> On Fri, 28 Mar 1997, Bryn Paul Arnold Jones wrote:
> >
> >the upper limit on the numbers of uid's you can have is 65536, because uid
> >on linux is an unsigned short (0 to 2^16).
> >
>
> Does this limitation apply to NIS?
> Isn't NIS the answer to this scaling?
>
It's still a you can only have N uid's, where N is set by the size of the
variable you store the number in. I know NIS dosn't do anything with
this, NIS+ dosn't claim to either (and the Linux port is only half
started).
The way Sun fixed this was to change the type of uid to long as opposed to
int, mainly because short is 2 bytes, long is 4 bytes, and int is either
2, or 4 bytes (ie 16 bits, 32 bits, 16 or 32 bits) depending on what the
people descided when they implamented int (btw on Linux it's 4 bytes, the
same as long).
If you write a program on SunOS (aka Solaris 1, or SunOS 4.x.x) uid is an
unsigned short (2 bytes), on Solaris (aka Solaris 2, or SunOS 5.x.x), uid
is long (4 bytes), (and an old uid (o_uid_t) is defined)
IMHO there should be a flag day (for distribution makers) where uid_t is
changed to long (or unsigned long ;), that way we can all upgrade, and
stay in sync pre-built binary wise. We don't all have to upgrade on the
same day, or some of us ever. Just as long as we keep source compatable
;)
> --
> g-r-a-t-e-f-u-l-l-y---[ email:<fishbowl@conservatory.com> ]---l-i-v-i-n-g
> d-e-a-d-i-c-a-t-e-d---[ http://www.conservatory.com/~fishbowl ]-----l-i-g-h-t
>
Bryn
--
PGP key pass phrase forgotten, \ Overload -- core meltdown sequence
again :( and I don't care ;) | initiated.
/ This space is intentionally left
| blank, apart from this text ;-)
\____________________________________