From newspost at deletethispart.hypercreations.com Mon Jun 2 08:29:31 2008 From: newspost at deletethispart.hypercreations.com (D. T.) Date: Mon Jun 2 08:30:03 2008 Subject: [Scmail] Messages sent using SC webmail contain invalid Message-IDs, Triggering Spamassassin "MSGID_MULTIPLE_AT" rule Message-ID: Not sure how long this has been going on, but when a message my wife sent to herself wound up in her Held folder, I analyzed the headers and found that the following Spamassassin rule had fired: MSGID_MULTIPLE_AT So, looking at the Message-ID generated by the SpamCop webmail system, I saw multiple @ symbols, which apparently isn't allowed by RFC 2822 sec 3.6.4. Here's a link to the SA Wiki article on this: http://wiki.apache.org/spamassassin/Rules/MSGID_MULTIPLE_AT and here's a sample Message-ID from a message I sent to myself: Message-ID: <20080602050631.9j8hdrb6hws84044- gbccvat@fcnzpbc.arg@webmail.spamcop.net> The problematic Message-ID is adding a score of 1.2 to messages we send, which is significant, and when combined with other possible attributes, might cause our messages to be dumped or trapped. This needs a "JT fix" ASAP, IMO, so I'll also email Support. DT From MikeE at ster.invalid Mon Jun 2 09:23:29 2008 From: MikeE at ster.invalid (Mike Easter) Date: Mon Jun 2 09:25:04 2008 Subject: [Scmail] Re: Messages sent using SC webmail References: Message-ID: Disclaimer: I don't use spamcop mail; I'm just commenting on the issues you've raised. D. T. wrote: > http://wiki.apache.org/spamassassin/Rules/MSGID_MULTIPLE_AT That link sez SA has an optional rule for scoring against multiple @ and gives a link where the MID is defined http://www.apps.ietf.org/rfc/rfc2822.html#sec-3.6.4 As I'm reading 2822's 3.6.4 on MID at the link, I don't see any stipulation against multiple @, just some suggestions about how to make a unique MID which involves the @. Conclusion1: I don't believe there is a 2822 rule against multiple @ in a MID, and everyone (who has control of it) should decide for themselves whether or not some particular or peculiar SA rule or another is any good for them or whether it should be discarded as a bad rule in their armamentarium. Conclusion2: Given conclusion1, I don't see any 'pressure' on JT to change how the SC mid is configured, but that's up to his interpretation of the 'existence' and implied 'guidance' of such a SA rule. Perhaps he might simply decide to not use that rule. PetPeeve: 'Bad' subjects. Important sentences or questions or statements belong in the message body. The message body is made of all complete sentences. A subject is a very very brief title for what is contained in the body, such as 'SC Mail MID problem' -- not a long sentence which belongs in the body, not the subject. -- Mike Easter kibitzer, not SC admin From MikeE at ster.invalid Mon Jun 2 10:07:59 2008 From: MikeE at ster.invalid (Mike Easter) Date: Mon Jun 2 10:10:03 2008 Subject: [Scmail] Re: Messages sent using SC webmail References: Message-ID: Mike Easter wrote: > D. T. wrote: > > > > http://www.apps.ietf.org/rfc/rfc2822.html#sec-3.6.4 > > As I'm reading 2822's 3.6.4 on MID at the link, I don't see any > stipulation against multiple @, just some suggestions about how to make > a unique MID which involves the @. RFC 2822 doesn't define the construct of the mid, but makes some rules about the various headers which may contain the mid. 2822 also 'notes' that, interestingly, the MID is 'optional', but if provided it has some MUST features (not the @ business) such as unique. Here's an RFC (of a particular type) which does define what the RFC 850 thinks the usenet mid should be: http://www.freesoft.org/CIE/RFC/850/index.htm Standard for Interchange of USENET Messages - This memo is distributed as an RFC only to make this information easily accessible to researchers in the ARPA community. It does not specify an Internet standard. - 2.1.7 Message-ID - In order to conform to RFC 822, the Message-ID must have the format "<" "unique" "@" "full domain name" ">" ... and then the 2.1.7 goes on to define how unique & fulldomainname strings don't contain @ and some other rules. Notice the reference to 822, which has been obsoleted by 2822. So the 1983 850 cites the 822 as must-defining the MID. 822 was 1982 and 822 did not, in fact, define the mid as 850 states, but only said that it should be unique. It seems there is some kind of 'assumption' going on about the mid being configured like 850 thinks, but 850 assumptions are both incorrect in the first place (about 822's definition of the mid) and also only informational and also superseded by 2822 which is silent on the number of @ for the MID which is optional in the first place and which only MUST be unique. -- Mike Easter kibitzer, not SC admin From newspost at deletethispart.hypercreations.com Mon Jun 2 10:45:31 2008 From: newspost at deletethispart.hypercreations.com (D. T.) Date: Mon Jun 2 10:50:03 2008 Subject: [Scmail] Re: Messages sent using SC webmail References: Message-ID: "Mike Easter" wrote in news:g20s87$918$1@news.spamcop.net: > Conclusion1: I don't believe there is a 2822 rule against multiple @ > in a MID ...but the SA folks seem to think so, and that's the point here. > Conclusion2: Given conclusion1, I don't see any 'pressure' on JT to > change how the SC mid is configured, but that's up to his > interpretation of the 'existence' and implied 'guidance' of such a SA > rule. Perhaps he might simply decide to not use that rule. ...which would be fine, assuming your interpretation of the RFC to be correct. If the SA interpretation is correct, then there's still a problem. > PetPeeve: 