Lilypond crash related to tuplets

[Andrew Bernard]

I'm not sure if this is the right list to discuss this on. Recently in my
200 page string quartet the tuplet brackets, which always used to be fine,
have started becomg too short, randomly, with no apparent obvious pattern,
and only a little bit, but enough to make the score very unprofessional
looking. Worse, for no reason I can fathom, suddenly I can't compile any
more due to lilypond dumping core.

So I went through the exercise of building lilypond in order to get an
executable with debugging symbols. learned how to run gdb, and was
interested to observe this in the (very long) backtrace:

#2  0x00007ffff74c640f in __assert_fail_base
    (fmt=0x7ffff7652858 "%s%s%s:%u: %s%sAssertion `%s' failed.\n%n",
assertion=0x555555915f81 "!is_empty ()", file=0x555555915f63
"../flower/include/interval.hh", line=227, function=<optimised out>) at
assert.c:92
#3  0x00007ffff74d6142 in __GI___assert_fail
    (address@hidden "!is_empty ()",
address@hidden "../flower/include/interval.hh",
address@hidden, address@hidden
<_ZZNK10Interval_tIdE6centerEvE19__PRETTY_FUNCTION__> "T
Interval_t<T>::center() const [with T = double]") at assert.c:101
#4  0x00005555556e4a06 in Interval_t<double>::center() const
(this=<optimised out>) at ./include/lily-guile-macros.hh:70
#5  0x00005555556e4a06 in Tuplet_number::calc_x_offset(scm_unused_struct*)
(smob=<optimised out>) at tuplet-number.cc:293

I am running 2.19.83 on Ubuntu 18.10. The same occurs with 2.19.82.

This seems to indeed indicate something in the tuplet machinery going wrong.

I have disabled ally my music functions that have anything to do with
modifying tuplets. I did this because the exact error came up a year or two
go and that turned out to be related to matters regrading setting the
stencil to #point-stencil or not. But that's not the current issue,
although I suspect it is close.

What calls calc_x_offset for Tuplet Number?

Do I need to compile a non-optimised version of lilypond to trace this
further?

Any assistance would be greatly appreciated. My string quartet is going
nowhere but I am learning a lot about debugging lilypond.

I am not sure what list this belongs to as I feel this may very well be a
conding error in my source, and not a lilypond bug, but it's very hard to
determine. I cannot make it happen in an MWE.

Andrew