On N4078 "Revised Proposal to enable the use of Combining Triple Diacritics in Plain Text"... (http://std.dkuug.dk/JTC1/SC2/WG2/docs/n4078.pdf)
There was an alternative way to introduce diacritics spanning multiple letters. It was possible to duplicate COMBINING XXXX diacritics with CONJOINING XXXX diacritics set with next simple rules: 1) CONJOINING XXXX diacritics yields the whole conjoining diacritics if it follows the same type COMBINING XXXX or CONJOINING XXXX diacritics*; 2) in all other cases CONJOINING XXXX diacritics behaves exactly like COMBINING XXXX one. The rule #1 looks like modification of COMBINING XXXX behavior, but legacy compatibility would be preserved. This would eliminate necessity in closing diacritics and in most cases necessity in opening diacritics as well, if one uses CONJOINING XXXX diacritics only. The only case when COMBINING XXXX diacritics would be really necessary in this scheme is two adjacent, but not conjoined blocks with the same XXXX diacritics — in this case the only way to start a new block would be using COMBINING XXXX diacritics. Everything encoded with COMBINING XXXX diacritics would be kept intact in this scenario. E.g. if we have the word ABCDEFGH, where we have (ABC)(DE)F(G)H distribution of macrons, it could be encoded as: <character A> CONJOINING MACRON / COMBINING MACRON <character B> CONJOINING MACRON <character C> CONJOINING MACRON <character D> COMBINING MACRON <character E> COMBINING MACRON <character F> <character G> CONJOINING MACRON / COMBINING MACRON <character H> Maybe I am wrong, but for me it seems to be more simple for the end-user. Best regards, Alex.
