#3454 Crash with malformed tremolo code

[Anonymous]

Originally created by: *anonymous

Originally created by: PhilEHol...@googlemail.com

{ \repeat tremolo 4 { b32 [ g'32 ] } }

causes a crash on Windows and Ubuntu.  Log file output:

GNU LilyPond 2.17.23
Processing `TremoloCrash.ly'
Parsing...
TremoloCrash.ly:1: warning: no \version statement found, please add

\version "2.17.23"

for future compatibility
Interpreting music...
TremoloCrash.ly:1:23: warning: stem does not fit in beam
{ \repeat tremolo 4 {
                      b32 [ g'32 ] } }
TremoloCrash.ly:1:27: warning: beam was started here
{ \repeat tremolo 4 { b32
                          [ g'32 ] } }
programming error: Stem already has beam
continuing, cross fingers
TremoloCrash.ly:1:29: warning: stem does not fit in beam
{ \repeat tremolo 4 { b32 [
                            g'32 ] } }
TremoloCrash.ly:1:27: warning: beam was started here
{ \repeat tremolo 4 { b32
                          [ g'32 ] } }
programming error: Stem already has beam
continuing, cross fingers
Preprocessing graphical objects...Segmentation fault

[Google Importer]

Originally posted by: dsch...@gmail.com

The crash is in Beam::whole_note_close_chord_tremolo, which uses stems[0], but the tremolo beam has no stems. The "stem does not fit in beam" warning is because the 32nd notes become quarter notes but still have their beam. "Stem already has beam" is because that beam collides with the tremolo beam.

This was introduced in d8dfa74, which was reverted in 2122a96 (for unrelated reasons), so the crash is fixed since 2.17.29. It may return if that patch is resurrected for issue 704, and other empty-beam bugs are plausible, so this should be added to the tests.

[Valentin Villenave]

This crash appears to have been fixed for at least five years; I move to close this issue.