Re: case mappings

[Alex Shinn <alexshinn@xxxxxxxxx>]

On 7/13/05, Matthew Flatt <mflatt@xxxxxxxxxxx> wrote:
> 
> Although I agree that the 1-1 mapping is of limited use, my impression
> and experience is that it's still useful (if, for nothing else,
> interfacing to older code that uses a 1-1 mapping). So, pending further
> discussion, I will leave the character-casing operations in place.

Since we are proposing a new character based case mapping that
doesn't exist in Unicode, we need to be clear what the semantics are.

Specifically, unlike the R5RS ASCII based mapping

  (char=? (char-upcase c) (char-upcase (char-downcase c)))

is no longer a tautology (nor the reverse direction).

  (char-ci=? a b)

can no longer be defined as

  (char=? (char-upcase a) (char-upcase b))

or equivalently with char-downcase.

Also

  (string-map char-upcase s)

is not the same as

  (string-upcase s)

-- 
Alex