meeting with SUGJ

[Andrew Tridgell]

I met with some members of the Samba Users Group of Japan last night
and they gave me a lot of useful information on what needs to be done
in Samba to properly support Japanese. We spent the evening running
some character set test suites that Monyo has written and working out
how we can merge the efforts of the Japanese Samba port with
samba.org.

The main things are:

- the head branch does mostly work with Japanese, with some
  exceptions. 

- iconv has some significant problems, including lack of a little
  endian UCS2 character set on older glibc versions (eg. RH6.2) and
  lack of some commmon character set encodings used in japan (eg. CAP
  and HEX).

- we need to make the name mangling global, not per smbd and it also
  needs to be persistent. I think this will be pretty easy with a
  global tdb to hold mappings (which will also allow us to more
  closely emulate the NT mangling algorithm). I think that we need to
  completely throw away mangle.c and start again, but I think the new
  mangle.c can be much simpler than the current one. One issue is the
  potential size of the mangling tdb.

- we need a "display charset" option, as some character sets commonly
  used on the unix side such as CAP are not suitable for reading by
  humans, so I plan on adding a disp_printf() function that will print
  to a file in the "display charset". That will replace the existing
  DEBUG() and printf() statements in smbclient, nmblookup etc where
  the output is supposed to be human readable. disp_printf() will also
  be used in SWAT.

- apparently docbook doesn't properly support Japanese. We didn't work
  out how to solve this, but it could be a pain for the translated
  docs.

Cheers, Tridge