For tomorrows call: Email handling in the datatracker
Robert Sparks <rjsparks <at> nostrum.com>
2015-03-09 21:52:20 GMT
As I hope you've seen on ietf <at> ietf.org, there is some pain around how
email is currently being sent through the aliases and from the tracker.
Here is some background for discussion tomorrow. Please read through
this before we get to the meeting.
Henrik and I have both been studying this, and have several steps we are
going to take right away. I've also been discussing how draft ballot
email is handled with Pete to prepare a proposal for the IESG to agree on.
I will send a separate message with more detail than you want (but you
need it) about where the tracker is sending mail right now.
In short, we have these factors primarily contributing to the pain:
1) In November, we started adding the working group list to the .notify
alias by default at the request of the IESG. This was driven by the IESG
wanting the group to see ballot positions, and make it likely that
responses to those positions would be in-thread.
2) Shortly after, we started migrating to <at> ietf.org aliases instead of
3) 10-11 days ago (depending on when you read this message), we started
forwarding the aliases <at> tools.ietf.org to use the aliases at <at> ietf.org
instead of expanding them on the tools server.
Now, there were several things related to those 3 changes that have
a) Adding the working group list to .notify had an unintended
consequence. What we didn't account for is that the contents of .notify
get expanded into the .all alias, so people sending mail to .all for
some draft end up sending to the list.
b) When tools.ietf.org was expanding aliases, it added Resent-To and
Resent-From headers. The expansion at ietf.org does not currently do
that, and we are working to fix it. Some people were filtering on these
headers. It's also likely (we haven't confirmed this yet) that Mailman
was using the headers to bypass putting mail to .all into moderation for
having an implicit destination.
c) The aliases at ietf.org are _NOT THE SAME_ as the aliases at
tools.ietf.org! The datatracker allows addresses to be marked active or
inactive. If an address that's in a draft is marked inactive, then the
ietf.org alias will not use it - it will find the most recently changed
active address for that person instead and use that. The tools.ietf.org
aliases are not aware of these bits, and used the address in the draft
always. Further, when addresses are learned from drafts by the submit
tool, with the currently deployed code, they are marked inactive (the
idea at the time was that we didn't want mail to start going to a
different address for a person each time they submitted a draft using a
different email address - several years ago we had a big problem with
people being surprised by such automatic changes. Instead, they could
activate each address they wanted the tracker to use.)
So, we are making the following changes right away:
Now - i) We are teaching the tracker about a "preferred" address to use
whenever context doesn't indicate a better choice (often context will
say to use the address attached to a chair role for instance). The
address the tracker is preferring now (specifically the last active
address that's been touched) will be marked "preferred".
Now - ii) We are marking the addresses that are in drafts that are
current as active. If someone with an account has an addresses active
state changed by this, we will send mail to the affected address and all
other active addresses for them letting them know it happened (in case
they really _wanted_ that address to be inactive). As new drafts go
through the submit tool, any learned addresses will be marked active
(but note that the "preferred" address will not be automatically changed).
After that release, most of the addressing that has been surprising
people in the last 10 days will be resolved. The ietf.org aliases will
much more closely match what had been being generated at tools.ietf.org.
We (primarily Glen and Matt) are also continuing to work on getting
Resent-To and Resent-From into mail forwarded through the aliases at
ietf.org. We should set the priority for this on tomorrows call (I
suggest it that it's priority should be high).
Now, early discussions with Pete and Henrik lead to this rough proposal
that will be sent to the IESG. This is a draft idea, and we won't
execute on it until the IESG says to (Pete, note this is slightly
different than what we discussed last). I expect the discussion with the
IESG to change this some.
Soon - i) The "Save and Send Email" page in the ballot editing workflow
currently offers the content of the .notify field with a checkbox
(selected by default). We plan to add another row, with the group's list
address, also with a checkbox selected by default. If the AD sending the
message doesn't want the list copied, they unselect the box next to the
list address. (We might, instead, have to turn this into a bit that
lives on the document that the sponsoring AD manages, making it their
choice rather than the balloting ADs choice).
Soon - ii) We will scrub all the documents and remove the list address
from the .notify field. We'll send notes to the document personnel of
each affected document so they can verify that the result is what they want.
Soon - iii) We'll add a .group alias for people to use to copy the WG
list for a draft without having to look it up. The contents of this
alias would not be included when sending to .all
Sorry for the long message, but there's a lot involved here. Hopefully
I've teased it apart into easier to follow chunks.
Talk to you all tomorrow,
Tools-discuss mailing list
Tools-discuss <at> ietf.org
Please reports datatracker.ietf.org bugs at http://tools.ietf.org/tools/ietfdb
Please reports tools.ietf.org bugs at http://tools.ietf.org/tools/issues or
send email to webmaster <at> tools.ietf.org