keymap has wrong encoding.

I get similar messages for my Estonian keyboard on Phoebe 2:

Jan 21 23:30:21 phoebe keytable: plus before udiaeresis ignored
Jan 21 23:30:21 phoebe keytable: plus before Udiaeresis ignored
Jan 21 23:30:21 phoebe keytable: plus before udiaeresis ignored
Jan 21 23:30:21 phoebe keytable: plus before Udiaeresis ignored
Jan 21 23:30:21 phoebe keytable: plus before otilde ignored
Jan 21 23:30:21 phoebe keytable: plus before Otilde ignored
Jan 21 23:30:21 phoebe keytable: plus before diaeresis ignored
Jan 21 23:30:21 phoebe keytable: plus before brokenbar ignored
Jan 21 23:30:21 phoebe keytable: plus before odiaeresis ignored
Jan 21 23:30:21 phoebe keytable: plus before Odiaeresis ignored
Jan 21 23:30:21 phoebe keytable: plus before odiaeresis ignored
Jan 21 23:30:21 phoebe keytable: plus before Odiaeresis ignored
Jan 21 23:30:21 phoebe keytable: plus before adiaeresis ignored
Jan 21 23:30:21 phoebe keytable: plus before Adiaeresis ignored
Jan 21 23:30:21 phoebe keytable: plus before cedilla ignored
Jan 21 23:30:21 phoebe keytable: plus before acute ignored
Jan 21 23:30:21 phoebe rc: Starting keytable:  succeeded
----------------------------------------------------------------------
# cat /etc/sysconfig/keyboard
KEYBOARDTYPE="pc"
KEYTABLE="et"

Same version of kbd - kbd-1.08-2

Look like these messages are created by the "dumpkeys | loadkeys -u" line
in the script "unicode_start".

fixed in latest rawhide releases

Bug still there in RHL 9 ( Shrike )

Created attachment 90757 [details]
Excerpt from /var/log/messages showing the "xxx ignored" messages

And here is my i18n file :

# cat /etc/sysconfig/i18n
LANG="en_US.UTF-8"
SUPPORTED="hr_HR.UTF-8:hr_HR:hr:en_GB.UTF-8:en_GB:en:en_US.UTF-
8:en_US:en:de_AT.UTF-8:de_AT:de:de_DE.UTF-8:de_DE:de:hu_HU.UTF-
8:hu_HU:hu:sr_YU.UTF-8@cyrillic:sr_YU:sr_YU.UTF-8:sr:sl_SI.UTF-8:sl_SI:sl"
SYSFONT="latarcyrheb-sun16"

Which rawhide release has this fixed ?
both rawhide and shrike have kbd-1.08-4.i386.rpm

*** Bug 82510 has been marked as a duplicate of this bug. ***

I use the dvorak layout and not just for X.  I too noticed the keymap, keytable
and oddly system font fail.  So I have regular en qwerty usage at the console,
but dvorak *is* working in X (KDE).  

A strange case observed in using rdesktop:  I have set my default keyboard
layout on the remote windows box to dvorak in the past but now with RH9, the key
layout seems to be translated to some other I'm not familiar with (not US or
dvorak), possibly dvorak for left or right handed. To fix this I have to change
the windows box to en in order for the local X dvorak to work in the remote
window.  In the past, dvorak did not effect the rdesktop.  This is why I needed
to set windows to use it also.

Changing the default back to en, all is well with no failures at boot time.

I use dvorak (HH kbd with dvorak keycaps) at home on my RHL 9-- works both in
console and X.

Q: are you using dvorak with dead keys? (i.e. are you using dvorak with a
language other than english?) How do you do that?

No I'm not using dvorak with dead keys.  I guess I should add that my RH9
install was an upgrade not a fresh install.

Me too! If I use Danish keytable, I also get errors after RH9 installation. But:
if I tweak /etc/sysconfig/i18n to look like:

LANG=C
SYSFONT=lat1-16
SYSFONTACM=8859-1

no errors appear. So it seems that keymaps giving this error are perfectly ok,
loadkeys just reports these things if using Unicode. Personally I find it
strange that the warnings are printed since in /etc/init.d/keytable where
loadkeys is called the line is:

loadkeys $KEYTABLE < /dev/ttyS0 > /dev/null 2>&1

Shouldn't that make these warnings disappear?

Anyway, this bug *does* exists with RH9 and is quite annoying yet totally
harmless. The warnings just means that Caps Lock won't have the normal effect
with the symbols mentioned.

I still get this error/warnings with the latest bits.

In an email conversation with the kbd maintainer he said this:

> > does the plus signs mean something
>
> It is not bad if you remove this plus. The idea is that
> if one uses CapsLock then all text IS IN ALL CAPS, but
> for example digits remain as they are. So CapsLock acts
> a bit like ShiftLock, but only for letters. Now what is a letter?
> Letters are A-Z and further those things that have a + in front.
> In Swedish you would regard Aring a letter.
>
> > plus before XXX ignored
> > where XXX is aring, Aring, odiaresis, Odiaresis, adiaresis and Adiaresis
>
> Here all works well, but I do not have redhat modifications.
> The code in ksyms.c says
>
> add_capslock(int code) {
>         if (KTYP(code) == KT_LATIN)
>                 return K(KT_LETTER, KVAL(code));
>         ...
>         fprintf(stderr, _("plus before %s ignored\n"), p);
> }
> 
> so the question is why KTYP(code) == KT_LATIN fails for you.
> Probably because you use unicode.
> 
> For ISO 8859-X character sets there is an easy correspondence
> between lower and upper case versions of a character, but
> Unicode does not have such an easy correspondence, and of
> course for many languages there is no such concept.
> So plus before unicode stuff will not work.
>
> Andries

My conclusion out of this is that the bug should be moved to kernel.

It could also be noted that CapsLock doesn't work as expected in UTF-8 locale
with regards to non-ascii characters. If i type å (aring) in the console with
CapsLock pressed the lowercase version is shown and not the uppercase version.

/noa

*** Bug 91273 has been marked as a duplicate of this bug. ***

*** Bug 91638 has been marked as a duplicate of this bug. ***

Same deal with Russian keytable (see also bug 79338).

*** Bug 82400 has been marked as a duplicate of this bug. ***

*** Bug 98824 has been marked as a duplicate of this bug. ***

*** Bug 100530 has been marked as a duplicate of this bug. ***

Created attachment 93294 [details]
my "/var/log/boot.log" file

I had changed the default keymap from "us" to Polish ("pl") and got the same 
problem. After I had rebooted the system - during the system startup - my 
screen got flooded with such messages:

plus before ecircumflex ignored
plus before Ecircumflex ignored
plus before oacute ignored
plus before Oacute ignored
(...and many, many more - all in attached /var/log/boot.log file)

# my "/etc/sysconfig/i18n" file:
LANG="en_US.UTF-8"
SUPPORTED="en_US.UTF-8:en_US:en:pl_PL.UTF-8:pl_PL:pl"
SYSFONT="latarcyrheb-sun16"

Luke

All national 8-bit keyboard layouts should be converted to utf-8 in the same way
like we did with Russian layout. See attachment for bug 79338 as an example of
utf-8 layout.

Warning spewer commented out from ksyms.c in release 9