Closed Bug 1471865 Opened 7 years ago Closed 7 years ago

autoconfig documentation

Categories

(Thunderbird :: Account Manager, defect)

Product:
Thunderbird
Component:
Account Manager
Type:
defect
Priority:
Not set
Severity:
normal

Tracking

(Not tracked)

Status:
VERIFIED INVALID

People

(Reporter: dpa-mozilla, Unassigned)

Details

User Agent: Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/605.1.15 (KHTML, like Gecko) Version/11.0 Safari/605.1.15 Epiphany/605.1.15 Steps to reproduce: https://wiki.mozilla.org/Thunderbird:Autoconfiguration says TB calls https://live.mozillamessaging.com/autoconfig/emailaddressdomain during auto-configuration. https://developer.mozilla.org/en-US/docs/Mozilla/Thunderbird/Autoconfiguration says TB fetches https://autoconfig.thunderbird.net/v1.1/domain/ under the same conditions. Update both pages to be consistent and both of them should state the same called URL.
We do not use bugzilla to track documentation issues. The relevant docs can just be directly updated. Note however, history shows these docs had an update earlier today that were cancelled. For example https://developer.mozilla.org/en-US/docs/Mozilla/Thunderbird/Autoconfiguration$history
Status: UNCONFIRMED → RESOLVED
Closed: 7 years ago
Resolution: --- → INVALID
I cannot directly change the documentation, because I don't know what the text should be at the end. The current state is inconsistent/controversal.
To his point, Wayne, if you have a problem with the documentation is there a way to comment? Where should that be raised? Do we need to create a place to do this?
Flags: needinfo?(vseerror)
One creates an MDN account if necessary, and just directly edits the document. In this case Ben has already done it. https://developer.mozilla.org/en-US/docs/Mozilla/Thunderbird/Autoconfiguration$history
So the way it works is one edits whatever s/he considers as appropriate and then somebody else possibly reverts the changes, if they are not good? Concerning the autoconfiguration site: it is stated, that first the author has to be contacted before doing changes. It is not clear from this prescription, if the author does not object within which time, any changes can be made without consulting him. Keep in mind, that if the time for waiting for an answer is too long, it can be a reason to loose the motivation and at the end not do any changes. Also trying to consult the author, getting the email rejected to him as spam, then doing changes on the wiki, after being unable to contact the author, and seeing the changes reverted, is not the way things shall run. How can one estimate the chances, that changes are not going to be reverted? Think of * adding a section, that other mUA consider _imap._tcp, _imaps._tcp, _submission._tcp, _submissions._tcp, _pop3._tcp and _pop3s._tcp.domain.TLD records during their autoconfiguration, so providing universal autoconfiguration, that anyway does not work with Outlook, means adding the mentioned records * adding a section, that only those domains will be inserted in the ISPDB, if they have the aforementioned *._tcp.domain.TLD records, as this is in the interest of their users and forgetting to add the records in not what ISPs want either. * clarifying the behaviour when non-existent email addresses is supplied as ?emailaddress= . Shall the MUA insist that invalid address is provided, or can the xml file return invalid xml? * adding a section stating which client implements to which extend this autoconfiguration. Currently nobody knows which clients support what and the procedure to extend the wiki is ... heavy. In fact there are articles some years old, stating Evolution and Kmail support this autoconfig.domain/mail/config-v1.1.xml?emailaddress system, but Evolution added only very recently support * In particular state that Evolution after 3.30 follows this autoconfiguration system, but passes to ?emailaddress=EVOLUTIONUSER@domain, instead of the supplied local part of the email address. A philisophical question: is this autoconfiguration site also a specification that other clients can follow, or it a specification for site operators, but not for MUAs? Provided that Thunderbird calls https://autoconfig.domain/.well-known/autoconfig/mail/config-v1.1.xml?emailaddress=xyz@domain shall other MUAs call ?emailaddress=xyz@domain too, or not? Not really wiki-related, but consider contacting the sites in the HSTS Preload List only over HTTPS. Finally, the comment field when doing changes in the wiki is two short. For me it was impossible to write at the same time why I do some changes and that I couldn't contact the author per email and I was told, that I was supposed to contact the author before doing changes.
The author of the code remains very much an interested party, and it is unusual for doc pages to state "contact the author before making changes". So most of your questions need to be directed to him.
Flags: needinfo?(vseerror)

