> If "hiding default ignorable" means substituting by zero-width space, > then the pdf file produced by xetex would not contain U+115F and > U+1160, resulting in invalid text string when texts are extracted from > the pdf.
Not exactly. As far as I understand, the correct output for UnBatang font should be [uni115F.ljmo04=1+1000|uni1161.vjmo02=1+0|uni112B.ljmo05=2+1000|uni1160.vjmo02=2+0] And if we omit default ignorables that are part of a valid cluster, then the output for JieubsidaBatang font would be [uni115F.lj4=1+0|uni1161.nt=1+833|uni112B.lj5=2+0|uni1160.nt=2+833] w/o having to "accumulate" the overall cluster advance in the cluster base. > Only isolated filler may be ignored. Anyway this is illegal input string. Indeed. That's why "placeholder" characters like Hangul fillers are default_ignorables. However, we have an option to preserve default ignorables for the case when the rendering system is able to show something useful for a broken input text or missing glyphs (i.e. to show the vowel on top of a dotted square to mention the fact that it is missing a leading consonant before, etc.). regards, Konstantin _______________________________________________ HarfBuzz mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/harfbuzz
