alerque / libertinus

The Libertinus font family
SIL Open Font License 1.1
979 stars 56 forks source link

Link Semibold Regular/Italic faces into one (legacy) family name #393

Open alerque opened 3 years ago

alerque commented 3 years ago

Somewhere between v6.11 and v6.12, the Serif Semibold became two families. The italic style is showing up with the family name as Libertinus Serif Semibold Italic and subfamily Regular instead of both fonts being Libertinus Serif Semibold, one with one Subfamily Regular and one Italic.

The snafu with font names in #390 brought this up again. I was so worried about getting past the broken release to working v6.12/v7.000 parity that I didn't look into it further, but this could definitely use fixing as well.

Tangentially related would be dropping the legacy support altogether, see #392. But as long as we have it, it might as well be working properly.

From Khaled's comment:

Family:              Libertinus Serif Semibold
Subfamily:           Italic
Full name:           Libertinus Serif Semibold Italic
PostScript name:     LibertinusSerif-SemiboldItalic

This is what I expect, and it was that way in 6.11, no idea what broke it in 6.12.

alerque commented 3 years ago

Somewhat unsurprisingly I guess, this issue bisects to 33bce0e6 as the first bad commit.

alerque commented 3 years ago

I think the offending code is going to turn out to be this monkey business in ufo2ft:

https://github.com/googlefonts/ufo2ft/blob/b796cb7d225768645fbd47101424cc298de8fdc9/Lib/ufo2ft/fontInfoData.py#L70-L76

khaledhosny commented 3 years ago

You need to check the generated UFOs, quite possibly they might have the wrong names set.

kenmcd commented 3 years ago

Tangentially related would be dropping the legacy support altogether, see #392. But as long as we have it, it might as well be working properly.

If you drop this the fonts will not work properly in LibreOffice on Windows.

alerque commented 3 years ago

@kenmcd Thanks, but I already came to that conclusion. Hence why the related tracking issue #392 was closed, see comments there.