From: Brad Templeton (brad@templetons.com)
Date: Sun Jul 19 1998 - 02:45:48 CDT
On Sun, Jul 19, 1998 at 03:09:59AM -0400, John Moreno wrote:
> Brad Templeton <brad@templetons.com> wrote:
>
> > On Sat, Jul 18, 1998 at 02:28:56AM -0700, Russ Allbery wrote:
> >
> > > You seem to be assuming that implementation of cancel locks means that
> > > people will turn off all unlocked cancels. So far as I know, that's never
> > > been the proposal, and that's certainly not what I'm claiming will happen.
> >
> > Yes, that's exactly what people have been proposing. It's certainly the
> > eventual plan, but some people want to do it right away.
>
> I don't think that's the eventual plan at all - I think unsigned cancels
> should be accepted unless the article in question is locked. Sites
> which currently don't accept cancel I believe would move back to this if
> there were a way for people who were actually having trouble to keep it
> from happening, in the meantime they do this as a intermediate step.
>
I've seen every variant of the method proposed, but none truly match the
real goal of the system -- that the people who are supposed to be able
to cancel *all* articles can cancel them, and that none of the people who
aren't supposed to can.
You imply that by sending out an article with no lock, I'm either saying
"Nobody can ever cancel this, including me" or "Anybody can cancel this."
depending on the site.
I'm saying neither. I'm saying, "I have old software, I would really like
the system to just work as one might expect -- as described above."
This mishmash of results, and this entirely unsatisfactory result for users
of old posting programs hardly seem like the sort of thing we should
design. What we design should get as close as possible to that real goal.
If I send out a cancel, it should work everywhere, if we can design it
that way, and this should happen as soon as possible. If an unauthorized
person sends out a cancel, it should work nowhere, if we can design it
that way, and that should happen as soon as possible too. As quickly
as we can, the behaviour should not be different from site to site and
from user to user.