO) is ignore the \bar "" in the
spacing problem and insert only one spring (between the notes).
Fortunately, we have a mechanism for that already (called loose
columns). It's just a matter of tweaking the inclusion test for this
mechanism to include \bar "" as a loose column.

Joe