[Tig] DMARC test
rob at colorist.org
Wed May 28 03:51:31 BST 2014
Thank you to those who responded to me personally that the test was received. However, that isn't a true test: the exigencies of DMARC are such that (in a nutshell) those subscribers to mailing lists who have accounts at major Email Service Providers like yahoo, aol, et alia (who have implemented stringent DMARC controls), who reply to a message sent via the mailing list to their personal address, and address that reply to the list address (using, for example. q, Reply-All*) are apt to, based on DMARC's various policy headers, have their reply-post bounce, incrementing the bounce-count at the mailing-list server, which will eventually cause their subscription to be disabled. The intricacies are involved, and are implemented in violation of the Internet Engineering Task Force's standards and precedents, in addition to bypassing the time-proven RFC process. Because bounce-count and the resulting non-post of the message reply aren't immediately obvious to the replying-poster, the r-p won't know her/his reply isn't going out to the group-at-large for some time (until the bounce threshold is exceeded).
The professional, extraordinarily adept and proficient developers of the industry-standard open-source Mailman suite of programs have been very quickly issuing complex workarounds that can be effective, but can require extensive study to understand. Current policy in the latest Mailman release (2.1.18-1), for the TIG, has the From: header rewritten with "<poster's email> Via Telecine Internet Group" as an experimental workaround. It should be noted, strongly, that this breaks email RFCs and causes unpredictable behavior.
Allowing the large Email Service Providers to write their own standards, not subject to IETF and RFC (proven) examination and measurement, is a dangerous precedent. It breaks mailing lists and forces workarounds that themselves break the RFCs.
More at ( and also do a search for "DMARC breaks mailing lists" ):
*depending on the configuration of the Mail User Agent (MUA), the Reply-To: header is sometimes not honored for Replies (RFCs state it takes precedence; but MUAs themselves are sometimes mis-coded).
Rob Lingelbach http://rob.colorist.org
rob at colorist.org
More information about the Tig