ietf-mta-filters
[Top] [All Lists]

SIEVE lunch summary

2003-11-12 20:11:10

Summary of SIEVE Lunch, IETF 58, Minneapolis 17-Nov-2003

Present in person: Cyrus, Rob, Chris, Phillip, Madan, Randy
Present on Jabber: Ned, Alexey, Ken, Arnt

current docs: <http://www.cyrusoft.com/sieve/documents.html>
jabber log: <http://www.xmpp.org/ietf-logs/sieve(_at_)ietf(_dot_)xmpp(_dot_)org/2003-11-12.txt>

I have summarised comments for each draft in their own section, even though the discussion of some of them was interleaved due to jabber lag and some of us taking a break to eat.

Discussion on existing drafts:

imapflags: some implementations exist. Should we last call? Not yet, its tied to variables very tightly and thus cannot proceed until variables is nailed down. Should name be changed because of recent incompatible changes? Final suggestion for new name: imap4flags.

vacation: new from address option proposed. Question is how to deal with MIME encoding. Suggestion: server MUST support utf-8 encoding of non-ascii from parameter, and MAY support encoding in some other appropriate character set. Client MAY do MIME encoding itself. Ned wanted alignment with Keith Moore's auto-responder draft: resent headers must be tested as well as regular to/cc/bcc.

variables: consensus that it is important to complete this asap. Cyrus has a new date draft (out soon - see below) that may overlap with date variables. Should date items be removed from variables: yes - lets pare variables down to the essentials. Date stuff can go in Cyrus' draft if it makes sense. Ned has partial draft on date tests - will send to Cyrus. Wait to see date draft before deciding exactly what to do.

regex: i18n problems - two solutions (1) ignore for now using case-insensitive ascii and octet-stream comparators, (2) wait for Chris Newman's comparator draft. Discussion on evil regex tests & utf-8 took place. Should we opt to publish as experimental and gloss over i18n issues? Rough consensus. Use reference to unicode technical spec in the draft. OK - lets rev the current draft and punt the i18n issue to the list.

managesieve: need to clean up - refs to latest sasl work, id nits, references etc. URL scheme and sasl profile name need to be changed to managesieve. STARTTLS issue with virtual hosts: can be done as extension. Ned will re-post his issues to the list. Rob will endeavour to get Tim to do a new draft asap, or will take over the job himself.

notify: draft is currently expired. Need to get it updated. Some implementations, e.g. sms messages. Rob will again prompt Tim for an update or take over himself.

body: Jutta will have an update out shortly. Will do list last call on that.

spamtest: in IESG last call right now.

include: how to deal with missing scripts? Error at runtime seems to be consensus. Need way to list global scripts in managesieve - Chris: allowing users to see the contents of global scripts is bad - could cause legal problems. Include will define managesieve extension for listing available global scripts only. It would be useful to have a managesieve extension to check a script against a supplied message to see errors or results - Randy volunteered to work on a draft.

multiscript: need more review of this document. Punt to list.

editheader: one 'hacked' implementation. Punt to list for further review.

copy: Consensus was to list last call it.

Discussion on new drafts:

Cyrus will submit mime header test draft in the next few days.

Cyrus will submit date test draft in the next few days.

Cyrus has report/trace draft in the works. Proposal: to allow users to turn on logging/tracing of actions on filtered messages. Two possible outcomes (1) addition of headers with trace info, (2) dsn with trace info sent to email address or fileinto'd a specific mailbox. Chris suggested just use (2). Hidden headers are bad. Ned has a similar capture extension - actually not the same. Any interest in also writing up capture - general indifference on that point.

Other issues:

fileinto + auto-create mailbox behaviour: consensus that the default behaviour is up to the implementation, but we need specific options to explicitly turn on or off auto-create as required by client. Not sure whether this was proposed in a draft before - possibly not. Phil will bring this to Jutta's attention.



--
Cyrus Daboo

<Prev in Thread] Current Thread [Next in Thread>