[Bug 871] request for idn (intl domain names) support

[bugzilla-daemon at mindrot.org]

http://bugzilla.mindrot.org/show_bug.cgi?id=871

           Summary: request for idn (intl domain names) support
           Product: Portable OpenSSH
           Version: 3.8.1p1
          Platform: All
        OS/Version: Linux
            Status: NEW
          Severity: enhancement
          Priority: P2
         Component: Miscellaneous
        AssignedTo: openssh-bugs at mindrot.org
        ReportedBy: egmont at uhulinux.hu


This entry is meant to be a feature request for handling host names with
accents in them.

I see at least two ways of possible technical approaches: either use the libidn
library, or use the AI_IDN and AI_CANONIDN flags or getaddrinfo() which are
available at least in glibc cvs and some brand new distros e.g. Fedora 2.

I send a quick & dirty & should-not-be-applied-as-it-is patch which is kind of
a case study that "okay it's really not that hard to make it work". It basically
works, however, there are some bugs/notes which might require some attention.

- it sure needs some autoconf'isation to autodetect whether AI_IDN and
  AI_CANONIDN are available.

- setlocale() is required, since getaddrinfo() assumes the accented host names
  are encoded in the current locale's charset.

- Currently the host name is put in .ssh/known_hosts exactly as the same byte
  sequence as it was given to ssh. So if I use a latin1 terminal to connect to
  an accented host, and later I use an utf-8 terminal then it's not recognized
  to be the same host, actually, it is added once again to known_hosts with a
  different encoding. It's even possible theoretically that two different hosts
  from two terminals using different encodings conflict each other in this file.
  So I'd recomment to always use utf-8 when storing host names in files,
  regardless of the current locale.

- When I have latinX terminal, everything is fine now. However, in UTF-8
  terminal I see this:
  $ ssh óx
  The authenticity of host 'óx' (...)' can't be established.
                            ^^ this is okay
  RSA key fingerprint is ...
  Are you sure you want to continue connecting (yes/no)? yes
  Warning: Permanently added '\303\263x' (RSA) to the list of known hosts.
                              ^^^^^^^^ this is not okay
  ...
  ...
  Connection to óx closed.
                ^^ okay again
  So that \ooo escaping should be removed.

I hope you don't find it a silly request.



------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.