Rsync and filename codepages

[Dylan Cuthbert]

Hi there

These are the kind of error reports I get from my cron job running rsync:
(the filenames show up as perfectly ok japanese filenames in this mail and
when I browse the linux harddisk via samba)

stat ntprofile/shin/Favorites/???? : No such file or directory
mkstemp ntprofile/shin/Favorites/???? failed: No such file or directory
stat ntprofile/shin/Favorites/???? : No such file or directory
rsync error: some files could not be transferred (code 23) at
/tmp/rsync-2.5.6/main.c(1045)

These files are copied onto linux via a samba share (which is the domain
user profile login share as I use linux as my PDC).

Then I am rsync'ing these files from that share to another windows box which
I use to make DVD backups from.  (out of a good 6 gig of files I only get
those two errors)

Would it be useful to have rsync support optional codepage translation for
filenames because it is seemingly used to transfer between remote systems
that might have different character sets etc. (and I've seen several
unanswered posts in the archives asking about this)

Regards

---------------------------------
Q-Games, Dylan Cuthbert.
http://www.q-games.com

----- Original Message ----- 
From: "jw schultz" <jw at pegasys.ws>
Newsgroups: gmane.network.rsync.general
Sent: Tuesday, September 16, 2003 11:23 AM
Subject: Re: Rsync and filename codepages


> On Tue, Sep 16, 2003 at 11:07:21AM +0900, Dylan Cuthbert wrote:
> > Hi there,
> >
> > I am running rsync in server mode on linux (redhat 8) and I am running
> > cygwin (latest version) on a mixture of win2k and winxp machines for
> > Japanese users.
> >
> > When I Rsync from windows rsync complains about a no. of Japanese
filenames
> > (not all of them) and this is due, as far as I can work out, to
> > incompatibilities/clashes between SJIS filename encoding (which Windows
> > uses) and certain character sequences in linux filenames that are
invalid.
> >
> > So I'd like to switch linux to be using UTF-8, does Rsync use the
> > locale/LANG settings to convert filenames?
>
> No.  Rsync doesn't care about any of that.  To rsync,
> pathnames are a null terminated sequence of bytes with / as
> a directory delimiter.  The only special characters are /
> and NUL.  Rsync does not translate anything.
>
> I have to wonder what filesystems you are using because
> Linux treats filenames the same way as rsync: 8 bit NUL
> terminated array of bytes.
>
> > Maybe I need to patch it and merge in a version of the useful linux util
> > "convmv" so the user can specify a "from" and a "to" codepage.
> >
> > Any ideas, as this is causing serious problems on japanese systems.
>
> There have been earlier reports of codepage issues so you
> might check the list archives to see the past fixes apply to
> your problem.
>
>
>
> -- 
> ________________________________________________________________
> J.W. Schultz            Pegasystems Technologies
> email address: jw at pegasys.ws
>
> Remember Cernan and Schmitt
> -- 
> To unsubscribe or change options:
http://lists.samba.org/mailman/listinfo/rsync
> Before posting, read: http://www.catb.org/~esr/faqs/smart-questions.html
>