Topic "UTF-8 directory listings broken"

Author Message
honzo

Guest


Hi,

I have several directories on a Debian 'testing' system [openssl 0.9.7g-2, ssh 3.8.1p1-8.sarge.4, server runs in en_US.UTF-8 locale] containing several directory names with Czech national characters in UTF-8 enconding. Every time when I try to list those directories, I get complaints about an error when decoding UTF-8 text and only a part of the directory contents is displayed.

I have not checked the directory for improper UTF-8 characters, but I think that even if there are some inconsistencies, WinSCP should be able to handle such cases without messing up the directory listing. Moreover, I can see these listings properly in FileZilla, putty (Unix shell 'ls' output), and when accessing these directories as Samba 3 shares. From what I have seen, I am not aware of any suspicious characters that would spoil the listing.

I have first encountered this bug with Servant Salamander beta 10 WinSCP plugin, just installed WinSCP 3.7.6 and it does the same thing. I have just downgraded to WinSCP 3.6.8 and it works like charm ...

-- jan
Advertisements
martin
[View user's profile]
Site Admin
Joined: 2002-12-10
Posts: 24530
Location: Prague, Czechia
Did you enabled UTF-8 support? WinSCP should not use UTF-8 with OpenSSH by default. OpenSSH does not support it.
_________________
Martin Prikryl
honzo

Guest


prikryl wrote:
Did you enabled UTF-8 support? WinSCP should not use UTF-8 with OpenSSH by default. OpenSSH does not support it.


Yep. No chance to get it working without forcing UTF-8 as OpenSSH supports only some archaic version of SFTP, AFAIK.

Meanwhile I tried WinSCP 3.8.2 build 330, it performs slightly better: It displays the Unicode characters without complaints, but it still goes ballistic in case it encounters something that is not UTF-8 (like, for example, filename containing '' in CP-1250, uploaded by some ignorant moron). I know that this is more a question of design, but I would still consider it more appropriate to just issue a warning and mask the illegal character with underscore or whaterver. This way users will be able to download the file, albeit with crippled filename, even if they forced UTF-8.

Actually, this is another important note: If I prevent WinSCP from using UTF-8, it will show the directory name containing (quite naturelly) gibberish instead of Unicode letters, but it downloads the file, containing non-ASCII and non-UTF-8 characters (my favourite '' in CP-1250) without complaints (again, quite naturally, as no character translation takes place).

-- jan
martin
[View user's profile]
Site Admin
Joined: 2002-12-10
Posts: 24530
Location: Prague, Czechia
honzo wrote:
I know that this is more a question of design, but I would still consider it more appropriate to just issue a warning and mask the illegal character with underscore or whaterver. This way users will be able to download the file, albeit with crippled filename, even if they forced UTF-8.

Unfortunatelly WinSCP does not work internaly in Unicode, so it cannot keep the original filename. Thus implementing your sugestion would imply major change. But it would be possible to simply skip the files with non-UTF8 names.
_________________
Martin Prikryl
Advertisements

You can post new topics in this forum






Search Site

What is WinSCP?

It is award-winning SFTP client, SCP client, FTPS client and FTP client integrated into one software program for file transfer to FTP server or secure SFTP server. [More]

And it's free!

Donate

About donations

$9   $19   $49   $99

About donations

Recommend

WinSCP Privacy Policy

WinSCP License