Subject: Re: new vfork(2) implementation
To: Jason Thorpe <firstname.lastname@example.org>
From: Bill Sommerfeld <email@example.com>
Date: 01/05/1998 15:52:02
> Well, there are several people who feel the same way, I'm sure... but
> one of the biggest points here is that if we're going to have an interface,
> it ought to at least have the intended semantics to that interface.
IMHO, while we're ranting about vfork.. the main problem of vfork() is
that its semantics make the userspace-kernelspace split of process
state visible to an application, since user-space state is copied and
kernel-space state isn't.
This severely constrains how one can partition an implementation of
(for instance) the POSIX API's between userland and kernel land. Ask
anyone involved with apollo's implementation of vfork() on domain/os
about what the letters "JLFU" stand for..