I asked on 29.06.2018 the author of the code per email to include in the Thunderbird-Autoconfig site, that other MUAs can use DNS SRV records, as described in RFC6186 und RFC8314. The purpose of this amendment is to inform the reader how to create a server-side setup, that works with all varieties of clients.

On 28.06.2019 I asked the author per email to remove the duplications at https://wiki.mozilla.org/Thunderbird:Autoconfiguration and https://developer.mozilla.org/en-US/docs/Mozilla/Thunderbird/Autoconfiguration .

On the same date I asked the author to include in the wiki, that Evolution also implements this protocol, but in a slightly different way.

I received no answer within reasonable time.

I asked above what is the procedure, when the wiki shall not be changed without consulting the author and the author does not reply within reasonable time. Also my argument, that having to wait for too long in order to get blessed, can demotivate potential contributors.

I'll pull in the author here to see if he can help clear this up.

Flags: needinfo?(ben.bucksch)

On a high level, the question is how to make progess, even when some persons do not participate.

I opened this case concerning the rarely seen text “Please do not change this document without consulting the author” on https://developer.mozilla.org/en-US/docs/Mozilla/Thunderbird/Autoconfiguration, asking what to do if the author does not answer within reasonable time.

What is reasonable time, is this three months? Do you expect that whoever contacts the author, does not get answer within three monts (reasonable time), tracks her/his correspondence and raises a hand, that it is not possible to consult the author? It is rather demotivating any contribution inititive, if the processes foresee consulting persons, that for a reason that is not relevant here, do not answer.

This simple think lasts now for a year and it costs also me resources, as I have to track it it my own system.

Please take this opportunity and rethink what is necessary to change, so that Mozilla, and TB in particular, can make progress without the need to send reminders or alike. I think this is the way to make exceptionally good software.

(I am not going to contribute in the foreseenable future, as I use only software that runs under Wayland).

assuming Ben takes a pass here

Flags: needinfo?(ben.bucksch)

I have read the emails he sent to me personally. They came in form of work orders, in a way a boss speaks to an employee. I did not feel like getting into an argument. Independent of that, I did not agree with the changes proposed, otherwise I would have made the changes anyways.

Some of these documents are protocol specifications. They define how things are supposed to work. That's why they are locked.

Other parts are documenting the original project, which options we had considered, discussed, and which solutions were eventually chosen. The page clearly documents "This is the project page. The most up-to-date and readable documentation is the Description for administrators and technical users at MDC.", so the reporter probably didn't read carefully enough. Had the reporter read that very page which he asked me to change, then his emails (both of them) would have been unnecessary.

Status: RESOLVED → VERIFIED

The behaviour of Thunderbird to fetch http://example.com/.well-known/autoconfig/mail/config-v1.1.xml?emailaddress=fred@exapmle.org is not documented. Or at least not as clear as fetching http://autoconfig.example.com/mail/config-v1.1.xml?emailaddress=fred@example.com .

I tried to contact the author of https://developer.mozilla.org/en-US/docs/Mozilla/Thunderbird/Autoconfiguration per email about this. The email was returned as undeliverable. Then I edited the page. The motivation for the change field was too short, so it has not allowed to write at the same time why I want to have a change and that I tried to contact the author, but my emails to him were rejected. So I skipped the second detail. Afterwards the change was reverted and I was told that I had first to contact the author of the page. As it turned later out he has indeed received my initial email, and I have received not-delivery report for my very same message to him.

You need to log in before you can comment on or make changes to this bug.