Reverse-direction ties?

[Dave LaDelfa]

Ties are traditionally drawn so that they curve out away from the stems, i.e. if
the stems go up, the tie curves downward between the note heads. This is the
basic rule followed unless something gets in the way of the tie, e.g. in divisi
writing the stems-up voice's ties curve up and the stems-down voice's ties curve
down.

Slurs work the same way and so therefore ties and slurs look a lot alike.

One composer who I did a lot of parts-copying for in my youth (with a
calligraphic  fountain pen, four Rapidographs, and a bottle of white-out!)
preferred a deviation from the standard notational practice for ties: he wanted
ties to curve TOWARDS the stems wherever possible -- the opposite of traditional
practice. His rationale was that ties would clearly be differentiated visually
from slurs, and that as a result things would get performed more correctly the
first time they were read, making for more efficient rehearsals.

After my inital reluctance, I became a convert to this practice. It looks
strange on the page for about 30 seconds, and then you get used to it and the
logic become instantly evident. I can't bear to go back to the old way now.

No notational software that I know of has a way to change the rule for tie
direction globally. Can it be done in Lilypond? I guess what I'd like to tell
the tie engraver is "once you've calculated out whether the stem goes up or
down, pretend it's the other way." 

The workaround in other software I've used (predominantly MusicPress) is to
manually set each tie in the "wrong" direction. The problem with this is that
once you transpose the notes (e.g. to extract a horn or clarinet part) you have
to go through and manually redo any of the ties whose notes have crossed the
center line and flipped stem direction. This seems like it would be a horrendous
nightmare in Lilypond due to the way the scores and parts typically derive from
the same notestream.

Dave LaDelfa