You are not logged in.

#1 2010-01-20 10:20:53

lomov_vl
Member
From: Russia, Irkutsk
Registered: 2010-01-20
Posts: 34
Website

File naming on Russian and limitation on length of file names

Hi.

I post this message as a note for any users for whom English is not native language and in everyday practice they use native language on Linux systems.

Torrents and Linux

System:
OS: Archlinux x86_64
FS: Reiserfs
Locale: UTF-8
Client: rtorrent
torrent: http://torrents.ru/forum/viewtopic.php?t=1445641

Problem:
if one try to download this torrent with the help of rtorrent he see message

Hashing: Storage error: [Hash checker was unable to map chunk: Слишком длинное имя файла]

(this in on Russian, on English this is: 'File name too long')

Namely rtorrent couldn't handle this file

Белов А. В. - Микроконтроллеры AVR в радиолюбительской практике. Автоматика. Радиоэлектроника. Связь. Радио. Радиосвязь. Любительская радиосвязь.(2007)(336).djvu

(on Russian again).
The same message prints Transmission when one try to download this particular file.

Reason:
the name is exceeded the limitation on length of file name: 255 bytes. See, e.g.

$ echo 'Белов А. В. - Микроконтроллеры AVR в радиолюбительской практике. Автоматика. Радиоэлектроника. Связь. Радио. Радиосвязь. Любительская радиосвязь.(2007)(336).djvu' | wc -c
279

The problem doesn't depend on type of FS, it is in Linux kernel (Linux VFS). Reiserfs has limit in 4032 bytes on length of file name, Reiser4 has limit in 3976 bytes.

In this case NTFS is more advanced FS:
the limitation on length of file name is 255 UTF-16 code units (255*2 bytes).

Of course if one use 8-bit locale (e.g. cp1251, koi8-r and etc) then this problem for THIS file is not appeared:

echo 'Белов А. В. - Микроконтроллеры AVR в радиолюбительской практике. Автоматика. Радиоэлектроника. Связь. Радио. Радиосвязь. Любительская радиосвязь.(2007)(336).djvu' | iconv -f utf-8 -t koi8-r | wc -c
162

Nevertheless there is no guarantee that someone won't face with longer named files and in that situation the advantage of NTFS against any Linux FS is evident.

Info about FS limitation is taken from wikipedia.

I would glad to hear comments and suggestions about this particular problem (because I don't know how to download files of this torrent and store them on my Linux box).

P.S. As I understand FreeBSD has the same problem in this case.

P.P.S. I think that this problem could appear for example on Linux which serves as samba server (file server) when windows user want to place file with native alphabet letters in name on server.

Offline

Board footer

Powered by FluxBB