Hi,

On Wed, 12 Mar 2008, Szabolcs Szakacsits wrote:

> Sorry but I meant to create the files on Linux. What you have found is 
> the expected, correct behaviour in the WIN32 namespace and documented for 
> instance here:
> 
>   http://www.reddragonfly.org/ntfs/concepts/filename_namespace.html
> 
> I expect CHKDSK not to warn when the files are created on Linux, i.e. we 
> don't have the submitted problem. 

I tried and CHKDSK has no problem with these file names. 

Btw, what I have meant under "submitted problem" was NTFS compatibility 
(NTFS does no Unicode normalization). I'm aware Microsoft has also Unicode 
problems (hint: search Microsoft employees' blogs). If they can't read a 
valid POSIX filename what CHKDSK accepts then they should fix their 
software.

However the test revealed that we do have problem creating file names 
having code 128 and up. I hope the new converter takes care about this 
too ;-)

Regards,
            Szaka


> On Wed, 12 Mar 2008, [ISO-8859-1] Jean-Pierre André wrote:
> 
> > Hi
> > 
> > Attached is my report, I zipped it to make sure character encoding
> > is kept unchanged.
> > 
> > In short I could not create files with the following chars :
> > 0x01-0x1f
> > 0x22 (should be escaped ?)
> > 0x2a *
> > 0x3a :
> > 0x3c <
> > 0x3e >
> > 0x3f ?
> > 0x5c \
> > 0x7c
> > File creation was done through a run time library which
> > may have done its own checking. I will retry with direct
> > use of Win32 API unless somebody else does it.
> > 
> > Then chkdsk found no error.
> > 
> > Regards
> > 
> > Jean-Pierre
> > 
> > 
> > 
> > 
> > Szabolcs Szakacsits wrote:
> > > Hi Barry,
> > > 
> > > Thank you for the email and patch.
> > > 
> > > Could you, or somebody else, please do the following test to decide how we
> > > should progress?
> > > 
> > > Create 254 files using the unpatched driver which file names are single
> > > character ASCII(i) where i=1,...,46,48,...255 then run CHKDSK and tell us
> > > all the file names CHKDSK finds to be incorrect. The log of the CHKDSK
> > > session can be viewed and copy-pasted this way:
> > > 
> > >   http://en.wikipedia.org/wiki/Chkdsk#Viewing_results
> > > 
> > > We do have known problems with the conversion and these all should be 
> > > fixed
> > > at the same time. 
> > > Bernhard Kaindl has just implemented a new converter recently, in case
> > > you're interested in the current development addressing the conversion
> > > problems:
> > > 
> > >   http://thread.gmane.org/gmane.comp.file-systems.ntfs-3g.devel/447
> > > 
> > > Regards,
> > >       Szaka
> > > 
> > > 
> > > On Wed, 12 Mar 2008, Barry Kelly wrote:
> > > 
> > >   
> > > > The problem:
> > > > 
> > > > NTFS-3G writes filenames containing '?' differently to how NT itself
> > > > handles filenames containing '?'.
> > > > 
> > > > Details:
> > > > 
> > > > When using the POSIX subsystem, now known as Interix or Services for
> > > > Unix (SFU), files with '?' (0x3F) in the name are silently converted to
> > > > use the private area Unicode character 0xF03F.
> > > > 
> > > > When a file containing a '?' (0x3F) created by NTFS-3G are manipulated
> > > > from Windows, even using the POSIX subsystem, they cannot be "found".
> > > > 
> > > > Solution:
> > > > 
> > > > I suggest that, in the interest of compatibility, NTFS-3G follows the NT
> > > > POSIX convention and silently transforms 0x3F to and from 0xF03F.
> > > > 
> > > > Rationale:
> > > > 
> > > > The alternative is that files created by NTFS-3G are in no way
> > > > accessible or even deletable from NT, as it appears that the kernel
> > > > doesn't support '?' in file names. I hope it's obvious that that
> > > > alternative isn't very desirable.
> > > > 
> > > > Patch:
> > > > 
> > > > I've attached a patch to the latest CVS sources from sf.net which does
> > > > exactly that, using 'cvs diff -u3'. The simple idea behind the patch is
> > > > that during MB->UCS-2 and UCS-2->MB conversions, 0x3F is replaced with
> > > > 0xF03F and vice versa. The invariant that should apply is that data
> > > > referenced with type ntfschar* should never contain 0x3F, and contain
> > > > 0xF03F instead.
> > > > 
> > > > Any further suggestions on how to proceed appreciated.
> > > > 
> > > > -- Barry
> > > > 
> > > > -- 
> > > > http://barrkel.blogspot.com/
> > > > 
> > > >     
> > > 
> > > --
> > > NTFS-3G:  http://ntfs-3g.org
> > > 
> > > 
> > > -------------------------------------------------------------------------
> > > This SF.net email is sponsored by: Microsoft
> > > Defy all challenges. Microsoft(R) Visual Studio 2008.
> > > http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
> > > _______________________________________________
> > > ntfs-3g-devel mailing list
> > > [email protected]
> > > https://lists.sourceforge.net/lists/listinfo/ntfs-3g-devel
> > > 
> > > 
> > >   
> > 
> > -- 
> > JP André
> > email [EMAIL PROTECTED]
> > 
> > 
> 
> --
> NTFS-3G:  http://ntfs-3g.org
> 
> 

--
NTFS-3G:  http://ntfs-3g.org


-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2008.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
_______________________________________________
ntfs-3g-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/ntfs-3g-devel

Reply via email to