Comment 49 for bug 283128

Well, it does not "trigger", the bug simply doesn't show up if there are no symbols on the third level.
I mean that [ letter otherletter ] == [ letter otherletter NoSymbol NoSymbol ] and the NoSymbol's being
added to the extra group are effectively wiped out by xkb.

The good news is there is a simple way to reproduce the bug (and maybe someone will find a simple
solution?):
1. Get a keyboard with multimedia keys (or any other keys with scancodes > 128)
2. Set up any layout with the third level (typo layout is good for making any layout being so)
3. Now press the multimedia key and you've got an extra group
4. Run setxkbmap and you don't have the extra group

Also, some xev output may be useful (the first two indicate that xkb sends new maps):
MappingNotify event, serial 43, synthetic NO, window 0x0,
    request MappingKeyboard, first_keycode 8, count 248

MappingNotify event, serial 43, synthetic NO, window 0x0,
    request MappingKeyboard, first_keycode 8, count 247

KeyPress event, serial 43, synthetic NO, window 0x3200001,
    root 0xbd, subw 0x0, time 3654219, (696,311), root:(698,424),
    state 0x2000, keycode 180 (keysym 0x1008ff18, XF86HomePage), same_screen YES,
    XLookupString gives 0 bytes:
    XmbLookupString gives 0 bytes:
    XFilterEvent returns: False

KeyRelease event, serial 45, synthetic NO, window 0x3200001,
    root 0xbd, subw 0x0, time 3654339, (696,311), root:(698,424),
    state 0x2000, keycode 180 (keysym 0x1008ff18, XF86HomePage), same_screen YES,
    XLookupString gives 0 bytes:
    XFilterEvent returns: False