Well, large UID support is somewhat useful, and you need to consider
the tradeoff between portability between regular uid systems and
32bit uid systems and the functionality of the old software. It's
certainly possible to build backwards compatability glue for libc4
so that a overly large UID will be either mapped to `nobody' or
will cause an immediate abort().
>Sorry for ranting but the "we don't care about binary compatibility"
>attitude that Linux people have gives Linux a very bad name and it
>creates a significant amount of work for people running systems with
>lots of users.
If the old system calls stick around, it's easy (if anyone cares to
do the work, that is) to write userland shims to libc that support
both sets of entry points, and then we can shove most of the
compatability glue up to userland and save the catastrophic changes
for the necessary change in time_t format (which, given my family's
lifespan, will happen while I'm still alive, and force me to do
deathbed upgrades of whatever 486 and Pentia machines I've still got
running at that time.)
>Peter, who's just discovered the joys of compiling code which mixes
>libraries compiled with gcc 2.8 and gcc 2.7.
Stick with a.out, it's good for the soul.
david parsons \bi/ no problem mixing that code (evil chuckle)
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to email@example.com
Please read the FAQ at http://www.tux.org/lkml/