Path: utzoo!mnetor!uunet!husc6!hao!boulder!sunybcs!bingvaxu!leah!uwmcsd1!ig!jade!ucbvax!CORY.BERKELEY.EDU!dillon
From: dillon@CORY.BERKELEY.EDU (Matt Dillon)
Newsgroups: comp.sys.amiga
Subject: Re: DEADKEYS
Message-ID: <8712290532.AA14690@cory.Berkeley.EDU>
Date: 29 Dec 87 05:32:40 GMT
Sender: daemon@ucbvax.BERKELEY.EDU
Lines: 46

:> *  NOTE: For some reason DeadKeyConvert()/RawKeyConvert() seem to get
:> *  confused about the REPEAT flag, therefore the flag bit should be
:> *  cleared from the qualifier set before calling this function.
:
:I suggest that the proper way to phrase this is that RawKeyConvert uses the
:REPEAT flag of the input event.  If the input event represents a repeat of
:a key, and the keymap specifies that the key is not, in fact, repeatable,
:then the resulting conversion is (correctly) of zero length.  You may, of
:course, know that you want the conversion anyway and force REPEAT to zero.
:
:If RawKeyConvert is, in fact, confused, I'd like to know more about it.
:
:- Bob

	I sent a note to Jim Mackraz on this after he asked.  Actually, I
made one mistake.... DeadKeyConvert() *does* handle the REPEAT flag 
correctly.  I thought it hadn't when DME stopped repeating the RETURN key
when I switched over to DeadKeyConvert() but as it turns out, the RETURN
key isn't supposed to repeat.  Still, it was something I wanted so I
defeated the flag before calling DeadKeyConvert() in DME.

	The two *real* bugs are:

	-(1) the new additions to the numeric keypad in the A500/A2000
	keyboard are not flagged as being in the numeric keypad.  This
	is an absolute *must* to distinguish the two without resorting
	to making assumptions on the keycodes (the keycodes are different,
	of course).

	-(2) It is difficult to use RawKeyConvert() as an inverse function
	due to it's lack of bounds checking and the duplication of many
	keys.  
	
	That is, rather than write a huge amount of code to decipher the 
	keymap, I simply call RawKeyConvert() with Code values 0 thru 127,
	and with and without the shift/numeric keypad qualifier flags set
	to create an upper/lower case keycode conversion table.  BUT since
	RawKeyConvert() doesn't return an error when I give, say, the keycode
	for the '*' on the NON-numeric keypad, but WITH the numeric keypad flag
	set, I get duplicates and have to make assumptions as to some of
	the keycodes.

	Also the RETURN key is mapped into two different keycodes.... for
	one key. 

					-Matt