zero.onum_pnum is a fine name. Some CompareFamily tests may be too tailored to the glyph names we use at Adobe. Would be good to know exactly what's the message you're getting and which test is generating it, so that we can improve the algorithm to account for such glyph names. Thanks.
FDK’s CompareFamily is blaming me for using underscores in glyphname extensions like in zero.onum_pnum. It interprets the glyphs as ligatures like a f_l glyph and it searches for the pnum glyph without any luck. Is it just FDK that is a little bit too sensitive here or does this affect decomposition behavior in Acrobat as well?