Re: [core] Interest in defining CoAP over Unix Domain sockets?

[Esko Dijk <[e-mail]>]

There is also the option to re-use the RFC 8075 URI embedding scheme.  (Section 5.4.1.1 and 5.4.2.1 have examples)
This is the syntax of a CoAP proxy server, but it can be used also for regular CoAP servers.

It could look like this then:

coap+unix:///some/local/abspath/coap:///another/path

Or, when doing CoAP over TCP:

coap+unix:///some/local/path/coap+tcp:///another/path

So then only one new additional scheme would be required.

In the above proposal the host component is not included, since it's not needed, as the host is the socket listener process itself. This follows the section 5.4.2 of RFC 8075 approach. In case we want a hostname for clarity or future extendibility reasons, it may look like e.g.:

coap+unix:///some/local/abspath/coap://-/another/path
coap+unix:///some/local/abspath/coap://[::1]/another/path
coap+unix:///some/local/abspath/coap+tcp://[::1]/another/path

And if the socket endpoint is able to operate as a CoAP reverse proxy then URIs like below can also be used:
coap+unix:///some/local/abspath/coaps://myserver.example.com/another/path

Esko

PS this works based on the constraint that your Unix path segment cannot equal a Coap scheme name plus ':' character.
Some percent-encoding may be needed like for the square brackets.

-----Original Message-----
From: core <[e-mail]> On Behalf Of Michael Richardson
Sent: Thursday, December 15, 2022 17:05
To: Carsten Bormann <[e-mail]>; Jon Shallow <[e-mail]>; [e-mail] WG ([e-mail]) <[e-mail]>
Subject: Re: [core] Interest in defining CoAP over Unix Domain sockets?

Carsten Bormann <[e-mail]> wrote:
    >> coap+unix:///some/local/path/another/path

    > Which part is the socket path, and which part is the URI path?

Oh. Yes.  I wasn't thinking that there would ever be a URI path, because in
Jon's use case, he is always just proxying to some servers.
But, of course there are other use cases.



_______________________________________________
core mailing list
[e-mail]
https://www.ietf.org/mailman/listinfo/core