Hi, Now I can confirm, that Samuel's patch works. Thank you Samuel, your patch is OK and thank you again for your help.
... but I do not understand another related issue: Computer boots, it has to load keymap from /etc/console/boottime.kmap.gz, but just now no accented characters are written using dead key. I make a copy of current keymap: dumpkeys > afterboot Now I run: loadkeys /etc/console/bottime.kmap.gz ... wow, now characters are written fine and now I dumpkeys > afterload diff -s afterboot afterload ... diff reports identical files :-( Is anything in init scripts resetting results of loadkeys running from /etc/rcS.d/S05keymap.sh? Pavel On Mon, 20 Apr 2009 20:40:46 +0200 Samuel Thibault <samuel.thiba...@ens-lyon.org> wrote: > clone 404503 -1 > retitle use unicode compose table ioctl > tags 404503 + patch upstream > thanks > > Pavel Vávra, le Mon 20 Apr 2009 20:22:38 +0200, a écrit : > > I've found that compose table in /etc/console/boottime.kmap.gz contains > > strange symbols 0x255 as result of each compose line, as described in > > http://bugzilla.kernel.org/show_bug.cgi?id=10143. I try to change my > > initial console map to my own file, but after rebooting dumpkeys still > > return 0x255 as composed characters. Does it mean that lenny kernel doesn't > > contain patch from kernel bug 10143? No, it seems this patch is implemented. > > Note that the bugzilla entry you mention is only about the predefined > compose table, not the one you load after booting. > > > I think that loadkeys has to set map and then dumpkeys has to return same > > (or similar) map back. > > Yes, it should be that way. > > > Any idea why some composed characters are rewritten? > > The problem is that kbd wasn't patched into using KDSKBDIACRUC. > http://bugzilla.kernel.org/attachment.cgi?id=12359 > > Samuel >
pgpohlFnkrcnB.pgp
Description: PGP signature
