Re: Migrating to larger numbers

Andi Kleen (ak@muc.de)
Mon, 31 May 1999 23:06:54 +0200


>
> (Putting on my Samba Development Team hat now)... One possible
>reason for going to 64 bit uid_t and gid_t would be to aid mapping Windows
>global ids to Unix uids and gids. The Windows IDs include information about
>the domain they're a part of. Windows IDs are also used for groups,
>users (a group ID is just another user ID - which is not a bad idea),
>and machines. Mapping them is a major pain. That would be important
>for ntfs as well as samba.

This unfortunately has two problems:
- Currently glibc API is 32bit gid_t, uid_t. Going for more needs recompiles
even for glibc programs.
On the other hand we have to have some backwards compatibility plan for libc5
programs anyways, if it works well perhaps it could be extended to glibc, but
it isn't pretty.
- A lot of programs (and I think POSIX too) require these types to be of
arithmetic type. Unfortunately non gcc compilers (lcc, Tendra C, KAI C++ etc.)
often don't have long long, which requires a structure and breaks this requirement
(glibc has 64bit dev_t, that is where this one was discovered)

-Andi

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/