Re: Assertion failure on current master

[address@hidden]

On Oct 1, 2011, at 7:01 PM, address@hidden wrote:

> On Oct 1, 2011, at 6:53 PM, Neil Puttock wrote:
> 
>> Hey guys,
>> 
>> I can't complete test-baseline due to an assertion error running
>> mozart-hrn-3.ly.  Here's the backtrace:
>> 
>> Drawing systems...lilypond: ../flower/include/interval.hh:226: T
>> Interval_t<T>::center() const [with T = double]: Assertion `!is_empty
>> ()' failed.
>> 
>> Program received signal SIGABRT, Aborted.
>> 
>> (gdb) bt
>> #0  0x00007ffff508fd05 in raise (sig=6)
>>  at ../nptl/sysdeps/unix/sysv/linux/raise.c:64
>> #1  0x00007ffff5093ab6 in abort () at abort.c:92
>> #2  0x00007ffff50887c5 in __assert_fail (assertion=0x70bd60 "!is_empty ()",
>>  file=<value optimised out>, line=226, function=<value optimised out>)
>>  at assert.c:81
>> #3  0x000000000041bf01 in Interval_t<double>::center (this=0x7fffffff7310)
>>  at ../flower/include/interval.hh:226
>> #4  0x00000000006d8b2c in Tuplet_number::calc_y_offset (smob=0x7ffff22bebe0)
>>  at tuplet-number.cc:83
>> #5  0x00007ffff7926ae4 in scm_dapply () from /usr/lib/libguile.so.17
>> #6  0x00000000004fc65f in Grob::try_callback_on_alist (this=0x108d150,
>>  alist=0x108d1b0, sym=0x7ffff2a6c080, proc=0x7ffff45ac530)
>>  at grob-property.cc:231
>> #7  0x00000000004fc398 in Grob::internal_get_property (this=0x108d150,
>>  sym=0x7ffff2a6c080) at grob-property.cc:188
>> #8  0x0000000000505efd in Grob::get_offset (this=0x108d150, a=Y_AXIS)
>>  at grob.cc:383
>> #9  0x0000000000505a67 in Grob::relative_coordinate (this=0x108d150,
>>  refp=0x184dc00, a=Y_AXIS) at grob.cc:312
>> #10 0x00000000005062a4 in Grob::extent (this=0x108d150, refp=0x184dc00,
>>  a=Y_AXIS) at grob.cc:427
>> #11 0x000000000043a1ad in add_boxes (me=0x108d150, x_common=0x18950a0,
>> 
>> Looks like the new tuplet collision avoidance code.
>> 
>> Cheers,
>> Neil
> 
> Hey Neil,
> 
> I just ran mozart-hrn-3.ly on an unoptimized binary and could not reproduce 
> the problem.  Could someone else please verify?
> 

A follow-up: I can't figure out how the error could come about.  
Interval::center should, in Tuplet_number::calc_y_offset, always be getting an 
interval for which it can find the center (it uses robust_scm2interval).  Does 
anyone with more computer chops than I have have any intuition about this?

Cheers,
MS