Fontbakery: [com.google.fonts/check/usweightclass] Fail with weightclass 100 in variable font

[arialcrime]

Observed behaviour

The (wip) spec recommends setting the OS2 weightclass for Thin to 100, but Google Fonts’ usweightclass check returns a FAIL if I follow the spec.

The error message reads:
FAIL: OS/2 usWeightClass expected value for '9pt Thin' is 250 but this font has 100.

Expected behaviour

Not sure. Either no fail in the check, or a clearer explanation what the value can/should be.

[m4rc1e]

Yep, we need to fix this.

For CFF we apparently want 250, for TT 200.

[tphinney]

For CFF we apparently want 250, for TT 200.

I understand why 250 is needed for CFF, but I can’t imagine why 200 would be a breakpoint. Did you mean to say 100—the normally correct value for Thin? Or is there some way in which TT behavior is particularly surprising (to me, anyway)?

[m4rc1e]

Apologies, For TT we want 100 for Thin and 200 for ExtraLight.

[m4rc1e]

I'm writing a wip spec/docs atm that has this info. Hopefully we can just use this as a single source of truth instead of github issues.

[thlinard]

Issue 1706 ("Establish mapping for usWidthClass to css font-stretch") is still open.

[m4rc1e]

It's wip atm

[m4rc1e]

We must prioritise fixing this issue. Every vendor complains about it, https://github.com/xotypeco/big_shoulders/issues/9#issuecomment-623573511

[m4rc1e]

Working on this today.

[felipesanches]

Thanks, @m4rc1e