Re: lilypond segfaulting

[Han-Wen]

address@hidden writes:
> Thanks for reporting this bug.  The problem in your example file is that
> staffAglobal creates an ambitus grob of its own (which does not get
> properly initialized and therefore crahses), competing with the one
> resulting from staffAlayerA (silly me!).  I will try to get this fixed
> until the end of this week.  I beg your pardon!

I have a nitpick:


              Pitch pitch = *unsmob_pitch (nr->get_mus_property ("pitch"));
              if (Pitch::compare (pitch_min, pitch_max) > 0) // already init'd?
                {
                  // not yet init'd; use current pitch to init min/max
                  pitch_min = pitch;
                  pitch_max = pitch;
                  ambitus_p_->set_grob_property ("pitch-min",
                                                 pitch_min.smobbed_copy ());
                  ambitus_p_->set_grob_property ("pitch-max",
                                                 pitch_max.smobbed_copy ());
                }
              else if (Pitch::compare (pitch, pitch_max) > 0) // new max?
                {
                  pitch_max = pitch;
                  ambitus_p_->set_grob_property ("pitch-max",
                                                 pitch_max.smobbed_copy ());
                }
              else if (Pitch::compare (pitch, pitch_min) < 0) // new min?
                {
                  pitch_min = pitch;
                  ambitus_p_->set_grob_property ("pitch-min",
                                                 pitch_min.smobbed_copy ());
                }
            }

for every note head, a new pitch smob is put into
pitch-{min,max}. That is wasteful. It is more efficient to postpone
writing ambitus_p_->set_grob_property() to
Ambitus_engraver::finalize(). This creates less garbage

-- 

Han-Wen Nienhuys   |   address@hidden   |   http://www.cs.uu.nl/~hanwen