Sender Policy Framework

On Wednesday 2005-05-04 at 21:00 UTC, the council held a meeting on IRC. There was a pre-planned agenda. Chuck, Julian, Mark, and Meng were present from the beginning; Wayne joined at about 22:00, and Meng left at about 22:55.

Chairman's report. Chuck had nothing to report.

Executive Director's report. Meng reported that he had been attending various anti-spam conferences, and would be participating in the IAjapan anti-spam conference in Tokyo on 2005-05-10, giving the keynote speech. Also, he had not yet heard from Douglas Maughan, the HSARPA program manager, about why the grant request had been denied. Meng said that he and Shevek had been working on the Karma reputation system and on related technologies. More importantly, Meng reported that he had been consulting EarthLink, helping them rolling out SPF and DomainKeys. When Julian asked why the earthlink.com domain did not even have a "v=spf1 -all" record when earthlink.net did have a detailed record, Meng explained that this was due to SPF's forwarding problem. He added that, given today's hindsight in the early days of SPF, he would probably have omitted the concept of default rejection (-all) from the specification in order to silence the detractors who believed that the forwarding problem was a show-stopper. Julian and Mark disagreed with that assessment, pointing out that then SPF could not be used to actually prevent sender address forgery and that thus nothing would have been achieved. Meng concluded his report by saying that he considered getting the SPF specification finalized and published as an RFC as the project's highest priority – to which everyone agreed – and that this was "best achieved in cooperation with Microsoft, rather than inviting a conflict with them".

Cooperation with Microsoft. As the council chairman, Chuck asked Meng why he believed that cooperation with Microsoft was possible in the first place. Meng replied that he believed cooperation to be possible with anyone in the e-mail space, and that he saw Microsoft's involvement as a "net plus to the SPF movement" because, he said, many SPFv1 records had only been published after Microsoft had adopted the SPF record format for Sender-ID/Caller-ID. However, this notion was not well accepted by the other council members. Julian asked what Microsoft had deliberately done to help the SPF project, and Mark asked whether Meng had received any specific indication that Microsoft was willing to cooperate. Meng replied that, among other things, they had asked him which SPFv1 specification draft – draft-lentczner-spf or draft-schlitt-spf-classic – they should push for with the IETF, and that if they were in fact unwilling to cooperate, they would have submitted their own version of the SPF specification.

Meng also said that if the SPF community submitted a draft to the IETF that was in conflict with Microsoft's work, the IETF was likely to reject both and ask for compatible submissions instead. He explained that it was the IETF who wanted Microsoft and the SPF project to work together, and that this was why he believed it best for the project to come to a compromise with Microsoft in order to get the SPFv1 specification published as an RFC. When Julian asked why Microsoft had not delegated one or two of their employees to discuss the issue of cooperation with the SPF project, Meng replied that they had indeed delegated Craig Spiezle and Harry Katz in 2004 already, but that they had only been in contact with him and Mark Lentczner. Mark pointed out that, as the author of draft-schlitt-spf-classic, at least Wayne, too, should have been contacted. Meng offered to ask Craig and Harry to make contact with several council members, but he did not expect them to come to an agreement with the council as a whole.

Then Chuck tried to regain control of the discussion, reminding of various acts that Microsoft had comitted that could be seen as deliberately uncooperative, such as them trying to actively influence the semantics of the SPF specification line chosen by the project, or their Sender-ID Framework (SIDF) public relations campaign, which implied that SPF was just a part of Sender-ID. He concluded that making a deal with Microsoft was deluded.

At about 22:00, Wayne joined the meeting.

Next, a short debate arised about whether the "Checking other identities against SPF records is NOT RECOMMENDED" clause was harming the SPFv1 specification's chances to achieve RFC status. Meng advocated that position and described the issue as a "question of ideological purity vs. getting-it-published pragmatism", but the others rejected the notion that the design of the SPFv1 specification, whose primary goal it was to accurately describe the historical v=spf1 record format, should be defined by the wish for compatibility with Sender-ID. Finally Mark pointed out that the council had already passed a resolution during its last meeting mandating that checking of SPFv1 records against non-RFC-2821 identities be explicitly discouraged, so the debate was essentially concluded and only old arguments were exchanged after that.

Before the entire complex was concluded, Wayne wondered whether the IESG believed the draft-schlitt-spf-classic specification to be a replacement for the abandoned draft-ietf-marid-protocol. Meng seemed not to know the IESG's thinking about that either and thus proposed that Wayne contact and ask them about it. In the end, the motion was unanimously passed.

Council election period. Julian invited the council to discuss the length of the council election period. Wayne said that he had planned to resign from the council after six months, but that when he had initially stated this, he had assumed that he would be able to work more on the SPF specification than he had. The others collectively expressed their wish for Wayne to remain with the council. No other comments were made, so Julian moved to define the council election period as one year, and the motion was accepted by all five council members.

General operation of spf-announce. Meng apologized for his inactivity with regard to sending the first two press releases through the spf-announce mailing list and issued administrator privileges to Wayne immediately, who then extended those privileges to the remaining council members. Thus the issue was resolved.

Press Releases on spf-announce. Meng proposed to send a message to spf-announce before the press releases, summarizing what had happened in and around the project since the last announcement in February 2004. Wayne promised to write such a message, which he promptly did later that day. Julian promised to then send the first two press releases through the announcement list, too, which he did a few days later.

Project website. This was not a planned agenda item, but Meng said he had to leave the meeting soon, and as he considered it an important issue, he suggested setting up a staging website for the current http://spf.pobox.com website. Julian remarked that this was probably an acceptable solution for the interim, but that at least a full UML, VServer, or similar virtual server would be required for a long-term CMS-driven website relaunch. After it became clear that there were free or inexpensive hosting offers from at least Shevek, Chuck, and Wayne, the issue was adjourned due to the lateness of time.

At about 22:55, Meng left the meeting.

The SPF specification draft and the IETF. Wayne reported that he had released a new revision of draft-schlitt-spf-classic to spf-discuss the week before, and that since then he had received only few reviews. He was expecting reviews from the other council members and from the regular reviewers in the community, and he felt that the draft was not yet ready for the next release to the IETF due to important issues that were still unresolved, some of them still left over from the MARID IETF working group. Julian and Mark promised to send reviews within the next few days. Wayne explained that he was going to adopt right away any uncontroversial changes that would be brought up, and refer the rest to the council for final judgement. He planned to wait for another week and then submit the draft to the IETF. Julian asked whether Wayne intended to contact the IETF about the draft's role before he submitted it, which Chuck seconded. Wayne replied that he intended to initially just send a quick status request to the IESG, as was decided earlier during the meeting.

Because Meng had left the meeting, Wayne objected to deciding on any material issues of the specification draft, such as whether the PermError result code "SHOULD" be treated identically to the None code. The others agreed to that and that another meeting should be held within a week.

The meeting was then adjourned at 23:25 UTC.