The lock is coterminous with the open filehandle / file descriptor unless released by LOCK_UN. A write by a process that ignores the lock doesn't release the lock, it just ignores it (which and breakage can occur). A write by the process that holds the lock also would not release the lock.
At least that's my recollection. The rules can be tricky though. And the fact that locks are advisory is both powerful and scary.
Dave
In reply to Re^3: Preventing multiple instances
by davido
in thread Preventing multiple instances
by Bod
For: | Use: | ||
& | & | ||
< | < | ||
> | > | ||
[ | [ | ||
] | ] |