Thoughts on Downgrading
G”day: The following comes from the (currently) draft notes from the ICANN60 gathering. There was general acceptance, acknowledging that you’ll never get perfect automated downgrading, that the following would be useful good practice to include in the UASG’s Quick Guide to EAI (as well as our forthcoming Introduction to EAI). Your thoughts, please, but the 10th of November. Thanks. Don The UASG will (after further consultation) include the following as a Good Practice (not best practice) guide for Downgrading: - The EAI Mailbox provider establishes an EAI address and an ASCII alias when the mailbox is set up. - The ASCII alias will have a display name equal to the EAI Address. - When a non-EAI compliant mail system is encountered, the EAI MTA will revert with the ASCII alias and Display Name. - The ASCII Alias should not default to a Punycode conversion of the mailbox name because bi-directional equivalence cannot be guaranteed. Don Hollander Universal Acceptance Steering Group Skype: don_hollander
On Fri, 3 Nov 2017, Don Hollander wrote:
The UASG will (after further consultation) include the following as a Good Practice (not best practice) guide for Downgrading:
- The EAI Mailbox provider establishes an EAI address and an ASCII alias when the mailbox is set up.
- The ASCII alias will have a display name equal to the EAI Address.
- When a non-EAI compliant mail system is encountered, the EAI MTA will revert with the ASCII alias and Display Name.
- The ASCII Alias should not default to a Punycode conversion of the mailbox name because bi-directional equivalence cannot be guaranteed.
Also note that only the initial sender can do this. If mail goes A->B->C, A can downgrade if the A->B hop can't do EAI, but B cannot if B->C can't. Regards, John Levine, john.levine@standcore.com Standcore LLC
Clarifying, since this question comes up from time to time: Also note that only the initial sender can do this. If mail goes A->B->C, A can downgrade if the A->B hop can't do EAI, but B cannot if B->C can't Is this a MUST or a SHOULD (recognizing that RFC-style SHOULD is still very firm)? /marksv -----Original Message----- From: ua-eai-bounces@icann.org [mailto:ua-eai-bounces@icann.org] On Behalf Of John Levine Sent: Saturday, November 4, 2017 7:21 PM To: Don Hollander <don.hollander@icann.org> Cc: ua-eai@icann.org Subject: Re: [UA-EAI] Thoughts on Downgrading On Fri, 3 Nov 2017, Don Hollander wrote:
The UASG will (after further consultation) include the following as a Good Practice (not best practice) guide for Downgrading:
- The EAI Mailbox provider establishes an EAI address and an ASCII alias when the mailbox is set up.
- The ASCII alias will have a display name equal to the EAI Address.
- When a non-EAI compliant mail system is encountered, the EAI MTA will revert with the ASCII alias and Display Name.
- The ASCII Alias should not default to a Punycode conversion of the mailbox name because bi-directional equivalence cannot be guaranteed.
Also note that only the initial sender can do this. If mail goes A->B->C, A can downgrade if the A->B hop can't do EAI, but B cannot if B->C can't. Regards, John Levine, john.levine@standcore.com<mailto:john.levine@standcore.com> Standcore LLC _______________________________________________ UA-EAI mailing list UA-EAI@icann.org<mailto:UA-EAI@icann.org> https://na01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fmm.icann.or...
On Mon, 6 Nov 2017, Mark Svancarek wrote:
Clarifying, since this question comes up from time to time:
Also note that only the initial sender can do this. If mail goes A->B->C, A can downgrade if the A->B hop can't do EAI, but B cannot if B->C can't
Is this a MUST or a SHOULD (recognizing that RFC-style SHOULD is still very firm)?
MUST, absolutely, not negotiable. B does not know what addresses A considers to be equivalent. To clarify, we're talking about forwarders, not mailing lists. Lists conceptually originate a new message (even if it looks a lot like the incoming message) and do all sorts of funky stuff to messages on the way through. Regards, John Levine, john.levine@standcore.com Standcore LLC
Right, and this produces one of the irrevocably broken downgrading scenarios: Sender 1 has an EAI account ("U1") and wants to mail a message to both Receiver 2, who is EAI-enabled (account "U2") and also Receiver 3, who is not (account "A3"). Sender has a downgrading email provider who has also assigned alias A1 to the same mailbox as U1. Receiver 2 sees: From:U1 Cc:A3 Receiver 3 sees: From:A1 Cc: <implementation-specific*> If Receiver 2 ReplyAll, Sender 1 sees: From:U2 Cc:A3 Receiver 3 doesn't see anything and Receiver 2 gets a nondelivery notice. Even if Receiver 2 also has a downgrading email service, it's not allowed for the "U1" address to be converted to something that Receiver 3 can accept! * NOTE: Depending on the implementation-specific details (did Sender 1 convert "U2" into ACE? And if so, what did Receiver 3 do with it?), there may be additional breaks. Just keep remembering that downgrading is always imperfect and that the primary scenarios are people emailing to others who use their same script, and it's not so bad. Ajay, do you want to weigh in on this? /marksv -----Original Message----- From: John Levine [mailto:john.levine@standcore.com] Sent: Monday, November 6, 2017 4:54 PM To: Mark Svancarek <marksv@microsoft.com> Cc: ua-eai@icann.org Subject: RE: [UA-EAI] Thoughts on Downgrading On Mon, 6 Nov 2017, Mark Svancarek wrote:
Clarifying, since this question comes up from time to time:
Also note that only the initial sender can do this. If mail goes
A->B->C, A can downgrade if the A->B hop can't do EAI, but B cannot if
B->C can't
Is this a MUST or a SHOULD (recognizing that RFC-style SHOULD is still very firm)?
MUST, absolutely, not negotiable. B does not know what addresses A considers to be equivalent. To clarify, we're talking about forwarders, not mailing lists. Lists conceptually originate a new message (even if it looks a lot like the incoming message) and do all sorts of funky stuff to messages on the way through. Regards, John Levine, john.levine@standcore.com<mailto:john.levine@standcore.com> Standcore LLC
On Tue, 7 Nov 2017, Mark Svancarek wrote:
Right, and this produces one of the irrevocably broken downgrading scenarios:
No question about that, but none of the experiments we tried for downgrading on the fly worked. They just got confused and ended up sending EAI mail to systems that couldn't handle it. Regards, John Levine, john.levine@standcore.com Standcore LLC
participants (3)
-
Don Hollander -
John Levine -
Mark Svancarek