Re: Issue 3307: Rest positions incorrect with non-standard line count (issue 8580045)

[dak]

On 2013/04/10 05:47:20, Keith wrote:
> You changed more than necessary to fix the reported bug.  This is good
because
> finally voiced r1-style whole rests are in the correct positions
(which I
> confirmed with Ted Ross' book just to be careful).

More by coincidence than by design.  That was likely the most
significant change affecting five-line staffs.

> Quarter-rests and shorter look a bit odd now, on the even-line-count
staves, as
> they are not in their usual position on a line. I'm not sure if this
is
> objectively 'wrong'.

Again, I don't know what the standards would be.  The non-centered rests
look bad particularly in tablature, but are also weird in two-line drum
staffs.

The best solution likely would be glyphs explicitly designed for the
off-by-half-line use case.

> The reported bug does /not/ show as fixed in the make-check output for
> 'tablature-full-notation.ly'  which confuses me. From the code, I
would have
> expected it to come out right.


https://codereview.appspot.com/8580045/diff/5001/input/regression/rest-on-nonstandard-staff.ly
> File input/regression/rest-on-nonstandard-staff.ly (right):


https://codereview.appspot.com/8580045/diff/5001/input/regression/rest-on-nonstandard-staff.ly#newcode39
> input/regression/rest-on-nonstandard-staff.ly:39: \new StaffGroup
\with {
> \override StaffGrouper.staff-staff-spacing.padding = 3 }
> The short-term effect is to make the image-comparision less useful in
seeing the
> effect of the patch.... but this long regtest is not very useful
anyway.

I can alternatively tell the regtest to expect overfull page warnings.
I actually don't have a clue why they would be there in the first
place: maybe pitched multi-measure rests don't count in the page
height estimates?

> https://codereview.appspot.com/8580045/diff/5001/lily/rest.cc
> File lily/rest.cc (right):


https://codereview.appspot.com/8580045/diff/5001/lily/rest.cc#newcode46
> lily/rest.cc:46: Rest::y_offset_internal (Grob *me, int duration_log,
int dir)
> Naming: something like  staff_position_internal()  to help us remember
that this
> returns a quantity in units of staff-positions, because Y-offset is
always in
> units of (global) staff-spaces.

Yes, that's reasonable.  The function at first returned an offset, but
it turned out that the multi-measure rest code needs staff-spaces.


https://codereview.appspot.com/8580045/diff/5001/lily/rest.cc#newcode78
> lily/rest.cc:78: if (duration_log > 1)
> Shorter rests look ... unfamiliar if they are not aligned, but I'm not
sure
> which is correct.

Neither am I.  I was flying this by "least ugly" according to my
subjective impression.

For ancient music (four-line staff), rests of small size would be
unusual.  For full tablature and drums, we'll need input from players
or (preferably) theory books and hand-engraved scores.

It is also clear that multi-measure rests, particularly in multi-voice
situations, are still bonkers.  I have problems finding where they get
their positions.


https://codereview.appspot.com/8580045/