> process #1 should not rely on conditions that it has not previously > ascertained to be true (eg by setting something up by itself, so it > exists and is safe to use/rely on it).
> sounds self-evident ? sadly many inits do not comply with that > postulation, a well known example that comes immediatly to mind > is the poorly designed/though out SysV init itself. it also relies on syslog for logging, though it cannot ensure someone listens to /dev/log and processes the log messages.
