Not all code points in the Devanagari Unicode block are assigned,
but canEncode(c) returns true for all characters c in that block.

(This is a part of the fix for 5065777)

###@###.### 2004-06-22