Re: lilypond bug 2368

[Janek Warchoł]

On Sat, Apr 28, 2012 at 7:49 PM, address@hidden
<address@hidden> wrote:
> There's been a lot of back and forth on a few LilyPond lists about improving
> ties.  I'm of the opinion that the system in place is sound (it is a similar
> system as that of slurs and beams), but what needs improvement is an
> understanding of how much factors are worth and where they make the
> difference.  We've made some headway in slurs in 2.15 by removing
> unnecessary factors and decorolating them but we haven't been able to do the
> same yet for ties.  So, my advice to everyone would be to play around with
> this details list.

Surely i will experiment when i'll be finishing my tie report :)
However, given how often current ties fail, i'm pretty sure that the
current system is not good.  I have a strong feeling that it's an
overkill; from what i've already seen i could probably write a
deterministic algorithm that would perform better in many cases and
use half as much parameters.
I may be wrong; however, i think that the core of the problem is that
while slurs are really complicated and need advanced heuristics and
guesswork, ties are significantly more straightforward.

cheers,
Janek