How to make the login shell xterm use utf-8

unicodexterm

I use xterm (X-Win32 2012 Build 30 from StarNet Communications Corp) to login from a Windows 7 PC to a Red Enterprise Linux 6 (RHEL6).

My problem is that all multi-byte utf-8 characters comes out garbled in the xterm login shell. For example, here's how the string "Wilhelm Röntgen" is rendered in the two shell instances (the font used is a Unicode font and is the same font in both shell instances):

 Login shell: Wilhelm Röntgen
 Second shell: Wilhelm Röntgen

If I've understood things correctly the software from rom StarNet Communications Corp implements (or emulates) a X terminal (a thin client that runs a X server). I.e. both shell instances runs in a X terminal window on the PC, and communicates with RHEL6 using the X11 protocol. Below is how both shells appear on my desktop, catenating a file with unicode multibye characters to the terminal.

enter image description here

Here is the command I've configured X-Win32 to use to start the login shell:

xterm -u8 -ls

However, after I login, I can do xterm in the login shell, and that command that will fork a new xterm instance where locale setting works as expected (i.e. utf-8 characters are rendered correctly).

Here is the relevant settings as they appear in the login shell:

$ locale
LANG=en_US.UTF-8
LC_CTYPE=en_US.UTF-8
LC_NUMERIC="en_US.UTF-8"
LC_TIME="en_US.UTF-8"
LC_COLLATE=C
LC_MONETARY="en_US.UTF-8"
LC_MESSAGES="en_US.UTF-8"
LC_PAPER="en_US.UTF-8"
LC_NAME="en_US.UTF-8"
LC_ADDRESS="en_US.UTF-8"
LC_TELEPHONE="en_US.UTF-8"
LC_MEASUREMENT="en_US.UTF-8"
LC_IDENTIFICATION="en_US.UTF-8"
LC_ALL=

$ printenv XTERM_LOCALE
en_US.UTF-8

I also have the following two lines in .Xresources:

xterm*locale: true
xterm*utf8: 1

It looks like the login shell xterm doesn't recognise the locale I've set, but I don't understand why not. My xterm is clearly capable of this, since all non-login shells do this by default.

Best Answer

At the time the sshd process on the remote computer forks to run /usr/bin/xterm there are very few environment variable set. In fact the LANG variable is not set. Hence the xterm process does not know that it should display characters in UTF-8. It falls back to xterms defaults. Whatever that might be.

However, the subshell running inside the xterm runs all setup scripts and alike. Including setting the LANG environment variable.

One needs to understand the difference between the remote xterm process and the shell process running inside of xterm.

The solution is to run the remote xterm process like this:

/usr/bin/env LANG=en_US.UTF-8 /usr/bin/xterm

env(1) is a utility to run a program in a modified environment.

Setting LANG will make the remote xterm display UTF-8 characters properly.

Eskil... :-)

P.s: Reading the xterm manual page I also found an easier way to achieve this:

xterm -en en_US.UTF-8

P.P.s: I do not think setting resources in ~/.Xresources will take effect unless you merge them in with xrdb. The xterm process on the Linux computer will query the X server running on your windows computer. At the time where xterm starts it is very unlikely that your X-Win32 server has the xterm* resources set. But you might be able to set resources in X-Win32 if it supports that.

Related Question