JG>
  >  >>> John is implying that there's not necessarily a one-to-one
  >  >>> mapping  between lower and upper case in all character sets.
  >
  >  >> The normal way of handling that case is to not map the characters
  >  >> that have no other-case equivalent.
  >
  >  >   Consider the Greek alphabet.
  >
  > The Greek alphabet has a one-to-one mapping between lower and upper case,
  > doesn't it?
JG>

  No.  Think about the letter sigma.

  Neither does Modern French, interestingly.  All of the accented
  lower-case letters have a specific collation sequence.  But according
  to what I believe to be the latest rules, you don't accent upper-case
  characters.

  So what order words are sorted into depends from what case they are
  in.  Converting to lower case and sorting could easily give a
  different ordering to converting to upper case and sorting.

  Which is why stricmp() and its ilk are far more difficult and full of
  pitfalls than they at first seem, which was the original point.

  ... And why they never made it into an _international_ standard.

  > JdeBP <
___
 X MegaMail 2.10 #0:

--- Maximus/2 2.02