If you have trouble with the character mapping of SAMBA

There is a tool known als validchr.com which I designed to archive following:

• Enable SAMBA to recognize the CASE-Mapping (UPPERCASE/LOWERCASE) of foreign character sets, such that you can access "DFÜ.TXT" as "dfü.txt" on a SAMBA drive.
• Write out a "valid chars =" line suitable for smb.conf
• Be easy to use in DOS mode

Where to find it

You can find it in the examples section of samba. Under SuSE(r)-Linux it can be found in /usr/doc/packages/samba/examples/validchars/.

Drawbacks:

• It does not run on fast processors. On a fast processor (above 300 MHz or so) validchr.com does nothing. In fact the borland startup code of the original valichr.com tool gets a division by zero.

  There is a fixed version (I hope)

• You cannot map characters such that, with "ls", the Unix(tm) side shows the correct "foreign" characters.

  Rumours tell me that there is character encoding support in the latest version of Samba (Samba 2), but I don't know if validchr.com is compatible with Samba 2.

• You have to run a binary on your computer which I cannot proof correctness myself. I do not have any current Windows(tm) C Compiler myself, so I cannot prepare any new version. Even if I had I cannot ensure that my own system is virus free ..

Micro HowTo

To get your "valid chars =" line into your smb.conf, please do following:

• Locate validchr.com. It's part of the SAMBA distribution. Try "locate validchr.com" on the unix shell.
• Open a DOS box. Create a scratch directory on you hard drive. Do not create this scratch directory on a network drive! Copy validchr.com to the scratch directory. Start validchr.com in the scratch directory. You should get many lines of output.
• If there is no output, see below!
• The last line of the output (it's very long) starts with "valid chars =". Take this line and copy it entirely into the [global] section of your smb.conf file.
• You may remove the scratch directory and the contents.
• Kill smbd (the Samba server process) and restart it.

But: There are rumours about that validchr.com fails for some other foreign character sets; I can only check it for german (western) codes. I am really pretty sure that it will fail for multi-byte-codes such as japanese or korean character sets.

If validchr.com gives no output

Then it looks like you have a processor which is too fast. The following might help you. I havn't checked it completely myself:

• Download validchr32.com. It was compiled by Maxim Sobolev (thank you for this).
• Create a scratch directory on your hard drive.
• Copy validchr32.com into this scratch directory.
• Start command.com in this scratch directory.
• Call validchr32.com > out.txt
  You find the "valid chars =" line as the last line in the file out.txt.
• Perhaps start windows in the MS-DOS-only-mode and retry this again.