W3C home > Mailing lists > Public > ietf-http-wg@w3.org > October to December 2013

RE: Questions about ALPN

From: Andrei Popov <Andrei.Popov@microsoft.com>
Date: Tue, 29 Oct 2013 17:58:50 +0000
To: Mike Bishop <Michael.Bishop@microsoft.com>, Mark Nottingham <mnot@mnot.net>, "Joseph Salowey (jsalowey)" <jsalowey@cisco.com>
CC: Martin J. Dürst <duerst@it.aoyama.ac.jp>, "ietf-http-wg@w3.org" <ietf-http-wg@w3.org>
Message-ID: <8ae3fbbfb31543aeadc40fd79a0a976e@BL2PR03MB194.namprd03.prod.outlook.com>
If protocol ID segregation is desired, perhaps ALPN protocol IDs could be segregated by prefix, e.g. "HTTP/X" protocol IDs for Web applications.

Also, the ALPN protocol ID registry operates under the "Expert Review" policy, the idea being that expert review will prevent inappropriate registrations.

-----Original Message-----
From: Mike Bishop 
Sent: Tuesday, October 29, 2013 10:45 AM
To: Mark Nottingham; Joseph Salowey (jsalowey)
Cc: Martin J. Dürst; Andrei Popov; ietf-http-wg@w3.org
Subject: RE: Questions about ALPN

HTTP(S) is allowed on any port -- 80 and 443 are used when the port's not specified, and most likely to work across intermediaries, but a URI can always specify a different port.  I would presume the same is true of any other protocol.  I don't think segregating the registry by port is going to fly very far.

-----Original Message-----
From: Mark Nottingham [mailto:mnot@mnot.net] 
Sent: Monday, October 28, 2013 10:16 PM
To: Joseph Salowey (jsalowey)
Cc: Martin J. Dürst; Andrei Popov; ietf-http-wg@w3.org
Subject: Re: Questions about ALPN

On 29/10/2013, at 4:57 AM, "Joseph Salowey (jsalowey)" <jsalowey@cisco.com> wrote:

> [Joe] This may be part of the disconnect.   The ALPN extension was not designed to be solely used by the HTTPbis framework.   At this point I don't think it is safe to assume that everything in the registry will be relevant to HTTPbis.   Is this a problem for HTTPbis?

Well, it's an interesting question. We need a registry that the various things can point to. If some of the contents of this registry aren't appropriate, we might need to have a separate registry. Which means that people would have to do *two* registrations, which seems like busy work.

What if you didn't have any registry at all -- what if the name space of possible ALPN tokens being used was scoped by the port in use? Just thinking out loud.

Mark Nottingham   http://www.mnot.net/
Received on Tuesday, 29 October 2013 17:59:42 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 17:14:19 UTC