No kerning for bitcoin character?

[harbortype]

I am having a hard time convicing Kern On to kern the bitcoin character with proportional figures, like this:

₿0
₿1
₿2
₿3
₿4
₿5
₿6
₿7
₿8
₿9

By default, Kern On does not kern these pairs, so I have tried:

1. Setting them as user-set autopairs.
2. Making them independent pairs.
3. Making them model pairs.
4. Changing the ₿ glyph from “Standard kerning” to “Special spacing”.
5. Generating the kerning with 0kB limit.

None of that worked. No matter what, Kern On will not generate kerning for these pairs. Is this expected?

Edit: Options 1–3 all work while the Kern On window is open and I can preview the kerning just fine. The issue is that Kern On does not keep these pairs on the final kerning.

I am running version 1.37 on Glyphs 3.4.1 (3436).

[Tim Ahrens]

In such a case, setting user-set autopairs is the way to go.

It is very strange that the pairs are not generated (or retained) during finalization, though, even with 0kB limit. Would you mind sending the me the file?

[harbortype]

I have just sent you the file via email with some more findings.

[Tim Ahrens]

Thank you for sending the file.

The reason why the bitcoin character is not kerned against the figures is simply the “No kerning” setting in the figures (which are tabular). Same for all the other currency symbols (all glyphs, in fact).

For tabular figures, “No kerning” is the most sensible setting, including the effect it has, no kerning against currency symbols.

If you have a list such as
₿10.0
₿20.0
₿30.0
then you would want the zeroes to be perfectly aligned. This is the whole point of tabular figures, of course.

Kerning between the ₿ and the tabular figures would mess up this alignment. This is why it is the correct practice not to generate it.

[harbortype]

Thank you for having a look, but I don’t think that’s the issue.

Yes, the default figures are tabular and I don’t expect them to have kerning. The issue I have is with the bitcoin character against the proportional figures.

Here’s some pairs during kerning. The bitcoin–seven.lf was manually set as an autopair.
https://kern-on.com/forum4/download/fil ... ew&id=1402

And here’s the resulting kerning against the proportional figures. This kerning was generated with a 0kB limit to rule out optimizations.
https://kern-on.com/forum4/download/fil ... ew&id=1403

As I mentioned in the email, this file has 4 masters that are set to be autokerned against other masters. If I remove these autokerned masters, the bitcoin kerning is kept on the final kerning.
https://kern-on.com/forum4/download/fil ... ew&id=1404

---

While writing this reply, I actually realized that I had not set the autopairs between bitcoin and the proportional figures on the autokerned masters. Setting them as autopairs in all masters solved my issue.

Having said that, I still think there is a bug here. Not setting a pair as auto on an autokerned master should not prevent that pair from being included on the “parent” master. I’d even argue that an autokerned master should inherit its user-set autopairs from its parent master. If I set a pair to be kerned, I probably want it to be kerned in all masters.

[Tim Ahrens]

Thanks for the clarification. You are right, the proportional figures should be kerned but the bitcoin character is not properly supported yet. I just fixed that, it will be out in the next version. I hope until then, user-set auto pairs will be sufficient.