xkeycaps.man man page

xkeycaps — graphically display and edit the X keyboard mapping


xkeycaps [-toolkitoption ...] [-option ...]


The xkeycaps program displays a keyboard.  Moving the mouse over a  key describes the keysyms and modifiers that that key generates.  Clicking left on a key simulates a KeyPress event.  Clicking right on a key brings up a menu of operations, including a command to change the keysyms that the key generates.  This program is, in part, a graphical front-end to xmodmap(1).


xkeycaps accepts all of the standard toolkit options, and also accepts the following options:

-keyboard keyboard-name or -kbd keyboard-name

Specifies the type of keyboard to display.  There are many different computer keyboards in the world, and xkeycaps must know which one you are using  in order to function correctly.  Case does not matter when specifying a  keyboard name.

If you're running on the console display of a Sun or HP, then xkeycaps will interrogate the attached keyboard hardware directly to determine what keyboard you're using.  But if you're running remotely, or on another type of machine, then you must specify a keyboard somehow.


Lists the recognized values for the -keyboard option.

-gutterwidth number or -gw number

Specifies the number of pixels of space to leave between each key.

-font fontname

Specifies the font to use to display the keycaps. The following standard X Toolkit command line arguments are commonly  used with xkeycaps:

-display host:dpy

This option specifies the X server to contact.

-geometry geometry

This option specifies the preferred size and position of the window.

-bg color

This option specifies the color to use for the background of the window. The default is light gray.

-fg color

This option specifies the color to use for the foreground of the window. The default is black.

-bw number

This option specifies the width in pixels of the border surrounding  the window.

-xrm resourcestring

This option specifies a resource string to be used.  This is especially useful for setting resources that do not have separate command line options.


The bottom part of the window is a drawing of a keyboard.  In the top left of each key is printed the string which actually appears on the surface of the key.  In the bottom right of the key is the (hexadecimal) keycode that this key generates.

At the top of the screen are several lines of text describing the key under the mouse (or the most recently typed key.)  These lines are:


This displays the text printed on the physical key, and the keycode generated by that key in hex, decimal, and octal.


This displays the set of KeySyms that this key currently generates.  


This displays the ASCII equivalent of this key, taking into account the current modifier keys which are down.


this displays the modifier bits which this key generates.  If a key generates modifiers, it is a chord-key like Shift or Control.


Whether the X server claims that this key autorepeats.  I say “claims” because the OpenWindows X server is the only one I have encountered for which this information is accurate.  The per-key autorepeat flag seems to be almost-universally ignored.


There are several buttons in the upper left corner of the window.  They are:


Exits the program.

Select Keyboard

Pops up a dialog box from which you can change which keyboard is displayed. The left column lists the known types of keyboards, and the right column lists the known layouts (mappings) of those keyboards.

Type At Window

After selecting this, you are asked to click on some other window.  After doing this, clicking on keys on the keyboard display will simulate key events on the window you selected.  Selecting the root window or the xkeycaps window turns this off.

If you are using a window manager (for example, twm(1)) in which you can lock the keyboard focus on a window and still click on  other windows without having the focus change, then you can accomplish  the same thing merely by focusing on another window and clicking on the  keys in the xkeycaps window.

Restore Default Map

This command restores the keyboard to its default state.  If you execute this command while displaying a keyboard which is not the type of keyboard you are really using, your keymap will be in a nonsensical state.  There is no way for xkeycaps to tell what keyboard you are using except by taking your word for it, so don't lie.

Write Output

This command writes an xmodmap input file representing the current state of the keyboard (including all of your changes) to a file in your  home directory.  

The file will be called .xmodmap-hostname, where hostname is the name of the machine you're running on. It will warn you if the file already exists.

It prompts you with a dialog box: you can either write an xmodmap  file representing the state of every key, or you can write a smaller file which describes only the changes.

The idea is that in the appropriate startup script, you would add a line like

xmodmap  /.xmodmap-`uname -n`

in the appropriate init file, so that those keyboard modifications are made each time you log in.  (If you're not sure where this command should go, ask your system administrator, as that tends to vary from site to site.)

