"Indic" glyph separation breaks mkmk

[simoncozens]

In the mark feature writer, marks belonging to Indic glyphs are placed in abvm/blwm features and other marks are positioned using mark/mkmk.

However... the Sharada script (which goes through the USE and is not considered "Indic" by ufo2ft) uses the Devanagari udatta (U+0951). When a Sharada font is compiled with ufo2ft, all the Sharada mark positioning rules go into mark/mkmk and the "Indic" rules (containing udatta) go into abvm. The glyphs are separated out into Indic/non-Indic before creating these rules, and this means there is no interaction between Sharada marks and Devanagari marks; i.e. there is no way to place a udatta on a Sharada mark such as a Candrabindu.

I don't know if that's linguistically something that you might want to do, but e.g. Noto Sans Sharada allows it, positioning all marks in the mkmk feature.

I don't know how best to fix this:

• Add USE scripts to the indicScripts list?
• Provide a way of opting out of the abvm/blwm writer?

[anthrotype]

thanks for reporting this.

Add USE scripts to the indicScripts list?

that'd be ok provided that all USE scripts are supposed to use abvm/blwm, which I honestly have no idea.

Provide a way of opting out of the abvm/blwm writer?

Maybe MarkFeatureWriter could have an option indicScripts (or abvmBlwmScripts?) that defaults to the current set, and one could set it to empty to skip the abvm/blwm generation

These are somewhat related: #565 and #565

[simoncozens]

USE uses both abvm/blwm and mark/mkmk; you can choose. (See https://docs.microsoft.com/en-us/typography/script-development/use#opentype-feature-application-ii)

[simoncozens]

I think you mean #522, and yes. Behdad suggests adding all the USE scripts in there. I think it's the obvious way to do it. If a script is either USE or Indic then we know abvm/blwm will be applied.

[anthrotype]

yes that one. Let's add them all then

[simoncozens]

I'll do a PR in a minute.