Re: Issue 1336 in lilypond: skipTypesetting segfaults when set during a skipBars-induced MultiMeasureRest spanner

[lilypond]

Comment #9 on issue 1336 by Carl.D.Sorensen: skipTypesetting segfaults when  
set during a skipBars-induced MultiMeasureRest spanner
http://code.google.com/p/lilypond/issues/detail?id=1336

Here's a patch that seems to avoid the segfault in what I think is a  
reasonable way.  It checks for a null grob to align on and returns an  
offset of zero if it's found, along with setting a programming error.


diff --git a/lily/self-alignment-interface.cc  
b/lily/self-alignment-interface.cc
index e339e80..23a89a0 100644
--- a/lily/self-alignment-interface.cc
+++ b/lily/self-alignment-interface.cc
@@ -66,6 +66,11 @@ Self_alignment_interface::aligned_on_self (Grob *me,  
Axis a, bool
 SCM
 Self_alignment_interface::centered_on_object (Grob *him, Axis a)
 {
+  if (!him)
+    {
+      programming_error ("cannot center: null grob");
+      return scm_from_int (0);
+    }
   return scm_from_double (robust_relative_extent (him, him, a).center ());
 }


It's probably not as "correct" as suiciding the part of the grob that goes  
away with the earlier programming error, but I think it fixes the critical  
issue.


Patch is also available on Rietveld:

http://codereview.appspot.com/3308043