On Tue, Jul 29, 2003 at 11:30:16AM -0600, Jamin W. Collins wrote:
> On Tue, Jul 29, 2003 at 10:03:45AM -0700, Zack Brown wrote:
> > On Tue, Jul 29, 2003 at 09:26:38AM -0500, kfogel@collab.net wrote:
> >
> > > When the two lists split, to not copy addresses over would therefore
> > > have been the equivalent of *unsubscribing* them from a list they'd
> > > asked to be on, without checking first.
> >
> > Mass subscriptions are just not friendly. You can't make assumptions
> > that people want to join a new list, regardless of the list's subject
> > or purpose. It's like getting spam from people you like and trust.
>
> That's not always true. For instance, what if the list changed servers?
Make a new list at the new server, and then be a net-cop on the old
list. Whenever someone posts to the old list, point them to the new one.
If you don't have that option (the old server dies and is not brought back),
then you hit a corner case and have to make a decision. But that's not the
situation here.
>
> > > Neither option is clearly right, here.
> >
> > I think a better way is to just create a new list, subscribe no one,
> > and then just be a net-cop on the dev list: when someone posts a user
> > question, gently point them to the user list. Ta da!
>
> Then there exists the chance that the user's list would have few people
> there to help the user with his/her question (possibly none).
It would find its own size. And the people who had been paying the most
attention (and would be more likely to answer questions on the new list)
would have known about the list, and subscribed early.
With the size of the dev list, and folks giving pointers to the user list
whenever user questions came up, the user list would have gotten a lot of
subscriptions right away. The inconvenience of being in an empty user list
would have been very very short-lived, and experienced by only very very
few people.
>
> > > And obviously, we didn't know that unsubscription would be broken!
> > > I mean, c'mon... :-)
> >
> > True enough, but even if it had worked, you still dumped a lot of mail
> > into a lot of INBOXes that weren't expecting it. Add it all up, and it
> > was a *huge* inconvenience. Virtually everyone had to clean up their
> > mailboxes, track down the details, and modify their configurations.
>
> And if they had been reading the original list (they were subscribed to
> it after all) they would have known the change was coming and could have
> been prepared for it.
A lot of people just archive the list. But if you wanted to count on
them knowing what was up, you could count on them subscribing themselves
when the time came.
>
> In short its a situation without a _perfect_ solution. Either way it's
> approached, someone is going to complain.
You make it sound like all solutions are equally good/bad. It's not so. Mass
subscritpions have serious problems, as we're seeing. The net-cop approach
is more respectful and less inconvenient for most people. It's not like
these things haven't been tried in the past.
Bah. It's over now. We're just arguing over principle at this point. But at
least this debate will go "on the record" for other projects that consider
mass subscriptions (as we speak, Tom Lord is thinking about doing it for
arch when he migrates to Savannah in the near future).
Be well,
Zack
>
> --
> Jamin W. Collins
>
> This is the typical unix way of doing things: you string together lots
> of very specific tools to accomplish larger tasks. -- Vineet Kumar
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
> For additional commands, e-mail: users-help@subversion.tigris.org
>
--
Zack Brown
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org
Received on Tue Jul 29 19:57:08 2003