Clef: Use `extra-space` for distances to time and key signatures

• Avoid horizontally oversized regression tests

  This improves viewing of test-results/index.html.

• changes.tely: Avoid overlong line

• Clef: Use extra-space for distances to time and key signatures

  Despite of being only three modified lines, this is a huge change, affecting a lot of scores.

  Here is a table of old and new widths as reported by ly:paper-column::print (at 20pt staff size).

  clef + time sig	old	new	difference	comment
  treble	7.70	7.55	-1.6%
  bass	7.70	7.66	0%
  alto	7.70	7.70	+0.5%
  GG	7.70	9.05	+17.9%	was far too narrow
  varC	7.70	7.23	-5.6%
  petrucci-c3	7.70	5.98	-21.8%	was far too wide

  The results for 'clef + key sig' are very similar.

  Note that, as before, the widest clef in a staff group determines the horizontal position of all clefs in a system; this means, for example, that a piano score containing a treble and a bass clef doesn't change at all.

  Fixes issue #3598 (closed).

assigned to @lemzwerg

added Patchnew label

changed title from Avoid horizontally oversized regression tests to {+Clef: Use extra-space for distances to time and key signature+}s

changed the description

Regarding regression tests: Almost everything changes. Note that it's hard to browse them; there are limits that prevent the loading of all images in one rush; it's probably one of the rare moments where it's easiest to download the test results (58.8MByte).

I've checked the changes in detail offline; everythings seems to be as expected.

added Patchreview label and removed Patchnew label

On the whole, this is a major improvement! Comparing the two images above, I also think that the first one (with alto clef +0%) looks better than the second one, and I think it’s a very reasonable change to make, even with the sweeping impact on most scores. Avoiding change just because it will be widely noticeable adds a moment of inertia to the project that’s not inherently helpful, provided the changes are well-reasoned. If we knew that the current values are indeed based on the alto clef’s width, we should definitely change spacing for the treble clef.

On the other hand, the scope of this MR is the relative spacing of different clefs, not the absolute spacing of any clef, so taking a conservative approach and opening a new issue regarding absolute values for extra-space seems fair. After all, those absolute values are also easily changed in a stylesheet, and overall density of horizontal spacing is definitely a matter of style.

added Patchcountdown label and removed Patchreview label

added Patchpush label and removed Patchcountdown label

Up to now, I got exactly the same number of responses for and against basing the distances on the alto clef. I'll wait another cycle before committing, asking again on the lists.

added Patchcountdown label and removed Patchpush label

Thanks Werner. I think this is a definite improvement (either version). I have a slight preference for the 0% treble clef values, for both aesthetic and practical reasons.

I prefer the alto-based (closer) spacing.

I prefer the treble-based spacing (second image). I think keeping the treble clef spacing the same and basing the others after that makes the most sense.

Question still seems unresolved, leaving just in case. at Countdown

added 1 commit

• ea60917e - Clef: Use `extra-space` for distances to time and key signatures

Compare with previous version

added Patchnew label and removed Patchcountdown label

Since the opinions are almost equally split between preferring the treble clef or preferring the alto clef, I came to the conclusion that perhaps something in-between is the best solution, and here it is: Now the bass clef stays unchanged, and the differences for treble and alto clefs are minimized.

added 39 commits

• ea60917e...ca8e27f1 - 36 commits from branch master
• aa69b34e - Avoid horizontally oversized regression tests
• 6609426f - changes.tely: Avoid overlong line
• a71e824e - Clef: Use `extra-space` for distances to time and key signatures

Compare with previous version