Re: [Jmap] I-D Action: draft-ietf-jmap-sieve-02.txt

[Ned Freed <[e-mail]>]

> On 11/19/20 10:34 AM, Ned Freed wrote:
> >
> >> On 11/19/20 9:54 AM, Ned Freed wrote:
> >> >> Don't we have the same single-script/multi-script client issue with
> >> >> managesieve?
> >> >
> >> > Yes. That's kind of the point.
> >> >
> >> >> Or perhaps I'm not fully understanding your concern.
> >> >
> >> >> A separate question that comes to mind, is do we want JMAP Sieve
> >> servers
> >> >> to be able to only support a single script?  I guess in this case,
> >> the
> >> >> server could simply have a single script with id "singleton" which
> >> can
> >> >> only be updated and (de)activated.  Any attempt to create another
> >> script
> >> >> or destroy the existing one would fail with a "singleton" error.
> >> >
> >> > Indeed. It seems like we're asking a lot of clients here. And if the
> >> > past is
> >> > any indication, we're likely to be disappointed by what we get.
> >
> >
> >> So, its sounds like you're making an argument that JMAP Sieve should
> >> always operate on a singleton.  Am I correct?
> >
> > No, not really. I'm saying that the managesieve model has issues that
> > warrant serious consideration before we adopt it in a different context.
> >
> > The single sieve model is the obvious alternative, but it may not be
> > the only
> > one. Since none of this is especially difficult server-side, I would
> > like to
> > hear from client devleopers what sort of model they would prefer. Or if
> > the multisieve model is what they really want, have them say that.
> >
> >> Obviously, this takes the Sieve "include" extension of the table.
> >
> > No, not really. It just means you'd have to use some other means of
> > resolving
> > the includes.


> A user would have no way of managing their own included scripts, but
> they could still include system-wide global scripts in their own single
> script if desired.

Or you could include the ability to have (and manage) named scripts strictly
for use with include, and have one fixed script that's always the active one.

At first blush I actually like this model a lot. Enough that I may have to
revise my opinion of include.

				Ned