[no subject]

["Mohammed Nasiruddin" <nasir@cs.umn.edu>]

Hi,

    We have a fairly big MACL system that is being used throughout
Honeywell. We are in the process of converting this system to run
under Kanji Talk for our Japanese users.

    As you are probably aware, Japanese characters are two bytes
long and need special i/o capabilities which are provided by
KanjiTalk and the script manager.

    I have found a problem that is holding up any progress on this
project. The problem is this:

    The read (or read-from-string) functions, when they come across
symbols, convert all characters to uppercase by subtracting 32 from
the char code. But they are not smart. They subtract 32 from Kanji
characters too which have codes greater than 128.

Example:
	I store two Kanji chars in a string (4 bytes).
	Their codes are (130 201 130 231).

	I do a READ-FROM-STRING to get back an incorrect symbol.
	The codes in the symbol-name string are (98 169 98 199).


All the other functions (such as CHAR-UPCASE, STRING-UPCASE INTERN)
seem to work correctly. READ-CHAR also returns the correct value.

The code for these functions is not accessible to me. Is there
anything you can do to help me? Any suggestions are welcome. I would
really like to avoid writing my own reader.

Thanks in advance for any suggestions (or patches).

	- Nasir

Ps: This happens in both 1.2.2 and 1.3.1