Just an adjustment of weights, so that "rotation" weights are lower than "fill" weights. The inherent "fill 1" and "rotation 1" can be kept. This is similar to the collation for case insensitive sorts that preserve the difference of diacritics, or sorts that swap these levels (all you have to do is to swap arithmetically the ranges of weights by a simple offset)
The DUCET seems to have given small level to to rotation variants by assigning them higher ranges so that they take prority over fil variants, and you'd like the reverse : this a basic tailoring, and not a problem of inherent values, where "fill 1" and "rotation 1" are made ignorable ni all levels (except the last implicit level on the code points in NFD form, then the optional implicit level on original code points in any non normalized form) 2015-02-24 18:38 GMT+01:00 Stephen E Slevinski Jr <[email protected]>: > Hi Unicode list, > > I am concerned that the SignWriting symbols as defined in Unicode 8 do not > sort properly. Making "*fill 1*" and "*rotation 1*" inherent values > causes sorting problems. > > Without inherent values for "*fill 1*" and "*rotation 1*", the symbols > sort properly. Consider these symbols in the correct sort order. > > symbol - fill 1 - rotation 1 > symbol - fill 1 - rotation 2 > symbol - fill 2 - rotation 1 > symbol - fill 2 - rotation 2 > > > When "*fill 1*" and "*rotation 1*" are inherent, the symbols above have > shorter names that sort incorrectly. > > symbol > symbol - rotation 2 > symbol - fill 2 > symbol - fill 2 - rotation 2 > > With the above list, "*symbol - fill 2*" will sort before "*symbol - > rotation 2*". This is incorrect. > > I believe it would fix sorting by setting the weights in the "DUCET" table > so that rotations sort before fills. If this addition was made, the > SignWriting symbols in Unicode 8 should sort properly. > > Regards, > ∼Steve > > _______________________________________________ > Unicode mailing list > [email protected] > http://unicode.org/mailman/listinfo/unicode > >
_______________________________________________ Unicode mailing list [email protected] http://unicode.org/mailman/listinfo/unicode
