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

[lilypond]

Comment #25 on issue 1336 by n.puttock: skipTypesetting segfaults when set  
during a skipBars-induced MultiMeasureRest spanner
http://code.google.com/p/lilypond/issues/detail?id=1336

> Neil, could you make a new patch which removes the TODO comment, and then  
> put it up on reitveld for an official 24-hour review?

Unfortunately, there does appear to be a problem with the patch, which I'm  
unsure how to fix without more investigation: setting skipTypesetting = ##t  
for a full score (or doing the equivalent using showLastLength = s1*0)  
causes a segfault.

Program received signal SIGSEGV, Segmentation fault.
0x00007ffff5b89103 in __dynamic_cast () from /usr/lib/libstdc++.so.6
(gdb) bt
#0  0x00007ffff5b89103 in __dynamic_cast () from /usr/lib/libstdc++.so.6
#1  0x000000000062c344 in get_column_description (cols=..., col_index=0,  
line_starter=true) at simple-spacer.cc:367
#2  0x000000000062c9b0 in get_line_forces (columns=...,  
line_len=89.341653543307075, indent=0, ragged=false) at simple-spacer.cc:421
#3  0x0000000000488b48 in Constrained_breaking::initialize  
(this=0x7fffffff8530) at constrained-breaking.cc:444
#4  0x0000000000487eb0 in Constrained_breaking::Constrained_breaking  
(this=0x7fffffff8530, ps=0xcb24e0, start=...) at constrained-breaking.cc:360

There's only one pair of paper columns in the score (created in  
Paper_column::initialize () before skipTypsetting can be read) which seems  
to cause all sorts of problems for the breaking algorithm.