Clicking left on a key simulates a KeyPress event.  Releasing the button simulates a KeyRelease event.  If you click left on a key and move the mouse while the button is down, KeyPress and KeyRelease events  will be simulated on every key you move the mouse over.  Think of the mouse  as your finger: if you drag the mouse over several keys, they will go down and up in turn.

Clicking left on a key which is associated with modifier bits (such as Shift or Control) causes that key to “lock” down.  Clicking left again releases the key.  In this way, you can generate key-chords with the mouse: to generate Control-C, click left on the Control key, and then  click on the C key.  Click on Control again to turn the control modifier off.

Typing a key on the real keyboard simulates  a KeyPress/KeyRelease event pair in the same way that clicking on a key does.

You can also combine mouse and keyboard input: for example, if you use the mouse to select the Shift key, and type a character, the event that is simulated will have the Shift modifier set.  And if you hold down the real Control key, and click on the C key in the window, a Control-C event will be generated.  (Assuming, that  is, that your window manager does not intercept control-left-button for its  own purposes.)

Clicking right on a key pops up a menu of commands for the given key. They are:

Exchange Keys

After selecting this menu item, you are asked to click on another key. That key and the key on which you brought up the menu will be exchanged. This changes the keyboard mapping immediately.

Duplicate Key

After selecting this menu item, you are asked to click on another key. That key will be made a copy of the key on which you brought up the menu. That is, the two keys will generate the same set of keysyms and modifiers. This changes the keyboard mapping immediately.

Disable Key

The key on which you brought up the menu will be made to generate no keysyms and no modifiers.  This changes the keyboard mapping immediately.

Restore Key To Default

The key on which you brought up the menu will be restored to its default state; no other key will be altered.  This actually changes the current keyboard mapping.

Edit KeySyms of Key

This pops up the "Edit Key" window, which allows you to arbitrarily change which keysyms and modifiers this key generates.

On the left side of the window is the list of the keysyms that this key currently generates.  (A key may generate up to eight keysyms; the  interpretation of these keysyms is described in the X protocol document, and is summarized later in the Keysyms and Keycodes section of this man page.)

The second column is a multiple-choice list of the eight modifier bits that this key may generate.  For example, if you want a key to behave as a “control” key, you should select the Control modifier.

The third and fourth column (the scrolling lists) are for changing the  keysym associated with the key.  When you select a keysym-position from  the first column, the character set and keysym will be displayed in the scrolling lists.  Clicking on a keysym in the “KeySym” column will install that keysym in the highlighted slot in the first column.

To select a keysym from a different character set, click on the character set name in the second column.  (The Latin1 and Keyboard character sets are the most commonly used.)

At the bottom of the window are three buttons: Undo, Abort,  and Ok.  Clicking on Undo reverts the Edit Key window to the current state of the key in question.  Abort closes the Edit Key window without making any changes.  Ok closes the Edit Key window and installs your changes (the current keyboard mapping is modified.)

Keysyms and Keycodes

To effectively edit your keyboard mapping, there are some terms you need to be familiar with:


This is a raw scan-code that is read from the keyboard; each physical key on  the keyboard has a different number associated with it; this mapping cannot be changed (but that's ok.)

Generally, every keyboard has its own set of KeyCodes, which is why you will probably need to have a different keymap for every system you use.


This is a symbol which can be generated by a single press of one key on the keyboard: for example, all letters, numbers, and punctuation are keysyms, and so are more abstract things like “shift” and “control”.

Each KeyCode (that is, key on the keyboard) is associated with certain  KeySyms.  The KeySyms are what give the keys their semantics (and makes the A key generate an A), not the KeyCodes.

Usually keys are associated with one or two keysyms, which correspond to the symbols generated when the key is pressed, and when it is pressed while the shift key is held down.  There is a special case, which is that if the key contains only one KeySym, and it is a letter, then the Shift key does the obvious thing that one does to letters.


Not to be confused with KeySyms, this refers to the text which is printed on the physical keys: it is immutable (unless you repaint your keyboard...)


This term refers to a set of two or more keys held down simultaniously (by analogy with piano keyboards.)  All but one of the keys will generally be Modifier Keys.  Sometimes Constellation is used to mean the same thing.

