Open thundernixon opened 6 years ago
The GF doc recommendation talking about vertical metrics and hhea is weird - there is a vhea table for vertical metrics, you know...
Argh, I see this sort of (mis-)usage in Glyphs' doc. That's unfortunate as "vertical metrics" has a different meaning in the opentype spec...
Sorry, maybe something like "platform-specific vertical metrics" would be a better term?
What you are talking about are vertical sizes/distances in horizontal layout... The more usual meaning of "vertical metrics" are horizontal sizes/distances in vertical layout...
Since MS Val predates fsSelection bit 7 use_typo_metrics, I'm guessing Ms Val wants all 3 sets of vert metrics to sum the same?
⚠️ WARN MS-FonVal: Ascender is different than OS/2.usWinAscent. Different line heights on Windows and Apple DETAILS: hhea.Ascender = 1443, OS/2.usWinAscent = 1708
If you enable bit7 which we recommend, the winAscent is not used. As the bit suggests, it will use the typo Metrics instead.
This Font Val test should be disabled imo.
The actual code is in https://github.com/HinTak/Font-Validator/blob/master/OTFontFileVal/val_OS2.cs , no need to speculate.
This is basically how the source code is organized - one file per opentype table, "val_*.cs"
That's unfortunate as "vertical metrics" has a different meaning in the opentype spec...
In the type design community, no one uses the meaning of the opentype spec. I expect outside of CJK fonts, almost no fonts have a vhea
table except Bungee.
I feel like I'm spending a bunch of time chasing around checks that I'm unsure of the validity of. :/
This Font Val test should be disabled imo.
Marc and Felipe, when you've finished going over all the FB checks to doublecheck their priority is set correctly, it would be good to do the same process on the FontVal checks, and make a comprehensive list of which Font Val checks you want to filter.
Observed behaviour
In the GF Docs spec on Vertical Metrics, it is recommended that vertical metrics be set in new fonts such that:
and
However, this prompts two FontVal errors (at least in any font with accents that have points higher than caps and ascenders, including Montserrat):
Expected behaviour
I would expect to set
Hhea
andtypo
values such that they match thewinAscent
values, to keep line heights across platforms.If line heights really aren't intended to be the same between platforms (or they stay the same even with different
Hhea
andwin
values), I would expect this warning to not show up once I set vertical metrics as recommended.Is this potential inconsistency documented anywhere? Is the FontVal error incorrect in stating that line heights will be different? If so, there doesn't seem to be an issue filed yet, but I could do this if it's valid.
Resources and exact process needed to replicate
===
If these warnings are invalid, please let me know so I can file an issue for FontVal.
In addition, could there maybe be a central doc of which warnings to ignore, in which scenarios? Or does this already exist somewhere? I feel like I'm spending a bunch of time chasing around checks that I'm unsure of the validity of. :/