Google's code predates all of these auto mechanisms and just has never been updated (unless someone finally got to it in the last couple years).
It basically caches successful attempts to configure for specific email domains across the entire user base, but doesnt expire entries... and it keeps several entries and ranks by successes iirc, which means it should eventually catch up with changes but it's less than stellar at it. Improvements were always below the line on the priority list. You might even be able to game it by doing successful configs some number of times, though for popular services... That could take some doing. Brandon On Wed, Sep 4, 2019, 6:41 AM Jesse Thompson via mailop <mailop@mailop.org> wrote: > Somewhat related: we had a similar problem with Gmail's "send mail as" > autoconfig for our SMTP MSA. The wrong info wasn't coming from our SRV > records, our autoconfig service, our autodiscover service, and it wasn't > using mail.<domain> or any other well-known. > > IIRC Google front-line support claimed they actively query DNS, but it > seems clear that they had it hard coded or don't actively update cached > information previously obtained from DNS. > > Could be a similar situation at Samsung. > > Cache invalidation is one of the 2 hardest problems in computer science, > after all. > > Jesse > ________________________________________ > From: mailop <mailop-boun...@mailop.org> on behalf of Ricardo Signes via > mailop <mailop@mailop.org> > Sent: Tuesday, September 3, 2019 4:27 PM > To: k...@wemonitoremail.com; mailop > Subject: Re: [mailop] seeking Samsung contact > > On Tue, Sep 3, 2019, at 14:58, Ken O'Driscoll via mailop wrote: > If it's the native email client that they bake into the handsets, I think > that gets its provider data from a DB maintained by Mozilla. If that is the > case, the good news is that it's updateable. > > Mozilla maintains ISPDB, which has no entry for Fastmail. They also > support the autoconfig.example.com mechanism, which we support. We could > get an entry in ISPDB, but since that's not what Samsung is using (since > they have old data), it seems not helpful. Or, maybe they're using some > ancient version in which we appeared… > > Our SRV records are correct. > > 🤷♂️ > > -- > Ricardo Signes (rjbs) > CTO, Fastmail > > _______________________________________________ > mailop mailing list > mailop@mailop.org > https://chilli.nosignal.org/cgi-bin/mailman/listinfo/mailop >
_______________________________________________ mailop mailing list mailop@mailop.org https://chilli.nosignal.org/cgi-bin/mailman/listinfo/mailop
