Irrational rationale for com.google.fonts/check/usweightclass

Observed behaviour

 >> com.google.fonts/check/usweightclass
    Checking OS/2 usWeightClass.
    with ./Fonts/vf/Petrona[wgth].ttf

    * FAIL: OS/2 usWeightClass expected value for 'Regular' is 400 but this font has 82.
 GlyphsApp users should set a Custom Parameter for 'Axis Location' in each master to ensure that the information is accurately built into variable fonts. [code: bad-value]

    Result: FAIL

Expected behaviour

Apart from the usWeightClass issue already described here and here, I don’t see how adding a Axis Location custom parameter would change anything. It feels like the rationale text is misleading here. As mentioned in the other issue referenced above, GlyphsApp users should rather set a weightClass custom parameter on the instances, or better yet, fontmake can set these values itself according to the weights set in the instances.

Resources and exact process needed to replicate

This is the VF: Petrona-ROMAN-MASTER-NEW-W.glyphs

[arrowtype]

A note for @felipesanches on the suggested fix: I helped get weight locations working in Nunito By setting axis location custom parameters in instances (and not masters). Setting this in masters resulted in avar values only being present for master weight values (200,800,1000), and as a result, the other instance values were a bit off (Regular was something like 432.34375, etc).

I’m not 100% sure that it also set weightClass values correctly, though – I assume so, but I’d have to check specifically.

[thundernixon]

Don’t know why it thinks a Thin Italic should be 250?

It shouldn’t be, at least not in a TTF.

https://github.com/googlefonts/fontbakery/issues/2644 is an issue to correct this, but ultimately it looks like it’s an issue of how the name parser isn’t yet built to parse style names that aren’t just weight names (so, the width descriptor is messing with it).

Interestingly, apparently old versions of windows auto-bolded CFF/OTF fonts with a weight of <= 249, and I guess that’s where the “250” comes from. https://typedrawers.com/discussion/2477/numeral-values-of-weights