Resent-Date: Fri, 02 Aug 2024 23:10:50 +0000
Resent-Message-Id: <E1sa1QA-00CKcS-0E@mab.w3.org>
Received-SPF: pass (pan.w3.org: domain of gmail.com designates 209.85.167.176 as permitted sender) client-ip=209.85.167.176; envelope-from=phluid61@gmail.com; helo=mail-oi1-f176.google.com;
MIME-Version: 1.0
References: <CAF3KT4QZzx+FXOUHZoy+gPqJjQ+4KdOC+_29vbUANNtZQS4c+A@mail.gmail.com>
 <ba56fad8-e121-4c06-9a2d-783ef82471e0@gmx.de> <CAJV+MGz8hUTqar51V9wV=WPnWETDK+ECjWCTXYS92xXM5HEF_w@mail.gmail.com>
 <cd25a358-1e8a-43b7-ba61-3d16ad28b1e4@gmx.de> <CAF3KT4Q=ezzA2aCHyPg=k583n6vP4gTGP+wxKz+sQezD=GnowQ@mail.gmail.com>
 <CAJV+MGxH4wPK__Z4mGfD3j-KaHpSVjQBFLyM1u+ZwfaSsZGMRQ@mail.gmail.com>
 <247701dae3b2$a1ea1f50$e5be5df0$@gmail.com> <CAJV+MGwZmeAhxK9ooOXJr_jdkqN-Mk_oLS8pWAsdTSCwToaAdg@mail.gmail.com>
 <CAF3KT4Qaw+0LJtgL5=+ojOsmpgz2SUa1=Oz6QTA2=xE6Fy-KMw@mail.gmail.com>
 <026321eb-057c-4f5c-9e2a-4d4b7b290a2d@gmx.de> <CAF3KT4Tm4wHRi3J6A2JEb2329o=prBh4UiUVx_eqOUVkN-ZoiA@mail.gmail.com>
In-Reply-To: <CAF3KT4Tm4wHRi3J6A2JEb2329o=prBh4UiUVx_eqOUVkN-ZoiA@mail.gmail.com>
From: Matthew Kerwin <matthew@kerwin.net.au>
Date: Sat, 3 Aug 2024 09:10:32 +1000
Message-ID: <CACweHNBsv_P7iA=7Q3RNpxPhof5vC7DyjbsCsD1KPATwjgY7hA@mail.gmail.com>
To: Josh Cohen <joshco@gmail.com>
Cc: Julian Reschke <julian.reschke@gmx.de>, Patrick Meenan <patmeenan@gmail.com>, cxres@protonmail.com, 
	ietf-http-wg@w3.org
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Subject: Re: Method Mania
Archived-At: <https://www.w3.org/mid/CACweHNBsv_P7iA=7Q3RNpxPhof5vC7DyjbsCsD1KPATwjgY7hA@mail.gmail.com>
Resent-From: ietf-http-wg@w3.org
Resent-Sender: ietf-http-wg-request@w3.org
Precedence: list

On Sat, 3 Aug 2024 at 05:49, Josh Cohen <joshco@gmail.com> wrote:
>
> Thanks for the clarification.  Looking at the http2 rfc9113[1], bottom of=
 section 8.3.1, says:
> "Individual HTTP/2 requests do not carry an explicit indicator of protoco=
l version. All HTTP/2 requests implicitly have a protocol version of "2.0""
>
> So, there's no longer an explicit version string, or header.
>
> The reason I asked was when considering new methods.  The option of revvi=
ng the version of the messaging/semantics layer isn't feasible.
>
> https://datatracker.ietf.org/doc/html/rfc9113#section-8.3.1
>

I'd do it the same way I would in HTTP/1.1 =E2=80=93 either
1. send ":method: FOOBAR" and see what the server does, or
2. use OPTIONS. Either check that 'FOOBAR' is in the Allow field, or
if I want more rich capability/compliance advertisement, send a
"X-WantToUse-Foobar: 1" header field, and have the server signal its
compliance appropriately

That would always have been safer than just assuming a server that
responds to "GET / HTTP/1.2" actually paid heed to the "2"

Cheers
--=20
  Matthew Kerwin [he/him]
  https://matthew.kerwin.net.au/