It looks like the problem you are describing lies in the interaction between the
keyboard and the program that handles keystrokes, not in the character encoding.
If the design is such that <a umlaut> should be entered by typing <a> then
<umlaut> without any special marking then you will have the described behaviour
no matter whether the character encoding is Unicode or 8859-1 or anything else.
On the other hand you may still design your interaction to type <a umlaut> as
one keystroke or using the dead key sequence - and use Unicode encoding in any
normalization form you wish (i.e. your keyboard handler will convert the
keystroke sequence to the character code sequence).
This archive was generated by hypermail 2.1.2 : Tue Jul 10 2001 - 17:20:53 EDT