Modifier Key

This is a key like shift or control, which is used to alter the interpretation of other keys which are held down at the same time.  Generally, pressing a modifier key without also pressing a non-modifier key does nothing.

A key is a modifier key if it has a Modifier KeySym on it.  (More specifically, if the KeyCode of that key is associated with a Modifier KeySym.)

Modifier KeySym

A KeySym is a modifier keysym if it has a Modifier Bit associated with it. But, the rules are a little more complicated than that.  It's easier to  describe by example:

For a key to behave as one expects a shift key to behave, the keycode should have the Shift modifier bit set; and the key should generate one of the keysyms Shift_L and Shift_R.  If either of these is not true, the key will not behave as a shift key.

Analogously, a control key must have the Control modifier set, and use one of the keysyms Control_L or Control_R.

This implies that if you wanted to swap your shift and control keys, it would not be enough to simply swap their modifier bits: you must swap their keysyms as well.  If you only swap the modifier bits, it might appear to work at first, but other things would malfunction.

Keys like Meta (and Super, Hyper, etc.) are a bit more complicated (see below.)

Modifier Bit

Modifier bits are attributes which certain keysyms can have.  Some modifier bits have predefined semantics: Shift, Lock, and Control.  The remaining modifier bits (Mod1 through Mod5) have semantics which are defined by the  keys with which they are associated.

That is, the Control modifier means Control if it is attached to Control_L or Control_R, and is illegal elsewhere.

But Mod1 means Meta if it is attached to Meta_L or Meta_R; but it would mean Alt if it were attached to Alt_L or Alt_R; or Hyper with Hyper_L or Hyper_R; and so on.  (It could not, however, be attached to Control_L, since the  Control modifier has already spoken for those keysyms.)

If you're thinking that this is all senselessly complicated... you're right.

X Protocol Document on Keymaps

The following is a more precise technical explanation of how keymapping works. This description is from the X Protocol document, and is reprinted here for your convenience:

A list of KeySyms is associated with each KeyCode.  If that  list (ignoring trailing NoSymbol entries) is a single KeySym “K”, then the list is treated as if it were the  list “K NoSymbol K NoSymbol”.   If the list (ignoring trailing NoSymbol entries) is a pair  of KeySyms “K1 K2”, then the list is treated as if it were the  list “K1 K2 K1 K2”.  If the list (ignoring trailing NoSymbol  entries) is a triple of KeySyms “K1 K2 K3”, then the list is treated as if it were the list “K1 K2 K3 NoSymbol”.

The first four elements of the list are split into two groups of KeySyms.  Group 1 contains the first and second KeySyms,  Group 2 contains third and fourth KeySyms.  Within each group, if the second element of the group is NoSymbol, then the group should be treated as if the second element were the same as the first element, except when the first element is an alphabetic KeySym “K” for which both lowercase and uppercase forms are defined.  In that case, the group should be treated as if the first element were the lowercase form of “K” and the second element were the uppercase form of “K”.

The standard rules for obtaining a KeySym from a KeyPress event make use of only the Group 1 and Group 2 KeySyms; no interpretation of other KeySyms in the list is given here.  (That is, the last four KeySyms are unused.)

Which group to use is determined by modifier state.  Switching between groups is controlled by the KeySym named Mode_switch.

By attaching that KeySym to some KeyCode and attaching  that KeyCode to any one of the modifiers Mod1  through Mod5.  This modifier is called the “group modifier”.   For any KeyCode, Group 1 is used when the group modifier is off,  and Group 2 is used when the group modifier is on.

Within a group, which KeySym to use is also determined by modifier state.  The first KeySym is used when the Shift  and Lock modifiers are off.  The second KeySym is used when  the Shift modifier is on, or when the Lock modifier is on  and the second KeySym is uppercase alphabetic, or when the Lock modifier is on and is interpreted as ShiftLock.  Otherwise, when the Lock modifier is on and is interpreted as CapsLock, the state of the Shift modifier is applied first to select a KeySym, but if that KeySym is lowercase alphabetic, then  the corresponding uppercase KeySym is used instead.

