In a gnome-terminal window where font used is from koi8-r charset 'less' precedes every string of cyrilic characters with an uppper case 'A'. This effect is at its best if one tries to send an output from 'od -c' into less. :-) Explicit changes in LANG environment variable (I tried a few, including ru_RU) do not have any visible effect apart of things like LANG=C where various characters are treated as binary. OTOH using -K with various 8-bit encodings, including 'less -K latin1', removes any display surprises regardless of a value of LANG. The above is likely to have a disconcerting effect on "uninitiated". If there is no handy test input around a file attached to #16978 can be used for such purposes. Michal michal
less uses the variables LESSCHARSET and JLESSCHARSET to determine the charset. Please set one of those to koi8-r for the time being. It seems that the japanese patch breaks things, we will look for an solution.
It also breaks latin1 charactersets... it puts an "A" in front of sequences of non-ASCII characters.
Created attachment 8619 [details] Norwegian text
I inserted an fallback if no charset could be found. Seems to work now, but please test less-358-14.i386.rpm and reopen this bug if this breaks anything.
It seems to work nicely now.