simplify previous patch-set by special casing 2-line staves

simplify previous patch-set by special casing 2-line staves

redesign dot placement of repeat sign

allow both dots in the same space only for two-line staves

an important example is TabStaff, which, for the present purposes, is
equivalent of a Staff with line-count 6 and staff-space 1.5.

new regtests for dots of a repeat sign

- test default tablature
- test set-global-staff-size and layout-set-staff-size combinations
  in default and extreme circumstances to make sure that real world
  usage is not hindered

[Keith, 2012-09-10 07:07:26]

This works, although in surprising ways.

It was a challenge for me to figure out why the tests in 'repeat-sign.ly' come
out as they do.  I was very surprised by the lines marked "dots in outer spaces"
and "dots outside" and assume we would prefer more centered placement.  The case
"dots in outer spaces" is different from a four-line staff in that the user
indicates eight scale-steps will fit between lines; if eight scale steps fit
legibly, two repeat dots will also fit.

You could simplify at this point, and work entirely in staff-positions, until
the end where you build the stencil.  I also suggest using a rule "prefer dots
in tight spaces to dots completely outside the staff" rather than the special
case for 2-line staves. 

(define (make-colon-bar-line grob)
  (let* ((staff-symbol (get-staff-symbol grob))
         (line-span (staff-symbol-line-span staff-symbol))
         (center (interval-center line-span))
         (line-pos (staff-symbol-line-positions staff-symbol))
         ;; Fold 'line-positions about its center 
         ;; to look for symmetric locations for the dots
         (folded-staff (sort (map (lambda (lp) (abs (- lp center)))
                                            (cons center line-pos))
                             <))
         (dist (or 
                 ;; Prefer dots in gaps of >1 staff-positions between lines
                 (any (lambda (x y) (if (> (- x y) 1.2)
                                      (+ x y)
                                      #f))
                      (cdr folded-staff)
                      folded-staff)
                 ;; ... but a gap of >0.5 staff-positions
                 (any (lambda (x y) (if (> (- x y) 0.6)
                                      (+ x y)
                                      #f))
                      (cdr folded-staff)
                      folded-staff)
                 ;; ... is better than outside the staff
                 (+ 2 (interval-length line-span))))
         (staff-space (ly:staff-symbol-staff-space grob))
         (staff-space (if (> staff-space 0.0) staff-space 1.0))
         (dot (ly:font-get-glyph (ly:grob-default-font grob) "dots.dot"))
         (stencil empty-stencil)
         (stencil (ly:stencil-add stencil dot))
         (stencil (ly:stencil-translate-axis
                    stencil (* dist (/ staff-space 2)) Y))
         (stencil (ly:stencil-add stencil dot))
         (stencil (ly:stencil-translate-axis
                    stencil (* (- center (/ dist 2))
                               (/ staff-space 2)) Y)))
    stencil))

[benko.pal, 2012-09-11 21:03:13]

> This works, although in surprising ways.
>
> It was a challenge for me to figure out why the tests in
> 'repeat-sign.ly' come out as they do.  I was very surprised by the lines
> marked "dots in outer spaces" and "dots outside" and assume we would
> prefer more centered placement.

that's sort of all right, I meant them to surprise,
i.e. to show the limits of the implementation.
"dots in outer spaces" was inspired by your comment
http://codereview.appspot.com/6488097#msg7
"dots outside" now changed behaviour and is accordingly relabelled to
"dots in middle".  as I hinted at in
http://codereview.appspot.com/6488097#msg8
I'm afraid improving this would more than such a contorted case is worth.

> The case "dots in outer spaces" is
> different from a four-line staff in that the user indicates eight
> scale-steps will fit between lines; if eight scale steps fit legibly,
> two repeat dots will also fit.

well, we know exactly whether two dots fit or not, but the current
issue arose just from that: though they fit, the user wants them outer.
as a theoretical consideration, I think repeat sign is a matter of the
staff and the dots, not the scale steps represented by the stafflines.
(are there chromatic or microtonal notations where these steps are
smaller than a third?)

> You could simplify at this point, and work entirely in staff-positions,
> until the end where you build the stencil.  I also suggest using a rule
> "prefer dots in tight spaces to dots completely outside the staff"
> rather than the special case for 2-line staves.

that rule is implemented by patch set two.  otherwise I think this
simplification is oversimplification as it cannot distinguish

\override #'line-count = #2

from

\override #'line-count = #2
\override #'staff-space = #2

then, if staff-space is taken into account, dot size must also be, see
http://code.google.com/p/lilypond/issues/detail?id=2648
and anyway that simplification buys us no more than replacing
gap-to-find by a constant.

[Keith, 2012-09-14 05:52:27]

Looks better than what is currently in master.