Icccm on the Modifier Mapping

The following is a more precise technical explanation of how modifier keys are interpreted.  This description is from the Inter-Client Communications  Conventions Manual, and is reprinted here for your convenience:

X11 supports 8 modifier bits,  of which 3 are pre-assigned to Shift, Lock and Control.  Each modifier bit is controlled by the state of a set of keys, and these sets are specified in a table accessed by GetModifierMapping() and SetModifierMapping().

A client needing to use one of the pre-assigned modifiers should assume that the modifier table has been set up correctly to control these modifiers. The Lock modifier should be interpreted as Caps Lock  or Shift Lock according as the keycodes in its controlling set include XK_Caps_Lock or XK_Shift_Lock.

Clients should determine the meaning of a modifier bit from the keysyms being used to control it.

A client needing to use an extra modifier,  for example Meta,  should:

Scan the existing modifier mappings. If it finds a modifier that contains a keycode whose set of keysyms includes XK_Meta_L or XK_Meta_R, it should use that modifier bit.

If there is no existing modifier controlled by XK_Meta_L  or XK_Meta_R, it should select an unused modifier bit (one with an empty controlling set) and:

If there is a keycode with XL_Meta_L in its set of keysyms, add that keycode to the set for the chosen modifier, then

if there is a keycode with XL_Meta_R in its set of keysyms, add that keycode to the set for the chosen modifier, then

if the controlling set is still empty,  interact with the user to select one or more keys to be Meta.

If there are no unused modifier bits, ask the user to take corrective action.

The above means that the Mod1 modifier does not necessarily  mean Meta, although some applications (such as twm  and emacs 18) assume that.  Any of the five unassigned modifier bits  could mean Meta; what matters is that a modifier bit is generated by a  keycode which is bound to the keysym Meta_L or Meta_R.

Therefore, if you want to make a “meta” key, the right way is to make the keycode in question generate both a Meta keysym, and some previously-unassigned modifier bit.

The Mode_switch Keysym

In case the above didn't make sense, what the Mode_switch keysym does  is, basically, act as an additional kind of shift key.  If you have four  keysyms attached to the A key, then those four keysyms will be accessed by  the chords: A; Shift-A, Mode_Switch-A; and Mode_Switch-Shift-A, respectively.

Like any Modifier Key, for Mode_switch to function, it must have a modifier bit attached to it.  So, select one of the bits Mod1 through Mod5 (whichever is unused) and attach that to the Mode_switch key.

The Multi_key Keysym

Not to be confused with Mode_switch, Multi_key allows the input of multiple character sequences that represent a single character (keysym.) A more traditional name for this keysym might have been Compose.

The Multi_key keysym is not a modifier keysym.  That is, for it to function properly, it should not have any modifier bits associated with it.  This is because it is not a “chording” key: you do not hold it down along with other keys.  Rather, you press Multi_key, then release it, then press and release another key, and the two together yield a new character.  

For example, one traditional binding would be for Multi_key, followed by single-quote, followed by A to produce  the Aacute keysym.

Not all vendors support the use of the Multi_key keysym; in particular, Digital, Sun, and HP support it, but the X Consortium does not.  (The reason for this, I am told, is that “Compose” sequences are considered obsolete; the party line is that you should be using Input Methods to do this.)

Whether Multi_key works is a property of the Xt library (not the X server) so it's possible that on a single system, Multi_key might work with some applications and not others (depending on how those applications were compiled and linked.)

If you use Lucid Emacs or XEmacs, then you can take advantage of Multi_key  sequences even if your version of Xt doesn't support it, by loading  the x-compose library, which simulates the traditional Xt behavior.   For more info, read the commentary at the top of the  file "/usr/local/lib/xemacs-*/lisp/x11/x-compose.el".

Dead Keysyms

Dead keys work similarly Multi_key, but they are two-keystroke commands instead of three.  For example, pressing the Dead_tilde key,  releasing it, then pressing the A key would generate the single  keysym Atilde.  (They are called “dead” keys because they do not, by themselves, insert characters, but instead modify the following character typed.  But HP likes to call them “mute” instead of “dead,” no doubt to avoid frightening the children.)

