Han-Wen Nienhuys writes:
> address@hidden writes:
> [...]
> >
> > This changes tie-ing behaviour so that not the exact pitch property
> > of two candidate notes is compared, but the normalized chromatic
> > pitch. (The old behaviour can, of course, be produced by not using
> > ~ in the first place.)
>
> Does this actually work? IIRC, the Tie code is hard-wired to assume
> that ties are always horizontal.
>
You're right, the tie is still horizontal, i.e. it will end half a
line too low or too high (I lack the expertise to change tie direction
as well). It still looks rather good to me - way better than the
alternative (tieing to an invisible note of the appropriate pitch)
because that would let the tie end much too soon.