You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
However, for automatic subset detection (gftools-add-font) this is a bit of a pain. Because there are so many presentation forms, if you don't support them, Arabic support falls below the threshold and does not get detected. I don't know if it's still true that lots of Arabic text is encoded using presentation forms; I don't think so, but others might know better.
The text was updated successfully, but these errors were encountered:
@nathan-williams Is there any way we could get a sense of what proportion of Arabic web pages include characters in the range U+FB50-FBFF U+FE80-U+FEFC compared to those which just use U+0600-06FE?
I’d do without Arabic Presentation Forms. Their use has been discouraged by Unicode from the day they were included and their use on the web falls into two categories 1) bad copy&paste (usually from PDF files, because PDFs are horrible at preserving text) 2) selecting a positional form in isolation, something that could be done by ZWJ if people knew any better.
Both are fringe use case IMHO and does not justify such requirement. They also can add quite a bit to file size, in one font where I measured this, they almost doubled file size (that was Arabic Presentation Forms A and B, so a subset of it might add a bit less than this). They don’t add much space for simple fonts that are built with composed glyphs.
The Arabic subset says:
https://github.com/googlefonts/glyphsets/blob/575273691bcc9c0dfc220c3c21b86bc64b052354/Lib/glyphsets/encodings/arabic_unique-glyphs.nam#L5-L8
However, for automatic subset detection (
gftools-add-font
) this is a bit of a pain. Because there are so many presentation forms, if you don't support them, Arabic support falls below the threshold and does not get detected. I don't know if it's still true that lots of Arabic text is encoded using presentation forms; I don't think so, but others might know better.The text was updated successfully, but these errors were encountered: