Comment 14 for bug 394570

Revision history for this message
AZaharia (f0rg3r) wrote : Backspace via SSH only deletes last byte of characters

Binary package hint: openssh-client

This is not a terminal-related problem: regardless of the terminal used and the way the backspace key is supposed to behave (ASCII DEL, escape sequence, Control+H), what happens is that via SSH the backspace key only deletes the last byte of a character, not the whole character itself.

You can notice this when you use UTF-8 characters that are encoded on more than 1 byte. Some diacritics from the Latin alphabet are encoded on 2 bytes, whereas some Japanese characters are encoded on 3.

As an example, assume the following scenario: the letter "ș" is encoded as "c8 99" and endline is encoded as "0a":

user@host:~$ cat > ș.txt
ș
^d
user@host:~$ hexdump -C ș.txt
00000000 c8 99 0a |...|
00000003

Now, assume you write "testș[hit backspace]test[endline]". On the physical host (i.e. without being SSH-ed into it), the hexdump of that file will be:

user@host:~$ cat > test.txt
testș[backspace]test
^d
user@host:~$ hexdump -C test.txt
00000000 74 65 73 74 74 65 73 74 0a |testtest.|
00000009

whereas the *same* thing via SSH would lead you to having the following hexdump:

user@host:~$ ssh user@localhost
user@localhost's password:
user@host:~$ cat > test.txt
testș[backspace]test
^d
user@host:~$ hexdump -C test.txt
00000000 74 65 73 74 c8 74 65 73 74 0a |test.test.|
0000000a

So notice how backspace only deleted via SSH the last byte of "ș" i.e. only the "99" out of "c8 99"; compare the expected hexdump:

74 65 73 74 74 65 73 74 0a

with the actual hexdump:

74 65 73 74 c8 74 65 73 74 0a

locale is set to en_US.UTF-8; changing it to ro_RO.UTF-8 (both on the host and via SSH) yields the same results.

Finally, last details:

Ubuntu 9.04
Kernel 2.6.28-13-generic
openssh-client: Installed: 1:5.1p1-5ubuntu1
openssh-server: Installed: 1:5.1p1-5ubuntu1