3 ms·
More likely it’s the effect of font subsetting. If you want your PDF to render the same everywhere, you have to embed font information (even if that font is so
by Someone 6y ago
More likely it’s the effect of font subsetting.
If you want your PDF to render the same everywhere, you have to embed font information (even if that font is something like Arial, which is available almost everywhere, as ‘almost everywhere’ isn’t ‘everywhere’, and because there are zillions of variations on Arial)
You don’t want to embed entire fonts, though, certainly not if a PDF uses only a few characters of a font.
The memory wise cheapest way of embedding a font leaves out the table that maps glyph numbers back to Unicode code points. If you do that, the PDF reader guesses a 1:1 mapping to ascii/Unicode.
Subletting means you can’t extract all glyphs in a font from a single file, but AFAIK, that’s just a side effect.
I guess this bug drops such a table, or messes up that translation table.