On 2012/09/11 21:03:13, benko.pal wrote:

> "dots in outer spaces" was inspired by your comment
> http://codereview.appspot.com/6488097#msg7

I was talking about a tab staff with larger staff-space, but with just two
staff-positions for each line.

Anyone writing #'line-positions = #'(-10 -2 6 14) would be planning to put a lot
of pitches in those eight staff-positions between the middle lines, and will be
surprised that LilyPond considers that not enough room for repeat dots.

More likely, someone might have removed the center line from a 5-line staff
  #'line-positions = #'(-4 -2 2 4) 
for some unusual instrument, and will not like the change.

But, if anyone complains, we can insert 0 into 'folded-staff

> (are there chromatic or microtonal notations where these steps are
> smaller than a third?)
seems crazy to me, but people do it,
<http://musicnotation.org/software/lilypondusers.html> 

> I think this
> simplification is oversimplification as it cannot distinguish
> 
> \override #'line-count = #2
> 
> from
> 
> \override #'line-count = #2
> \override #'staff-space = #2
> 

Why do you want to distinguish these two cases?

[benko.pal, 2012-09-14 06:16:57]

>> "dots in outer spaces" was inspired by your comment
>> http://codereview.appspot.com/6488097#msg7
>
> I was talking about a tab staff with larger staff-space, but with just
> two staff-positions for each line.
>
> Anyone writing #'line-positions = #'(-10 -2 6 14) would be planning to
> put a lot of pitches in those eight staff-positions between the middle
> lines, and will be surprised that LilyPond considers that not enough
> room for repeat dots.
>
> More likely, someone might have removed the center line from a 5-line
> staff
>   #'line-positions = #'(-4 -2 2 4)
> for some unusual instrument, and will not like the change.
>
> But, if anyone complains, we can insert 0 into 'folded-staff
>
>
>> (are there chromatic or microtonal notations where these steps are
>> smaller than a third?)
>
> seems crazy to me, but people do it,
> <http://musicnotation.org/software/lilypondusers.html>
>
>
>> I think this
>> simplification is oversimplification as it cannot distinguish
>>
>> \override #'line-count = #2
>>
>> from
>>
>> \override #'line-count = #2
>> \override #'staff-space = #2
>
> Why do you want to distinguish these two cases?

because (maybe depending on set-global-staff-size and
layout-set-staff-size) the latter has enough room for two
dots, the former hasn't.  and that brings me back to my
main argument: I can't guess why the user chose a
specific way of manipulating the staff and how (s)he
interprets it, I also don't know at that point why the staff
space is that big as is (staff-size, set-global-staff-size,
layout-set-staff-size), but I know exactly whether one
or two dots have enough room or not.

I admit that the request to restore the look of the original
repeat sign of a TabStaff makes all that not enough.

I also think that users would be surprised that

\override #'staff-space = #0.5

and

\override #'line-positions = #'(-2 -1 0 1 2)

makes the repeat sign look different, even if they know that
pitches are at different places - but who thinks that the dots
of a repeat sign mark pitches?

p

[Keith, 2012-09-14 06:53:47]

On 2012/09/14 06:16:57, benko.pal wrote:
> main argument: I can't guess why the user chose a
> specific way of manipulating the staff and how (s)he
> interprets it, 

If the user writes 
> \override #'staff-space = #0.5
he wants to scale things.

If he writes
> \override #'line-positions = #'(-2 -1 0 1 2)
he wants to re-arrange things

[benko.pal, 2012-09-14 10:11:28]

On 2012/09/14 06:53:47, Keith wrote:
> On 2012/09/14 06:16:57, benko.pal wrote:
> > main argument: I can't guess why the user chose a
> > specific way of manipulating the staff and how (s)he
> > interprets it, 
> 
> If the user writes 
> > \override #'staff-space = #0.5
> he wants to scale things.
> 
> If he writes
> > \override #'line-positions = #'(-2 -1 0 1 2)
> he wants to re-arrange things

perhaps yes, perhaps not (using different weird microtonal thingies).  in this
comparison the relevant thing is whether he wants different repeat signs.  note
that, unfortunately, LilyPond has several scaling possibilities, enabling the
user to scale some and not other things.  if everything (dots included) is
scaled, the code under review will do the right thing; if some is not, then
divining the intent is not so easy.

to me the repeat sign is a purely engraving matter.  making it depend on the
musical meaning of the staff is just making it more fragile.
if we want repeat signs in a TabStaff look as requested, then I want every
repeat sign (with identical dot size) in an identically looking staff
(regardless how it's achieved by mixing line-positions, set-global-staff-size,
layout-set-staff-size, line-count, staff-space) look identical.