SSH UTF-8 character mangling
Bug #10728 reported by
Jan
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
openssh (Ubuntu) |
Invalid
|
Medium
|
Unassigned |
Bug Description
When I do SSH from UTF-8 enabled Hoary box (LANG=cs_CZ.UTF-8) to ISO-8859-2
Warty machine (LANG=cs_CZ) Czech characters are replaced by blanks
(gnome-terminal) or mangled (VGA console).
Changed in openssh: | |
assignee: | kamion → nobody |
To post a comment you must log in.
Yes, this is kind of inevitable really ...
The first problem is that ssh doesn't pass locale information across to the
server, and even if it did sshd wouldn't do anything with it. This has been
fixed in OpenSSH 3.9p1; you'll be able to pass selected environment variables to
the server, and we'll probably configure it to pass LANG and LC_* by default.
You'll need a 3.9p1 server to take advantage of that, though.
The second problem, over and above this, is that text doesn't get recoded from
ISO-8859-2 to UTF-8 when it reaches your ssh client. This is much harder to fix,
and I don't think it's likely to be fixed any time soon, because many programs
such as scp expect ssh to be 8-bit-clean. You can use the 'luit' program to work
around this.
Much of this, of course, goes away when we have UTF-8 everywhere, and we're
trying to make a step towards that by defaulting to UTF-8 in Hoary. However, it
will take many, many years to get rid of legacy encodings entirely. :-(