[subset] Default to all glyphs?

A lot of people use our subsetter to simply convert ttf to woff/woff2. That’s a lot of work for such a conversion! Ideally we’d have a fonttools.convert module that wouldn’t incur the subsetting cost (and the surprises that come with subsetting) doing that.

But then sometimes people also want to provide --no-hinting and other options for such operations. So yeah, this borders on subsetting, so maybe is fine that subset module is being used for all kinds of conversions. We can even have a very light fonttools.convert module that calls subsetter if necessary…

In such uses it’s common for users to pass --glyphs='*'. If no glyph or Unicode codepoints is provided, I’m not sure what the subsetter does currently, but it feels like would be nice defaulting to all glyphs. Though, one can argue that (and I think I agree!) we should default to all Unicodes and throw out unreachable glyphs.

[anthrotype]

changing the current default values for the subsetter is not advisable, in my opinion, as it would break all existing scripts or tools that use it without explicitly setting all values (i.e. a lot)

[behdad]

why throw out glyphs without unicodes by default?

That’s not what I said. 😃 I said “unreachable glyphs”. Those are glyphs that are not accessible via any feature…

Marc is correct that by default the features pyftsubset keeps are features are used by default in shaping. Ie. any extras are not retained.