Again, these are not supported by all versions of the Xt library (but can be simulated by XEmacs.)

Also note that different vendors have different names for the dead  keysyms.  For example: depending on your vendor, X server version,  and the phase of the moon, you might find that the name of “dead-tilde”  is Dead_Tilde, Dtilde, SunFA_Tilde, SunXK_FA_Tilde, DXK_tilde,  DXK_tilde_accent, hpmute_asciitilde, hpXK_mute_asciitilde, or even  XK_mute_asciitilde.  It's a mess!  You'll have to just try it and see what works, if anything.

Things You Can't Do

People often ask if xkeycaps or xmodmap can be used to make one key generate  a sequence of characters.  Unfortunately, no: you can't do this sort of  thing by manipulating the server's keymaps.  The X keyboard model just  doesn't work that way.

The way to do such things is to set translation resources on particular  widgets.  It has to be done on an application-by-application basis.  For  example, here's how you would convince xterm(1) to insert the string next when you hit F17:

xterm*VT100.Translations: #override \
    <Key>F17: string("next")

Other applications may have different mechanisms for accomplishing the same thing, and some applications might not support it at all.  Check the relevant man pages for specifics.

Likewise, you can't convince one key to generate another key with  modifiers (for example, you can't make F1 behave like Ctrl-A except by  using translations, as above.)

It is also not possible to make a keyboard key behave as a mouse button.

Loser Vendors

Both HP and S.u.S.E. ship their systems with broken keyboard settings by default.  They really should know better, but they don't.

As explained above, it is undefined behavior for one modifier bit to be shared between two keys with dissimilar semantics.

By default, HP uses Mod1 for both Meta and Mode_switch. This means that it's impossible for a program to tell the difference between, for example, Meta-X and Mode_switch-X.

So, to repair this mess, you need to give the Mode_switch key a different modifier bit (mod2, for example.)  Or, you could just remove it from the keymap entirely.

S.u.S.E. Linux is even more screwed up than HP: whereas HP's default keymap contains only one bug, S.u.S.E.'s default map contains three completely different errors!

First, their default keymap has the Control modifier attached to both  the Control key and the Multi_key.  This is completely crazy,  because not only is Multi_key not a control key, it's not even  a chording key!  It mustn't have any modifier bits attached to it  at all.

Second, they attach Mod1 to Meta_L and also to Alt_R. Some people think that “meta” and “alt” are synonyms, but the fact is that the X Window System does not agree.   Those are distinct keys. It's possible to have both “meta” and “alt” keys on the keyboard at the same time, and to have programs interpret them distinctly. But of course only if they don't bogusly share the same modifier bit, making the interpretation of that bit be ambiguous.

Third, they attach Mod5 to both Scroll_Lock and  to Hyper_R, which is wrong for reasons that should by now be obvious.

The easiest way to fix your S.u.S.E. configuration is to: remove control from Multi_key; change the left Alt key to generate Alt_L instead of Meta_L; and delete the Hyper_R keysym from the keyboard.

If you have any pull with these vendors, I encourage you to encourage them to get their act together.

X Resources

XKeyCaps understands all of the core resource names and classes as well as:

*Keyboard.keyboard (class Keyboard)

Which keyboard to display; this is the same as the -keyboard  command-line option.  If this is not specified, the default keyboard is guessed, based on the server's vendor identification string.

*Keyboard.Keyboard.selectCursor (class Cursor)

The cursor to use when selecting a key or window with the mouse. The default is the crosshair cursor.

*Keyboard.Key.highlight (class Background)

The color to use to highlight a key when it is depressed.  If this is the same as the background color of the key, it is highlighted with a stipple pattern instead.

*Keyboard.Key.keycapColor (class Foreground)

The color to paint the keycap string.

*Keyboard.Key.keycodeColor (class Foreground)

The color to paint the keycode number.

*Keyboard.Key.borderColor (class Color)

The color of the box around each key.

*Keyboard.Key.keycapFont (class Font)

The font to use to draw the keycap string.

*Keyboard.Key.keycodeFont (class Font)

The font to use to draw the keycode number.

