Joerg Schilling wrote at 11/30/06 05:42:
I'm still hoping that we could work on transparent code conversions at
libc so that people will use Unicode at file systems and yet such Unicode
will be transparently converted to user land whatever codeset that
people are using at their environment as the user locale. I think I forwarded
the URL before but just in case, the idea is described in the following
PDF file, esp. slide pages 15-25:
http://developers.sun.com/techtopics/global/products_platforms/solaris/reference/presentations/IUC29-FileSystems.pdf
This would at least create incompatibilities with long filenames.
ISO-8859-1 is the low 8 bits of UNICODE and if I use a ISO-8859-1
coded filename I am currently able to have up to 255 ISO-8859-1
characters in a filename.
After e.g. UFS is converted to UTF-8, the max file name length depends
on the content of a filename and may be reduced to only 127 ISO-8859-1
characters. As a result, you may be unable to restore a backup.
Jörg
That's true and so migration to UTF-8 FS and switch on or off of
the feature is not mandatory. The migration is also aided by tools such as
fsexam(1) that people can do dry run and customization after the dry run.
Ienup
_______________________________________________
opensolaris-discuss mailing list
opensolaris-discuss@opensolaris.org
