On Oct 1, 2011, at 9:29 PM, David Kastrup wrote: > "m...@apollinemike.com" <m...@apollinemike.com> writes: > >> But I just pushed the actual cure. >> >> The problem was that positions stores two Y positions - a left and a >> right. If the left is higher than right, than the interval is read as >> empty. So, the center needs to be calculated manually. >> >> Interval is not really the right thing to use for positions if one >> wants to use the instance methods of interval (like is_empty), but I >> use it because it is already used in tuplet-bracket.cc for the print >> function (in a way that doesn't risk an assertion error). It could >> also be put in a real drul array, but there is currently no >> robust_scm2realdrul. This would definitely be a welcome addition to >> avoid confusion in the code base. > > This sounds really really icky. The sort of code I hope never to need > to deal with. Could you file a suitable "Enhancement" issue so that > redoing this in a less hackish manner does not get forgotten? That way > I have a better chance for my hopes becoming true. >
Actually, I found a robust_scm2drul that does the trick and pushed it to master. So now, no more kludgy code. Sorry for the panic - I have no clue why my unoptimized binary wasn't failing :( Cheers, MS _______________________________________________ lilypond-devel mailing list lilypond-devel@gnu.org https://lists.gnu.org/mailman/listinfo/lilypond-devel