*Keyboard.Key.borderWidth (class Int)

The thickness of the box around each key.

*Keyboard.Key.gutterWidth (class Int)

How many pixels to leave between this key and it's neighbors to the right and bottom.

The class of each key widget is “Key,” as you see above.  The name of each key is the string(s) printed on its face.  So if you wanted (for example) the Shift keys to have wider borders, you could specify that with

xkeycaps*Keyboard.Shift.borderWidth: 2


It is possible to rebind the actions which happen when a key or mouse button is pressed or released.  These actions are available on the Keyboard widget:

HighlightKey(condition, arg)

This places the key in question in the highlighted state.

If no argument is passed to this action, then the key is determined by the event which invoked this action.  If this action is invoked by a KeyPress or KeyRelease event, the key-widget is the key corresponding to the key that the event represents.  If it is  a ButtonPress, ButtonRelease, or PointerMotion event,  then the key-widget is the one under the mouse.

The argument may be one of the words mouse, highlighted, or displayed, meaning the key under the mouse, the key most recently highlighted, or the key currently being described in the “Info” area at the top of the window, respectively.

The condition may be one of the words ifmod, unlessmod, iftracking, unlesstracking, ifhighlighted, or  unlesshighlighted.  If ifmod was specified and the key in question (as determined by the argument or by the invoking event) is  not a modifier key, then this action is not executed.  The unlessmod  condition is the opposite.  The iftracking and unlesstracking  conditions allow you to do some actions only if (or unless) the key is  being “tracked” with the mouse (see below.)  The ifhighlighted and unlesshighlighted actions allow you to do some things only if (or unless) the key in question is currently in the highlighted state.

UnhighlightKey(condition, arg)

This places the key in question in the unhighlighted state.  Arguments are as above.

ToggleKey(condition, arg)

This makes the key be highlighted if it is unhighlighted, or unhighlighted  if it is highlighted.  Arguments are as above.

SimulateKeyPress(condition, arg)

This action makes a KeyPress event corresponding to the key be synthesized on the focus window.  Arguments are as above.

SimulateKeyRelease(condition, arg)

This action makes a KeyRelease event corresponding to the key be synthesized on the focus window.  Arguments are as above.

TrackKey(condition, arg)

This makes the key in question begin being “tracked”, which means that moving the mouse off of it will simulate a button-release action, and then will simulate a button-press action on the key that the mouse has moved on to.  This action may only be invoked from a ButtonPress or ButtonRelease event.

UntrackKey(condition, arg)

This makes the key in question no longer be “tracked.”

DescribeKey(condition, arg)

This action causes the key and its bindings to be displayed in the “Info” section at the top of the window, if it is not already described there.

The default actions for the Keyboard widget are:

<Motion>:   DescribeKey(mouse,unlessTracking)      \n\
<KeyDown>:  HighlightKey()                         \
            DescribeKey(unlessMod)                 \
            DescribeKey(displayed)                 \
            SimulateKeyPress()                     \n\
<KeyUp>:    UnhighlightKey()                       \
            DescribeKey(displayed)                 \
            SimulateKeyRelease()                   \n\
<Btn1Down>: HighlightKey(unlessMod)                \
            ToggleKey(ifMod)                       \
            TrackKey(unlessMod)                    \
            SimulateKeyPress(ifHighlighted)        \
            SimulateKeyRelease(unlessHighlighted)  \n\
<Btn1Up>:   UntrackKey(highlighted)                \
            SimulateKeyRelease(highlighted,unlessMod) \
            UnhighlightKey(highlighted,unlessMod)  \n\
<Btn3Down>: XawPositionSimpleMenu(keyMenu)         \
            MenuPopup(keyMenu)                     \n

If you don't want a key to be described each time the mouse moves over it, you can remove the <Motion> action.  In that case, you should probably add DescribeKey() to the <Btn1Down> and <KeyDown> actions.

If you want the key under the mouse to be described even while the mouse is moving with a button down, then remove the unlessTracking parameter from the DescribeKey action bound to <Motion>.

