Re: bug with tupletFullLength

[Thomas Morley]

Hi Colin,

2012/2/27 Colin Hall <address@hidden>:
> On Mon, Feb 27, 2012 at 02:12:06AM +0100, David Kastrup wrote:
>> Colin Hall <address@hidden> writes:
>>
>> > On Sun, Feb 26, 2012 at 06:36:05PM +0100, Thomas Morley wrote:
>> >> The following code gives a log warning:
>> >>
>> >> \version "2.15.30"
>> >> \new Staff {
>> >>    \set tupletFullLength = ##t
>> >> %   \override TupletBracket #'full-length-to-extent = ##f
>> >>    \time 1/8
>> >>    \times 2/3 { c'16 c' c' }
>> >> \mark "123456789"
>> >> }
>> >>
>> >> log:
>> >> warning: Found infinity or nan in output. Substituting 0.0
>> >>
>> >> If the mark contains less than 8 characters the warning disappears.
>> >> Same, when commenting in \override TupletBracket #'full-length-to-extent 
>> >> = ##f
>> >>
>> >> Compiles without warning using 2.12.3
>> >>
>> >
>> > Thanks for the bug report, Thomas, but I'm not sure that I should
>> > create an issue tracker for this.
>> >
>> > As far as I can tell you are reporting that 2.15.30 gives a warning
>> > and 2.12.3 does not.
>> >
>> > Do 2.15.30 and 2.12.3 both give good typeset output?
>> >
>> > In that case, I think the difference in behaviour is not a bug, but
>> > just a change in the way the software handles a particular case.
>> >
>> > Anyone from bug squad or dev care to comment?
>>
>> The warning describes a problem with data generated internally, not a
>> problem with the user input.  This is definitely a bug.  It is not
>> necessarily a regression since it is conceivable that 2.12.3 just did
>> not diagnose the problematic situation.  So it will require some
>> analysis by a developer to decide whether this is also a regression.
>
> I've created a new issue tracker 2362 for this bug, Thomas. Thanks again for 
> the bug report.
>
> http://code.google.com/p/lilypond/issues/detail?id=2362
>
> Cheers,
> Colin.
>
> --
>
> Colin Hall

I noticed the bug when working on http://lsr.dsi.unimi.it/LSR/Item?id=705
After further investigation I think the bug is more related to the
RehearsalMark.

The following code gives no warning with 2.14.2 or 2.15.30 but a weird
output. And again all's fine when commenting in \override
TupletBracket #'full-length-to-extent = ##f


\version "2.14.2"

\new Staff {
   \set tupletFullLength = ##t

%   \override TupletBracket #'full-length-to-extent = ##f

   \time 1/8
   \times 2/3 { c'16 c' c' }
   \override Score.RehearsalMark #'break-visibility = #'#(#t #t #t)
   \override Score.RehearsalMark #'direction = #DOWN
   \mark "12345678"
}



Cheers,
  Harm

bug.png
Description: PNG image

2.12.3.png
Description: PNG image