Support public.skipExportGlyphs in font.info

[benkiel]

Could makeotf support the public.skipExportGlyphs in font.info? See discussion of it here: unified-font-object/ufo-spec#84 (and, yes, we need to add this to defcon still...)

[josh-hadley]

Could you elaborate a bit on how you'd like to see this supported in makeotf (or elsewhere in AFDKO)?

[benkiel]

@josh-hadley Sorry for the slow reply. In brief, it would be great if makeotf would look for public.skipExportGLyphs if it ingests a UFO and if present it removes those glyphs from the generated OTF after decomposing the UFO.

[frankrolf]

I am curious how this would correspond with the GlyphOrderAndAliasDB’s -gs option (which excludes glyphs not mentioned within).
Which of the two methods would take precedence?

[benkiel]

Hmmmmm, that's a fair point. Use case for this in makeOTF is likely low, so feel free to close.

[miguelsousa]

Which of the two methods would take precedence?

The option would take precedence over the key. I think it's the approach we've taken in other comparable cases. See #765 (comment)

[josh-hadley]

If/when we implement this, we should be sure to document which takes precedence and maybe issue a warning at runtime when both the public.skipExportGlyphs key and GOADB with -gs option are specified to avoid users making the wrong assumption.

[twardoch]

My view is that since skipExportGlyphs travels with the font source, it really is informative about what constitutes the "full glyph repertoire" of the final font. The glyphs in skipExportGlyphs are simply "not intended" to be included in the binary font at all.

The -gs option is a positive filter for creating subsets, kind of equivalent to pyftsubset.

But I agree that the -gs option should take precedence and "turn off" skipExportGlyphs because there is a valid use case that you may want to include the otherwise skipped glyphs in GOADB.

Basically, in a UFO worfklow, GOADB is a mechanism for overriding specifications found in the UFO — glyph set, glyph names Unicode codepoint assignments. Right? So it always "turns off" an aspect present in the source file.