If you don't want the modifier keys to toggle, then change the Button1 actions to

xkeycaps*Keyboard.actions:  #override               \
        <Btn1Down>: HighlightKey()                  \
                    TrackKey(unlessmod)             \
                    SimulateKeyPress()              \n\
        <Btn1Up>:   UntrackKey(highlighted)         \
                    SimulateKeyRelease(highlighted) \
                    UnhighlightKey(highlighted)     \n

Remember that these actions exist on the Keyboard widget, not on the Key widgets.  If you add actions to the Key widgets, things will malfunction.



to get the default host and display number.


to get the name of a resource file that overrides the global resources stored in the RESOURCE_MANAGER property.


to get the location of the XKeysymDB file, which lists the  vendor-specific keysyms.  


The latest version can always be found at  http://www.jwz.org/xkeycaps/

See Also

X(1), xmodmap(1), xset(1), xdpyinfo(1)


Because this program has default colors that aren't "black" and "white", the -rv command-line option doesn't work.  But the incantation

% xkeycaps -fg white -bg black -bd white

will do what you want on a monochrome screen.

The NeXT default map is believed to be incorrect; someone with access to  a NeXT will need to debug this.

There is no portable way to be sure what keyboard is being used; this means it will often not default to the correct one, and if the user makes changes to the keymap while displaying a keyboard which is not the right one, very bad things can happen.

If you depress more than a dozen keys at a time, many X servers get confused, and don't transmit enough KeyRelease events; the result of this is that the xkeycaps keys will get “stuck” until they are pressed again. (Don't go like that.)

The AutoRepeat flag is apparently useless on all X servers except the OpenWindows one (I've never seen another server that didn't ignore it.)

You don't get to select from the set of Vendor keysyms (those keysyms which are defined in the XKeysymDB file) unless you're running X11r5 or newer.

NCD's non-US keyboards do not use the standard R4/R5 mechanism for attaching more than two keysyms to one key; instead of simply having three or four  keysyms attached to the keycode in question, the Compose key changes the actual keycode of the key (it turns its high bit on.)  The xkeycaps program doesn't really understand this.  Someone from NCD support told me that in future releases they will do things the R4/R5 way instead of the way they do things now, so hacking xkeycaps to understand the current behavior is  probably not worth the effort.

The Type at Window command doesn't seem to work on the WreckStation version of XTerm.  I assume some variation of the normal XTerm's Allow SendEvents command is necessary.

If you can't select anything from the right-button popup menu, it might be because you have NumLock or CapsLock down.  I'm not sure how to fix this, it seems to be some dumb Xt thing.

If the popup menu is always greyed out, or doesn't correspond to the key  that you clicked on, it might be because you're running xswarm, an old version of xautolock, or some other program that antisocially interferes with event-propagation.  (Don't go like that.)

Because of the nonlinear way in which this program uses XLookupString, there's no sensible way for it to do compose processing, and show you the  results of “dead” key or Multi_key sequences.

It needs to know about more keyboard types (and no doubt always will...)

L-shaped keys aren't drawn accurately.  We should use the Shape extension for that.

In addition to displaying the ASCII version of the given character, it should display the corresponding character in the character set (Latin2, Kana, Greek, etc.)  This would require having fonts for all of those  character sets, though, and as far as I can tell, they don't all come standard.

When running on a Sun and talking to an OpenWindows server, we should parse the appropriate file from $OPENWINHOME/etc/keytables/ to determine the default keymap.  No doubt there are system-specific ways of doing this in other environments as well.

The HP C compiler complains about "invalid pointer initialization" in the header files.  This is a bug in that compiler, not in xkeycaps.  This compiler bug goes away if you invoke HP's cc with the the -Aa (ANSI) option.

The xmodmap program still sucks.  Since its ADD and REMOVE directives take keysyms as arguments instead of keycodes, there are things that you can do with XKeyCaps that you can't represent in an xmodmap script (at least, not without great pain.)

The xmodmap program has no commands for changing the autorepeat status of keys, so that information is not written in the output.  Perhaps we could write out an appropriate xset command instead.  (For example, to turn on autorepeat on PgUp (which happens to have key code 103) on Solaris, you would do: "xset r 103".)

