On 4/7/2013 6:48 PM, Stephan Bosch wrote:
> Hi,
>
> I made a new version of the "duplicate" draft. It should address the
> comments by Ned, Kristin and Alexey. I added the example I gave to
> Alexey.
>
> This resulted in quite a bit of restructuring in the description of
> the new test command, so it is best to review at least that section
> entirely.
One other thing: we recently defined this great new Sieve use-case
called IMAPSieve. Is there any application imaginable for the duplicate
test in this context, i.e. do we want to allow it there? Either way, I
think we should state its applicability explicitly.
I think there are legitimate use-cases, but it's also incredibly easy to misuse
in this context.
The basic problem is that moving messages around in IMAP doesn't carry with it
any sort of inherent special semantic and people move messages around for all
sorts of reasons (or for no real reason at all). This is quite unlike message
delivery.
So, for example, you could define a Sieve in IMAP to prevent, say, the
inclusion of a second copy of the same message in a given folder. But bad
things are going to happen if, say, you forget the filter is there and move
some of the messages out and back in a couple of times.
We'd have to extend the semantics of the extension substantiantially to take
care of this, and even if we were to do that actually using it would be
very tricky indeed.
Since I see the main benefit of this extension as how simple it makes
duplicate checks in Sieve, this usage makes me very nervous, so nervous
that I could certainly accept a "NOT RECOMMENDED" label for this case.
Ned
_______________________________________________
sieve mailing list
sieve(_at_)ietf(_dot_)org
https://www.ietf.org/mailman/listinfo/sieve