Re: [lemonade] draft-ietf-lemonade-imap-sieve-00

[Cyrus Daboo <cyrus at daboo.name>]

Hi Barry,

--On December 21, 2005 11:05:52 AM -0500 Barry Leiba <leiba at watson.ibm.com> 
wrote:

> > Running a Sieve script explicitly over a set of messages would be nice,
>  > and would allow for all sorts of useful behaviour, commonly handled by
>  > clients locally:
>
> That's a good idea, which might be good to add here: an IMAP RUNSIEVE
> command, with a corresponding cause=RUNSIEVE.  I like it.  Do others?
> Shall I add it?

Actually I have to say that despite my initial enthusiasm for the idea of 
incorporating sieve into IMAP, given the complexity we already have and all 
the arguments about simplifying things and not doing ACAP in IMAP etc, I am 
now leaning towards not using sieve.

Instead, what I would propose for clients being able to 'run scripts', is a 
combination of the following:

1) Enhance the SEARCH command with better comparators (globbing, regex etc)

2) Introduce a 'pipe' operation that will allow the result set of the last 
(or named) search command to be fed into a subsequent fetch/store. I think 
there was already a proposal for that.

3) Introduce a 'macro' facility that would allow clients to define commonly 
used commands via a keyword and store those on the server, and then execute 
the commands via the keyword. This would allow for 'saved searches' - 
frequently used search commands stored on the server that can be quickly 
executed without having to uplaod the search each time. Anyone who has 
written a client that attempts to do traditional mua filtering using IMAP 
search will appreciate the benefit of something like that. These 'macros' 
could execute more than one command at a time with some limitations. e.g. a 
UID SEARCH followed by a UID COPY, followed by a UID STORE, followed by a 
UID expunge would emulate an mua filter action that could move messages 
matching certain criteria to another mailbox.

These three things combined together can do the majority of what is needed, 
and won't require clients/servers to have sieve script handling capability 
- just a few (pretty basic) extensions to the IMAP protocol.

That then leaves the question of whether it makes sense to use sieve for 
notifications. I would argue not - if its just for notifications. Lets 
define another IMAP extension for setting notification information. Maybe 
it results in something that is 'script-like' and if so perhaps SIEVE 
syntax could be borrowed for that, or maybe we stick with parenthesised 
lists as being more IMAP-like.

--
Cyrus Daboo


_______________________________________________
lemonade mailing list
lemonade at ietf.org
https://www1.ietf.org/mailman/listinfo/lemonade