Improve the bounding box around r and g, reducing the need for kerning

[pelson]

Improve the bounding box around r and g, reducing the need for kerning (especially no need for kerning all letters)

I think we need to do the same for e and j. The spacing between gj is particularly large.

[dummy-index]

I don't think it's smart to adjust things so that certain letters can avoid kerning settings. If you're going to adjust the left bearing of 'g' until the bowl sticks out, are you also planning to do the same for the left side of 'o'?

For example, if you perform proactive width adjustments only on the right side, and handle processing for tails of letters like 'g' and 'j' on the left side, could autokern be unified in the form of (xxx, all_chars, 'x')... I wonder.

[pelson]

I don't think it's smart to adjust things so that certain letters can avoid kerning settings. If you're going to adjust the left bearing of 'g' until the bowl sticks out, are you also planning to do the same for the left side of 'o'?

I think you're misunderstanding my intent - I believe that kerning is acceptable and necessary to make the font clean and convincing, however I also believe that if we are applying kerning to all letter pairs for a given letter, then we have the bounding box wrong in the first place. With your example, I would adapt the bounding box for o until it fits well next to B, and then kern when it needs it for a pair such as To.

[dummy-index]

I agree with that way of thinking. If that's the case, isn't the g curve sticking out by about 20 units a sign that we need to reconsider space = 20 in the first place? Actively incorporating ligatures from handwriting_minimal inevitably means aiming to reproduce the spacing of handwriting_minimal. First, change the right side bearing from space * 2 to space…

Please continue with this PR as is.