On 2024-11-20 at 14:10:42 UTC-0500 (Wed, 20 Nov 2024 14:10:42 -0500)
Miles Fidelman via mailop <mfidel...@meetinghouse.net>
is rumored to have said:
Hi Folks,
Not really sure where to ask this question, so I'll start here.
I've been thinking of migrating our mail infrastructure to a virtual
server, running in the Web3 IPFS cloud - without having a physical IP
address attached to it.
The obvious question becomes: How do I publish an address to it?
I'm almost concerned that the above query may just be a joke: a LLM
hallucination turned into a question. IPFS is a storage layer, not an
email protocol. What does it even mean to have a mail server "running"
on IPFS? It might *USE* IPFS, but the mail server itself never exposes
its storage layer to clients. So the whole think is nuts but I'll act
like it isn't...
You can't. If it has no IP address at which it is listening for TCP
connections on port 25, no one will ever know how to connect. SMTP is
defined as a protocol that uses TCP on port 25. IMAP is a protocol that
uses TCP over ports 143 or 993.
How do I set up an MX record to point to a socket listener that has
nothing but an IPFS CID to identify it?
You can't. SMTP uses TCP over IPv4 or IPv6. As specified, it looks for a
MX record that resolves to a *HOSTNAME* that MUST have an A record.
You are looking to define something else, which would by necessity be
entirely new.
I can set up a DNS_Link TXT record, pointing to an IPNS record - but
MX records have to resolve to an FQDN.
Any thoughts?
SMTP uses an A record if there is no MX record. It has no provision for
using a TXT record of any sort. If you cannot create a valid MX and/or A
that explains how to connect on port 25 using TCP to one or more
specific IP addresses, you cannot use SMTP.
SMTP might be able to use other underlying transport protocols, if they
act a lot like TCP.
Any thoughts on where to ask the question?
If you want to do this, you will need to figure out a replacement for or
major enhancement to SMTP on your own and convince others to work with
your alternative protocol. Maybe seek out a relevant IETF WG?
I wish you good luck, even though I think the whole idea is a pointless
waste of energy which will go nowhere. If you actually get anywhere with
the concept, it might be interesting.
--
Bill Cole
b...@scconsult.com or billc...@apache.org
(AKA @grumpybozo@toad.social and many *@billmail.scconsult.com
addresses)
Not Currently Available For Hire
_______________________________________________
mailop mailing list
mailop@mailop.org
https://list.mailop.org/listinfo/mailop
