Mailinglist Archive: opensuse (3464 mails)

[Zhang Weiwu <zhangweiwu@xxxxxxxxxx>]

WE cannot avoid it: some day you need to send an archive to a Windows
user, or you are on Windows and need to send an archive to Linux user.
In most cases your Linux is using a different charset than your Windows
receipt, because usually Linux use UTF-8 and non of known version of
Windows by default use UTF-8 on file system. So, you need to choose an
archive format that can deliver the file without corrupting file name.
Unfortunately there is only one known way to do this on Linux and only
one known way to do this on Windows. Which will be explained later in
this email. FYI: the famous ZIP format cannot do it.

(after searched for all possibility for weeks and discussed on this
list...)

The only one known way to deliver archived files to Windows user with
proper charset conversion/handling is to use my patched version of
patched lcab (originally from http://coding.wooyayhoopla.be/lcab/) and
my patched version only work correctly if your current locale is using
UTF-8. (the patch already sent to maintainer of lcab). (My patch is not
well programmed because I am perl idiot. I just wish to solve problems)

The only one known way to deliver archived files from Windows to Linux
with proper charset conversion is to use winRAR, as detailed below.

P.S. someone might suggest to use convmv before packing the package,
this only work if you know the receipt's Windows computer's
character-set. Here I suppose you don't know receipt's Windows version
and want to deliver the file safely whatever the format is, which is the
case I often met.

Detail:

I have tested a lot different formats:
                                                        WATCH FOR 'YES' IN THIS FIELD
                                                        ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Format  Created by      Opened by       compressible?   Filename readable after unpackaged?
zip     7-zip_win32     FileRoller      yes             no
zip     FileRoller      7-zip_win32     yes             no
rar     WinRAR 3.6      FileRoller      yes             yes!
rar     FileRoller      WinRAR 3.6      yes             no
iso     mkisofs -J      WinRAR 3.6      no              partial
jar     FileRoller      7-zip           yes             no
tar     GnuTar          WinRAR 3.6      no              no
tar     7-zip           GnuTar          no              no
cab     IExpress        cabextract      yes             no
cab     lcab(patched)   Explorer        no              yes!

Note about partial filename readable:

        Although the created ISO image have Joliet extension and all
        filenames in UTF16LE, winRAR (incorrectly) produce junk text for
        the first two ideographs, while following ideographs are
        correct. This should be winRAR's bug.
        
        The same ISO image, once burnt on CDR, works fine (no junk
        filename) on Windows.

Note on software being used: 
      * WinRAR 3.6 is actually WinRAR 3.60-Beta
      * 7-zip_win32 is a Windows software, the open-source alternative
        to WinRAR, can unpackage zip/tar/rar/7z formats and can create
        zip/tar/7z archives
      * IExpress is a Microsoft utility pre-installed on most Windows
        computers. This utility can create CAB format archive
      * FileRoller is SuSE 10.2's default gnome archive manager
      * in all above archive formats, TAR and ISO are non-compressing
        format, e.g. they only archive (collect files into one file). To
        compress them you can use gzip.

So the conclusion is:

     1. To send an archive file to other windows user, the only known
        way that can ensure file names are NOT corrupted due to locale
        difference is to make the archive with my patched version of lcab,
        unfortunately no very easy graphical user interface for this
        purpose and not compressed. If compression is needed, compress
        with gzip and Windows user who have WinRAR installed can open it.
        It shouldn't be difficult at all to add DEFLATE compression using
        zlib to lcab to produce compressed CAB archives, provide someone
        have time and knowledge to do it.
     2. To send an archive file from Windows to SuSE Linux user, the
        only known way that can ensure filenames are not corrupted due
        to locale difference is to make RAR archive with WinRAR (only
        tested v3.60-beta).
     3. Non of known Linux archive formats can handle character-set
        conversion. Only 3 archive formats have known support for 
        unicode thus are able to handle character-set conversion, 
        unfortunately all are non-free formats. They are
           i. Joliet ISO image format which is a Microsoft format. 
              Unfortunately opens partially correct with WinRAR;
          ii. Microsoft CAB format. Unfortunately CAB archives made
              on Windows using IEXPRESS do not use unicode filenames.
         iii. RAR format, unfortunately RAR format made on Linux do
              not support character conversion.
     4. 7z format is not tested because most Windows & Linux users
        cannot directly create/open 7z format archive without install
        special software so it cannot be used as a general solution.

-- 
Zhang Weiwu
Real Softservice
http://www.realss.com
+86 592 2091112

-- 
To unsubscribe, e-mail: opensuse+unsubscribe@xxxxxxxxxxxx
For additional commands, e-mail: opensuse+help@xxxxxxxxxxxx