Re: UUIDs (and devfs and major/minor numbers)

Richard Gooch (rgooch@atnf.csiro.au)
Sat, 19 Jun 1999 10:08:08 +1000


H. Peter Anvin writes:
> Richard Gooch wrote:
> >
> > H. Peter Anvin writes:
> > > Richard Gooch wrote:
> > > > >
> > > > > That's your assertion. I claim it is starting with a solution going
> > > > > looking for the problem, and finding the solution is not what the
> > > > > problem required -> kluge.
> > > >
> > > > Credit me with having thought about the issues before I started
> > > > coding. I even discussed things on the list before I started
> > > > coding. Devfs is most definately a response to a set of real problems.
> > >
> > > You did, and I agree it is a response. I don't think it is the
> > > *correct* response (as I have explained many times on this list many
> > > times both before and after you first started on this.)
> >
> > You ignored the point I'm making. You said that devfs is a solution
> > looking for a problem. That is incorrect. There are (were) several
> > problems for which I came up with a solution.
> >
> > You are entitled to have your opinion on whether devfs is the right
> > solution or not. However, it's just plain rude to insinuate that I
> > coded before thinking.
>
> No, but you did go out and say "I'm going to write devfs". Then
> when encountered with the problems with the approach, you build
> hacks around the problems.

I decided to write devfs after I heard/read about other alternatives
and decided they didn't measure up.

And I deny that there are problems specific to the devfs approach.
Those "problems" (i.e. persistence) also exist with a disc-based
dynamic /dev, where you lose persistence information when a device
node is deleted after a driver or device is removed.

The current tar solution is indeed a hack. But that's not a problem
with devfs. And besides, a more elegant solution is in the pipeline.

> The problem with devfs is that if your configuration is vanilla
> enough, it seems quite compelling. It's the Microsoft "one size
> fits all" approach.

Strange. There are plenty of people with big systems or unusual
configurations that use devfs precisely because it just "happens" to
fit them.

And with devfsd, there is enormous scope for configurability. So it's
not a "one size fits all" solution. Have a closer look and you'll see
that.

Regards,

Richard....

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/