> > there is no such thing as "NFD decompositions". > > Sorry for the confusion. Still even with a NFKD decomposition,
And there is no such thing as NFKD decomposition either. It goes as follows, in steps: 1. Canonical and compatibility decomposition mappings (one-step), and canonical classes. 2. Canonical and compatibility full/recursive decompositions and canonical reordering. The compatibility (full) decompositions make use of both the canonical and compatibility decomposition mappings. 3. Canonical and compatibility equivalences. 4. The four Unicode normal forms (NFD, NFC, NFKD, and NFKC). Please don't turn it upside down, that's only confusing! Ok, the formal definition of equivalences and normal forms are a bit backwards in The Unicode standard, defining NFD (in practice, though not the name) before the equivalences. Normally, a normal form is defined as a particular representative element in an equivalence class... But there is no need to aggravate the backwardsness into cyclicity. ... > It's true that not all (only most) combining non-spacing > characters have a non-combining spacing counterpart. Only a *few* g.c. Mn characters have spacing counterparts! /kent k