(Since the archive at http://www.imc.org/ietf-mta-filters/mail-archive/
seems so empty....)
Here is a copy of the minutes of the mail filtering ("anti-spam") BOF
at the 37th IETF meeting in San Jose, Thu 12 Dec 1996.
Discussions are also reported to be happening on spam-list(_at_)psc(_dot_)edu
(a majordomo list).
Cheers,
Neal McBurnett <nealmcb(_at_)bell-labs(_dot_)com> 503-331-5795 Portland/Denver
Bell Labs Innovations for Lucent Technologies
Formerly AT&T's systems and technology business
http://bcn.boulder.co.us/~neal/ (with PGP key)
-------------------
Date: Fri, 20 Dec 1996 17:53:36 -0500
To: "Jack De Winter" <jack(_at_)wildbear(_dot_)on(_dot_)ca>,
kirpal_khalsa(_at_)ccmail(_dot_)com,
snowhare(_at_)netimages(_dot_)com, steve(_at_)esys(_dot_)ca,
dwm(_at_)xpasc(_dot_)com,
yarong(_at_)microsoft(_dot_)com, rapier(_at_)psc(_dot_)edu,
earhart(_at_)cmu(_dot_)edu,
srw+filtering(_at_)cmu(_dot_)edu, chris(_at_)innosoft(_dot_)com,
randy(_at_)qualcomm(_dot_)com,
lyndon(_at_)esys(_dot_)ca, wall(_at_)cmu(_dot_)edu,
jgm(_at_)cmu(_dot_)edu, neald(_at_)glyph(_dot_)com,
ned(_at_)innosoft(_dot_)com, dan(_at_)innosoft(_dot_)com,
peter(_dot_)taylor(_at_)ncl(_dot_)ac(_dot_)uk,
jwnz(_at_)qualcomm(_dot_)com, nealmcb(_at_)bell-labs(_dot_)com
From: "Jack De Winter" <jack(_at_)wildbear(_dot_)on(_dot_)ca>
Subject: regarding the minutes for the filtering meeting...
have been swamped with work, and I apologize for the tardiness.
If I spelled any names wrong, perhaps some of you should have
printed them neater on the sign in sheet. ;-)
regards,
Jack
Chaired: Chris Rapier
Minutes: Jack De Winter
Attended:
Jack De Winter <jack(_at_)wildbear(_dot_)on(_dot_)ca>
Kirpal Khalsa <kirpal_khalsa(_at_)ccmail(_dot_)com>
Benjamin Franz <snowhare(_at_)netimages(_dot_)com>
Steve Hole <steve(_at_)esys(_dot_)ca>
David Morris <dwm(_at_)xpasc(_dot_)com>
Yaron Y. Goland <yarong(_at_)microsoft(_dot_)com>
Chris Rapier <rapier(_at_)psc(_dot_)edu>
Rob Earheart <earhart(_at_)cmu(_dot_)edu>
Sam Weiler <srw+filtering(_at_)cmu(_dot_)edu>
Chris Neumann <chris(_at_)innosoft(_dot_)com>
Randy Gellens <randy(_at_)qualcomm(_dot_)com>
Lyndon Nerenberg <lyndon(_at_)esys(_dot_)ca>
Matt Wall <wall(_at_)cmu(_dot_)edu>
John Meyers <jgm(_at_)cmu(_dot_)edu>
Neal A Dillman <neald(_at_)glyph(_dot_)com>
Ned Freed <ned(_at_)innosoft(_dot_)com>
Dan N. <dan(_at_)innosoft(_dot_)com>
Peter Taylor <peter(_dot_)taylor(_at_)ncl(_dot_)ac(_dot_)uk>
John Noremburg <jwnz(_at_)qualcomm(_dot_)com>
Neal McBurnett <nealmcb(_at_)bell-labs(_dot_)com>
[Editor's notes:
- this meeting started out as a discussion group about anti spamming
techniques
- these are my written recollections of the meeting
]
Chris Rapier
- need standardized filtering language
- [himself] is a sys admin with some programming
- [he] would like to see consensus on what to do and how to do it
Matt Wall
- evolved from Seattle IMAP conference filtering BOF
- recognized the need in infrastructure, standardized on client and server
- want to develop filters at the delivery/transport/MTA levels
- Seattle BOF did not want to consider client side filtering
- ACAP and LDAP are avenues of access, need filtering language
Ned Freed
- AOL is actively pursuing this, might want to talk to them
- complex relations can ensue depending on filtering goals
- systemwide rules may be illegal, need user input to make legal
David Morris
- should be distributed on behalf of the client
Kirpal Khelsa
- may be a delivery agent function at delivery time
Steve Hole
- operations like CC should be included for sorting
- distribute the work so that a delivery agent could be included
- shifting from proprietary to internet standards may prompt
the need for proprietary rules
- John Meyers presented a language in Seattle, SafeTcl and ActiveMail
also options
- examples exist today, must represent to client easily
Rob Earhart
- everyone has different goals and requirements
- filters could be URLs
- would put the focus on gateways and contain lots of options
Jack De Winter
- [we] need to decide the goals we want to pursue
Matt Hole
- need a good list of requirements for usage
Chris Rapier
- do we want to do this at some level? [consensus was yes]
- do we want to do this under the IETF? [consensus was yes]
- keep it open, simple, extensible
- move bulk of discussion to a IMC list on the subject
[Chris Rapier and Ned Freed volunteered and were accepted as
tentative chairs for the group]
Ned Freed
- domain of information to filter needs to be decided: all, header,
envelope
- facilities of the language
- actions that can be taken by that language as a result
John Meyers [explaining about filtering language]
- if/then statements
- looks at envelopes, headers, the size of the body
- some pattern matching using REGEX or GLOB
- actions are: file message, forward message, toss message,
reply with file to message, place message in mailbox
Randy Gellens
- when is the filtering applied?
