[NiLuJe]

Okay, I hadn't yet played with the new and "improved" Bookerly, so, that was the perfect excuse to go ahead and try my hand at handling the Panose stuff like FontForge, but with fontTools instead: fix-panose.py.

And, apparently, it works .

My usual shenanigans for Bookerly, which I used to do by hand via FontForge:

fonts=("${(@f)$(ls -1 ./Bookerly-*.ttf)}")

for font in "${fonts[@]}" ; do
	gftools fix-vertical-metrics -a 802 -d -198 -l 0 "${font}"
	mv -v "${font%.*}.fix.ttf" "${font}"
	gftools nametable-from-filename "${font}"
	mv -v "${font}.fix" "${font}"
	fix-panose.py "${font}"
	mv -v "${font}.fix" "${font}"
done

(That's ZSH. Too lazy to dumb that down ;p).

And it behaves .

Sidebar, since I'm using an unhinted CFF Literata 3, and I tried the latest unhinted TTF Bookerly:

ACCESS (Monotype's iType?) *really* doesn't like unhinted fonts, CFF or TTF.
RMSDK appears to deal with unhinted CFF mostly properly (makes sense, it's Adobe), but deals with unhinted TTFs only slightly less badly than ACCESS.

TL;DR: You'll probably want to stick to hinted TTFs, or eventually unhinted CFFs if you read plain ePubs.
(Official Literata builds ship as TTF anyways).