Fingering: a few observations
-
It is not possible to turn off automatic placement for fingering without losing the all-important hard-coded offsets. Suggestion: Allow AP to be turned off and retain hard-coded offsets. This would be handy for certain situations in which the user wants to move the fingering onto the staff or across a note beam, for example.
-
The height of fingering above a beamed note can be inconsistent over the stem. e.g. if you change the last note in the following beamed series from A to B, the height of the fingering above the beam increases slightly; if you move the note up to C, it changes again, and so on.
Note: This probably affects downstem beamed notes as well
Comments
I'm not understanding what you mean about preserving offsets. If you turn off autoplace, those offsets are no longer relevant, because they would be relative to a completely different starting position. If you want to overlap something, you'll need new offsets, not just for fingerings, but for anything. There is no way to have autoplace "on" enough to still choose the default position but "off" to allow overlaps - the whole way it chooses the default position is by avoiding overlaps. FWIW, though, I do plan to submit a PR soon that will allow Alt+drag to automatically disable autoplace, which I think probably gets at the same basic desire. I suppose it would be possible to devise a new "turn off autoplace but simulatenously convert the old default position into the appropriate offset needed to keep the element where autoplace would have placed it" - this is basically what you are asking for?
Regarding beams, I suspect this is because we can only approximate the shape of a beam during collision avoidance using a series of boxes. We could increase the number of such boxes for a better approximation, but then layout gets slower. Maybe there is better a tradeoff, though.