Some versions of OpenWound use a nonstandard mechanism for specifying which keys have toggle (lock-like) behavior (whereas most other X servers base this behavior on the keysym: if Caps_Lock or Shift_Lock is generated, the key locks, otherwise it does not.)  XKeyCaps doesn't know how to change the lock status on these servers.  This is because I don't know how, either. If you know what system calls are necessary to hack this behavior, tell me.

The XKB interface of X11R6 looks to provide most of the information which xkeycaps needs to know, but I haven't had time to investigate this yet.


Jamie Zawinski <jwz@jwz.org>, 10-nov-91.

Please send me any changes you make!  Especially new keyboards.  The strength of this program lies in the fact that it knows about so many different keyboards, thanks to the hundreds contributions I've received over the years.  If you have to make your own modifications, please do your part! Send the changes back to me so that I can incorporate them into a future release.

Thanks to:

Jonathan Abbey, Alon Albert, Vladimir Alexiev, David Arnold, David Axmark, Ruediger Back, Pedro Bayon, Corne Beerse, Eric Benson, Christoph Berg, Markus Berndt, Roger Binns, Stefan Bj|rnelund, black@westford.ccur.com, Mark Borges, Volker Bosch, Dave Brooks, Lorenzo M. Catucci, Michel Catudal, Francois Regis Colin, John Coppens, Cesar Crusius, Bart Van Cutsem, Matthew Davey, Christopher Davis, Albrecht Dress, Kristian Ejvind, Michael Elbel, Joe English, Eric Fischer, Morgan Fletcher, Olivier Galibert, Carson Gaspar, Andre Gerhard, Daniel Glastonbury, Christian F. Goetze, Dan R. Greening, Edgar Greuter, John Gotts, Berthold Gunreben, Jens Hafsteinsson, Adam Hamilton, Magnus Hammerin, Kenneth Harker, Ben Harris, Mikael Hedin, Tom Ivar Helbekkmo, Mick Hellstrom, Neil Hendin, Andre Heynatz, Mike Hicks, Alan Ho, Hide Horiuchi, Dirk Jablonowski, Alan Jaffray, Anders Wegge Jakobsen, Chris Jones, Jorgen Jonsson, Peter Kaiser, Heikki Kantola, Tufan Karadere, Benedikt Kessler, Philippe Kipfer, Edwin Klement, John Knox, Haavard Kvaalen, Frederic Leguern, Simon Leinen, Michael Lemke, Tor Lillqvist, Torbj|rn Lindgren, Tony Lindstrom, Richard Lloyd, Ulric Longyear, Ulf Magnusson, Cliff Marcellus, John A. Martin, Tom McConnell, Grant McDorman, Hein Meling, Jason Merrill, Aleksandar Milivojevic, Manuel Moreno, Ken Nakata, Pekka Nikander, Todd Nix, Leif Nixon, Christian Nybo, Antoni Pamies Olive, Edgar Bonet Orozco, Steven W. Orr, Martin Ouwehand, Daniel Packman, John Palmieri, Chris Paulson-Ellis, Eduardo Perez, Michael Piotrowski, Andrej Presern, Jeremy Prior, Dominique Quatravaux, Matthias Rabe, Garst R. Reese, Peter Remmers, Todd Richmond, Ken Rose, Pavel Rosendorf, Gael Roualland, Lucien Saviot, Johannes Schmidt-Fischer, Andreas Schuch, Larry Schwimmer, Joe Siegrist, Jarrod Smith, Tom Spindler, Robin Stephenson, Joerg Stippa, D. Stolte, A. A. Stoorvogel, Juergen Stuber, Markus Stumpf, Jeffrey Templon, Jay Thorne, Anthony Thyssen, Christoph Tietz, tkil@scrye.com, Juha Vainikka, Poonlap Veeratanabutr, Ivo Vollrath, Gord Vreugdenhil, Ronan Waide, Jan Wedekind, Bj|rn Wennberg, Mats Wichmann, Stephen Williams, Barry Warsaw, Steven Winikoff, Carl Witty, Stephen Wray, Endre Witzoe, Kazutaka Yokota, Yair Zadik, and Robert Zwickenpflug.


16-Aug-98 X Version 11