Ned Freed
- depends on the type of information to be acted upon and the
desired scope
- [example was given about not being able to check headers if a
filter was installed at the MTA level, checking the SMTP replies]
Benjamin Franz
- need one filter for MTAs and one for receipt of the message
John Meyers
- [said it was] trivially extensible
Randy Gellens
- IMAP's annotate may be useful
John Meyers
- could have an action of annotation in the filter, possible extension
David Morris
- manager may want to use this for filtering clients
- list needs to be refined all of the time
- make sure to provide easy input to filtering rules
Sam Weiler
- may want information on something that passed the filter
Ned Freed
- stats require a turing complete language
Sam Weiler
- may want complex actions such as [PGP] signature verification
Chris Neumann
- want extensibility requirement, and get it right in the base and
then worry about extensions later
Randy Gellens
- may have opposition on server side about being turing complete
Ned Freed
- [concurred as] may want to run on systems with little extra resources
Steve Hole
- want language, but need to make it easy to display
Matt Wall
- base mechanism be very safe
Rob Earhart
- may/not want turing complete language
Chirs Rapier
- may have good reasons for requirements, but want a single, simple
language
Chirs Neumann
- need MIME types and labels for storage
Ned Freed
- keep the venue and audience in mind
- may want the user to have the ability to use another language
Chris Neumann
- needs forward, vacation, discard, file into
Lyndon Nerenberg
- reject is a compound application [reply with and discard]
- multiple instances of reply should be allowed
Chris Rapier
- need reject as a separate option, sends DSN type reject message
Yaron Goland
- easy to share rules
- able to use admin rules; but not to share with anyone on the planet
Chris Neumann
- some kind of include mechanism
Jack De Winter
- may want some manner of promoting from one store to another store
- case of user wanting to submit a rule for inclusion on a company wide
basis
Randy Gellens
- useful as it is; may get too complex and harm concept
- work on a simple version, then expand
John Neremberg
- explicit sharing may be too much
David Morris
- get all of the requirements, and then pare down
Chris Rappier
- would like some way to identify messages
Yaron Goland
- can make it harder to spam, but not impossible
- want to identify who a given message is from
Randy Gellens
- may want to use SUBMIT protocol to limit input to the system itself
Ned Freed
- [seem to be] only attacking spam through filters. other tools?
Lyndon Nerenberg
- what do we show to filters? envelope, headers, body?
Rob Earhart
- IMAP is flexible and can support this
David Morris
- need to analyze the problem more thoroughly
Chris Neumann
- more ways to keep it down at the SMTP level
Ned Freed
- examples that are out of scope: mail to news and vice-versa
- if gateways had a lag to allow for cancel messages to catch
up with them, would help
- should be exclusively about mail for now
Chirs Rapier
- IETF may be against a spamming BOF/WG
- focus on mail based filtering languages
Lyndon Neremberg
- if make so it applies to RFC822, can apply to others later
Matt Hole
- not main requirement
David Morris
- holding an article for some time is a filter
Yaron Goland
- access control is an issue
Ned Freed/John Meyers
- storage mechanism is out of scope
Chris Neumann
- two camps: anti-spam and mail
- both are good as it will generate good filtering
Jack De Winter
- need to consider that we may need multiple filters at multiple stages
Chris Neumann
- need to extract from the language the envelope rules
-------------------------------------------------
Jack De Winter - Wildbear Consulting, Inc.
(519) 576-3873 http://www.wildbear.on.ca/
Author of SLMail(95/NT) (http://www.seattlelab.com/) and other great products.