Why do we include space, nb space and some control chars in extended subsets?

[m4rc1e]

These character are already covered in the non-extended subsets so I don't see the point in duplicating them?

[m4rc1e]

@garretrieger would you be opposed to me removing the following glyphs from every extended subset, https://github.com/googlefonts/nam-files/blob/main/Lib/gfsubsets/data/greek-ext_unique-glyphs.nam#L5-L8.

My issue is that gftools add-font will include the greek-ext subset for all families because our threshold for including an extended subset is only requiring 1% of characters. greek-ext only has 241 chars so 4/241=0.0165%. If I remove these glyphs, it won't trigger. If we can't remove these glyphs, I guess I could discount them from the set in gftools add-font.

[garretrieger]

Removing codepoints from a subset that have been included in pushed fonts can be risky, so if the only problem these are causing is with the tools then I'd prefer we try updating the tools first before resorting to subset changes.

[garretrieger]

I think it would be reasonable to just ignore those four characters for the purpose of subset selection as they occur frequently in various subsets and don't really signal any specific script.

[vv-monsalve]

My issue is that gftools add-font will include the greek-ext subset for all families

I think this is also the case for cyrillic-ext see gftools#829

Edit: At a revision of the Latin-ext_unique-glyphs.nam, we do want it on standard fonts as it covers what is now required under Latin Core. However, the central question about whether characters like space or nb space should be repeated on this subset remains valid.

[m4rc1e]

I'll make a fix for this today. Plan is to simply make the mod to the tools.