Path: utzoo!attcan!uunet!zephyr.ens.tek.com!uw-beaver!mit-eddie!bloom-beacon!LIGHTNING.MCRCIM.MCGILL.EDU!mouse
From: mouse@LIGHTNING.MCRCIM.MCGILL.EDU
Newsgroups: comp.windows.x
Subject: Re:  need help to add accented letters to US keyboard
Message-ID: <9010120141.AA01273@lightning.McRCIM.McGill.EDU>
Date: 12 Oct 90 01:41:35 GMT
Sender: root@athena.mit.edu (Wizard A. Root)
Organization: The Internet
Lines: 55

> I am using X11r4 on a Sun 3/80 running SunOS 4.0.3, with a type 4 US
> keyboard.  I would like to use the Mod1 key (Meta Left key according
> to Sun terminology) combined with the Shift key to generate accented
> letters with the [ ] ; ' ` , . and / keys.

> I tried to add the following lines to my .keycode.map file:

>    keysym bracketleft=bracketleft braceleft egrave udiaeresis
[...]
> but couldn't get xmodmap to accept that. I also tried:
>    keycode 48=equal plus ediaeresis idiaeresis

> but couldn't get the keyboard to generate an accented letter, even
> though an "xmodmap -kp" would show the modified entries, i.e.
>  48  0x003d (equal)  0x002b (plus)   0x00eb (ediaeresis)  0x00ef (idiaeresis)

> I was however able to generate accented letters by redefining the
> "Shift-ed" definition of the function keys (F1 to F12) with the
> following lines in my keycode.map file:

>    keycode 12=F1 eacute
[...]

The problem is that keysyms after the first two in a key binding do not
have standard interpretations, so you're at the mercy of the client
program.  I seem to recall that there is some sort of de-facto
pseudo-standard that the third keysym is to be used when Mod1 is
active, but I don't think the default XLookupString() does this;
clients must handle it themselves.

Keyboard handling is IMO one of the weakest points of X.  I have been
mulling over some of the possibilities, as have doubtless many others,
but I have nothing concrete to suggest yet.

> Could any kind soul point me to the adequate documentation or
> directly tell me how to achieve my goal.

Toolkit clients generally accept Translation resources to allow you to
bind actions to keys; these actions can, I believe, include things like
inserting a Latin-1 accented character into a Text widget and such
things.  However, the mechanism is seriously flawed.  In particular,
each client must have its Translation resources set up separately;
there is no way to (say) set F9 to generate the string "foo" for all
clients that simply call XLookupString, or to specify that
Mod2-Shift-semicolon is to generate a capital thorn in all clients.

I suspect that the situation is bad enough that the first semi-solution
proposed will rapidly become a de-facto standard, which is part of the
reason I'm holding off proposing any of the various ideas I've had - I
don't feel they're good enough.

					der Mouse

			old: mcgill-vision!mouse
			new: mouse@larry.mcrcim.mcgill.edu