Assar Westerlund <[EMAIL PROTECTED]> writes:
>Randell Jesup <[EMAIL PROTECTED]> writes:
>> Sounds like what we'd want to build it upon. If the FS doesn't
>> support it, use st_dev/st_ino.
>
>Actually, since it's in the kernel, the default implementation of the
>vnode operation might be:
>
>int
>vop_default_cmp (struct vnode *v1, struct vnode *v2)
>{
> return v1 == v2;
>}
Sure, depending on what's in a vnode (I haven't looked).
I was really thinking of the kernel; I don't know how much of the
interior filesystem structure is exposed to the kernel; I was assuming
that vnodes are something that's opaque (or mostly so) to the kernel,
and are interpreted by the filesystem that created them.
>Or did you mean a fallback in the library function for when the kernel
>doesn't provide the fdcmp (or whatever) system call? That could be
>something like:
That's what I was thinking of, partially. It makes binaries
more transportable, and source if we can get Linux/etc to add it to
their libraries.
>> The real problem is getting people to switch.
>
>You mean application programs? Sure, but the only thing we can do
>about that is implementing support for it, right?
Right; that's where to start.
--
Randell Jesup, Worldgate Communications, ex-Scala, ex-Amiga OS team ('88-94)
[EMAIL PROTECTED]
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-hackers" in the body of the message