'Bad' subjects. Mea culpa....I lazily copied both the Topic Title and Topic Description I used in my SC Email forum post on this issue. In the forum, it helps to be as specific as possible in the Title in order to get people to click into the messages, but YMMV. I cross-posted this issue as a courtesy to people who read the ng but not the forum, so they'd be notified about the problem....which is very real and needs *some* sort of fix. DT From MikeE at ster.invalid Mon Jun 2 11:10:55 2008 From: MikeE at ster.invalid (Mike Easter) Date: Mon Jun 2 11:15:03 2008 Subject: [Scmail] Re: Messages sent using SC webmail References: Message-ID: D. T. wrote: > "Mike Easter" >> Conclusion1: I don't believe there is a 2822 rule against multiple @ >> in a MID > > ...but the SA folks seem to think so, and that's the point here. SA (and SpamPal, which I use) make all kinds of 'rules' and scoring based on someone's experience with spam and the ability to make a rule and a score out of it. That doesn't mean that all of the rules are based on some kind of RFC MUST condition or another. SA makes an 'assumption' about there being an RFC 2822 rule about 1 @ and gives a link to the RFC 2822 section on mid, but in fact the link and the 2822 do /not/ so stipulate about 1 @ and neither does the old RFC 850 nor the old RFC 822. SA has made a rule. The rule isn't supported by the RFC which is used as a basis for the rule. There isn't a 'requirement' that a SA rule/points be supported by any RFC. If there is a useful SA rule then it should/can be used. If there is a 'problematic' SA rule, then maybe it shouldn't be used, depending upon how problematic it is and how useful it is. SC has found itself in RFC-ignorant before. Someone might decide that senders who are in RFC-ignorant should have points scored against their mail. My arguments are that 2822 doesn't say only 1 @ and that the SA rule makes an incorrect assumption that it does. That doesn't mean there shouldn't be such a SA rule and it doesn't mean that SC's mail shouldn't contain 2 @s. >> Conclusion2: Given conclusion1, I don't see any 'pressure' on JT to >> change how the SC mid is configured, but that's up to his >> interpretation of the 'existence' and implied 'guidance' of such a SA >> rule. Perhaps he might simply decide to not use that rule. > > ...which would be fine, assuming your interpretation of the RFC to be > correct. If the SA interpretation is correct, then there's still a > problem. Not exactly a problem. A 'condition'. Like I say, a SA rule doesn't have to have an RFC to support it. > I cross-posted this issue as a courtesy to people who read the ng but > not the forum, so they'd be notified about the problem.... The forum is a much more likely place for JT to see your issue. > which is very > real and needs *some* sort of fix. You and I might not see that the same way. If there is a real condition that multiple @ is useful for scoring for spam and the score isn't so great that it causes any problems for goodmail (which apparently it did in the test example you cited at the beginning) then the SA rule caused more harm than good in the specific instance you cited. I'm a big believer that spamrules should not cause goodmail to get into the spam -- so my belief is that there should be 0.0000% false positives. That causes me to believe that the SA rule is a bad rule if it caused a false positive all by itself. -- Mike Easter kibitzer, not SC admin From MikeE at ster.invalid Mon Jun 2 18:00:48 2008 From: MikeE at ster.invalid (Mike Easter) Date: Mon Jun 2 18:05:04 2008 Subject: [Scmail] Re: Messages sent using SC webmail References: Message-ID: Mike Easter wrote: > D. T. wrote: >> "Mike Easter" > >>> Conclusion1: I don't believe there is a 2822 rule against multiple @ >>> in a MID >> >> ...but the SA folks seem to think so, and that's the point here. > Not exactly a problem. A 'condition'. Like I say, a SA rule doesn't > have to have an RFC to support it. > The forum is a much more likely place for JT to see your issue. > >> which is very >> real and needs *some* sort of fix. The current forum word/report is that the SC MID has reduced the @s to one. -- Mike Easter kibitzer, not SC admin From newspost at deletethispart.hypercreations.com Mon Jun 2 20:01:42 2008 From: newspost at deletethispart.hypercreations.com (D. T.) Date: Mon Jun 2 20:05:02 2008 Subject: [Scmail] Re: Messages sent using SC webmail References: Message-ID: "Mike Easter" wrote in news:g21qi5$2b8$1 @news.spamcop.net: > The current forum word/report is that the SC MID has reduced the @s to > one. Correct. Trevor (from CES/SC Email) considered it a bug and squashed it immediately. Happy ending. DT From nobody at devnull.spamcop.net Sat Jun 7 00:57:37 2008 From: nobody at devnull.spamcop.net (Wazoo) Date: Sat Jun 7 01:00:04 2008 Subject: [Scmail] SCMail newsgroup Archive subscriptions Message-ID: Do any of the folks attempting to subscribe to the Mailing List for this newsgroup's Archives actually read this newsgroup, either direct from the newsgroup server or the Archive server? Out of the 500+ attempts over the last year, only four or five have actually followed the instructions to make a 'real' attempt .... none of them have apparently read the old traffic in which I clearly stated that I wasn't ready to entertain Subscriptions. That I do not recognize any of the names / e-mail addresses involved (even those done correctly) doesn't leave me with a feeling that I would want to change my mind on this either. I simply do not grok that over 95% of the Subscription requests are for "this" newsgroup .. one of two that are essentially dead ...?????