On Thu, 2009-08-06 at 23:03 +0100, Neil Puttock wrote:
> 2009/8/6 Joe Neeman <[email protected]>:
>
> > As per usual, it turned out to be more complicated than I expected. But
> > it's fixed in git now.
>
> Nice work, Joe; it's a great fix.
>
> I had a go at fixing it myself, but gave up after ending up with notes
> colliding with clefs and multiple `no spring ...' errors. :)
>
You probably made the same mistake that I made initially: if you prune a
breakable column then you've pruned it's broken pieces (ie. the
line-ending and line-beginning pieces, which shouldn't be pruned) in
addition to it's mid-line piece.
> A few comments:
>
> Don't you think \bar "empty" should behave in the same way as \bar ""?
> Though it's treated differently in bar-line.cc (it's ignored
> completely, so returns an empty stencil), I think users will expect it
> to reflect its naming.
How about just checking for an empty extent, like in the attached patch?
>
> I notice the spacing for volte which don't start or end at a barline
> is tighter (it matches the behaviour without volte), though this is
> probably preferable
That wasn't intentional; I'll take a closer look to make sure it doesn't
have undesirable consequences.
> (as an aside, the volta brackets are right at the
> top of the page in the pdf, which looks like a spacing bug).
Yeah, grobs that live in System mess up the spacing problem.
Joe
diff --git a/lily/spacing-determine-loose-columns.cc b/lily/spacing-determine-loose-columns.cc
index 90db1e0..7fddb84 100644
--- a/lily/spacing-determine-loose-columns.cc
+++ b/lily/spacing-determine-loose-columns.cc
@@ -111,7 +111,8 @@ is_loose_column (Grob *l, Grob *col, Grob *r, Spacing_options const *options)
{
extract_grob_set (h, "elements", helts);
for (vsize k = helts.size (); k--;)
- if ("" != robust_scm2string (helts[k]->get_property ("glyph-name"), ""))
+ // Only non-empty bar lines count.
+ if (h->extent (h, X_AXIS).length () > 0)
return false;
}
}
_______________________________________________
lilypond-user mailing list
[email protected]
http://lists.gnu.org/mailman/listinfo/lilypond-user
