Re: [art] Auto-configuring Email Clients via WebFinger

["John Levine" <[e-mail]>]

Some do 6186, most don't.  I publish the srv records but my users all have to config manually. The real question of course is whether mua makers care about this enough to use it.Autocorrectly,John
-------- Original message --------From: Dave Cridland <[e-mail]> Date: 7/23/19  10:25  (GMT-08:00) To: John R Levine <[e-mail]> Cc: "General Area Review Team ([e-mail])" <[e-mail]> Subject: Re: [art] Auto-configuring Email Clients via WebFinger On Tue, 23 Jul 2019 at 17:27, John R Levine <[e-mail]> wrote:On Tue, 23 Jul 2019, Dave Cridland wrote:
>> There are places where, say, mailboxes with local parts starting with
>> A-M are on one IMAP server and those starting with N-Z are on another.
>> It's hard to see how you could do that with SRV records tied to the
>> domain.
>>
> Sure. I'm just thinking that anywhere you say "IMAP server" we'd put an SRV
> resolution.

In general I like SRV just fine but in this case, the config already 
provides a hostname and port so what does it add?  I don't thinkml MUAs 
are likely to use a backup IMAP server if the main one is unreachable.Hmmm... "thinkml"? Sounds familiar.But to the point,  I don't know if MUAs will do SRV or not, if it's presented to them in a standard. On a quick skim, some decided RFC 6186 wasn't useful: https://mail.gnome.org/archives/commits-list/2012-November/msg01976.html and others (like K9, for example) decided it was.I do know, however, that Google deploy RFC 6186 records for gmail.com, so there's hope - it's a useful mechanism for allowing multiple endpoints with preferences etc, and I'm quite sure that we do have cases where multiple IMAP hosts are deployed.We could bake that functionality into the autoconfig fomat, of course, but it feels like refreshing the config can happen at a different cadence to a SRV lookup, and besides, SRV works well.Dave.