[Samba] Re: AW: unicode conversion

[Jérôme Fenal]

Michel Jouvin wrote:
> Jérome,
> 
> No, you have skipped something from the discussion. The pb is that 
> before running convmv, the filename is incorrectly truncated because he 
> was encoded by Samba v2 (not UTF8) and Samba v3 looks for UTF8. And if 
> you convert the filename with convmv, the UTF8 version of the previous 
> encoding (supposed to be iso-8859-1) is not correct (doesn't correspond 
> to the expected character).
> 
> I made a new test : configure Samba v3 with unix charset = iso-8859-1. 
> In this case, the character displayed is still not the right one but if 
> you create a file with a french é and convert it with convmv, you obtain 
> the same encoding as with Samba v3 running with unix charset = utf8.
> 
> Looks like if the problem is that encoding used by Samba v2 was not 
> iso-8859-1. And convmv is working properly.

Was it CP850 (e.g. default client charset for Samba 2.2) ?
You can check it by running testparm -v if you still have your samba 2 
binaries around, on the old smb.conf.

If yes, remember that cp850 != iso-8859-1.
See 
http://oss.software.ibm.com/cgi-bin/icu/convexp?conv=ibm-850_P100-1995&s=ALL 
and http://oss.software.ibm.com/cgi-bin/icu/convexp?conv=ISO-8859-1&s=ALL.

Does convmv know how to remap cp850 to UTF-8 (never tested it ?)

Regards,

J.


-- 
Jérôme Fenal - Consultant Unix/SAN/Logiciel Libre
Groupe Expert & Managed Services - LogicaCMG France
http://www.logicacmg.com/fr/ - <mailto:jerome.fenal AT logicacmg.com>