Unicode filenames in an ISO9660 file-system

Christophe Huriaux c.huriaux at gmail.com
Wed Jul 13 18:37:42 UTC 2011


  I'm about to implement the Joliet extension in the ISO9660 filesystem as
part of my GSoC project. This extension removes the limitations on the
character set used (the ISO9660 standard implies a character set comprising
only uppercase letters, digits and underscores in a filename.extension
format), and brings the internationalization thanks to Unicode filenames.

  In order to maintain backward compatibility for systems which doesn't
handle the Joliet extension, every entry in the directory structure of the
ISO9660 volume is recorded twice :
     - One which is fully ISO9660 compliant (for older systems)
     - Another one with Unicode and long filenames support

  Since the RTEMS iso9660 implementation is going to be aware of those
Unicode encoded records, I'll have to make a decision on how to handle them,
should I :
     - use Joliet records and leave the filenames unchanged to the upper
layers (and thus to the final application) : the end application should be
able to interpret the character set (which is not the case, RTEMS is not
designed to handle Unicode strings AFAIK, does it ?)
     - use Joliet records and somehow convert the filenames in an other
character set in order to avoid compatibilty breaks
     - don't use Joliet at all : "classic" filenames (lowercase, special
characters, ...) would not be handled in this case and it limits data

  I need comments from RTEMS users since I don't know all the internal of
the OS, maybe I'm wrong regarding the Unicode support.

Thanks for your help,
Best regards,
Christophe Huriaux
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.rtems.org/pipermail/users/attachments/20110713/d08fe8de/attachment.html>

More information about the users mailing list