- From: Melvin Carvalho <melvincarvalho@gmail.com>
- Date: Fri, 11 Sep 2026 15:05:53 +0200
- To: "Roy T. Fielding" <fielding@gbiv.com>
- Cc: Martin Thomson <mt@lowentropy.net>, Poul-Henning Kamp <phk@phk.freebsd.dk>, Tim Bray <tbray@textuality.com>, Mark Nottingham <mnot@mnot.net>, ietf-http-wg@w3.org
- Message-ID: <CAKaEYhLWFSrswQEDKL_ximMNfZZqfpPUmjEh5Bm7anAeRgFoLg@mail.gmail.com>
On Tue, Sep 1, 2026 at 9:10 AM Roy T. Fielding <fielding@gbiv.com> wrote: > On Aug 30, 2026, at 10:30 PM, Martin Thomson <mt@lowentropy.net> wrote: > > On Sun, Aug 30, 2026, at 22:07, Poul-Henning Kamp wrote: > > That smell is "Pay-Per-View on the Internet", > > > That's *a* smell here, but this doesn't need to be that. Services with > subscription-based membership might be able to use this to ask clients to > show that their subscription is up to date, without the implication that > authentication is involved. For that, see Privacy Pass and its use in the > context of VPNs as an example where proof of payment does not lead to > authentication. > > However, it's hard to disentangle 401 from this. A decision to deny > access on the basis of (a lack of) payment is still fundamentally an > authorization decision. > > > Hmm, not sure I agree with that, but I don't think it matters much. > 402 was originally defined before SSL (and maybe even Basic auth). > It certainly predated me. > > 402 had a very simple definition that, fortunately, nobody deployed. > === > > *PaymentRequired 402* > > The parameter to this message gives a specification of charging schemes > acceptable. The client may retry the > > request with a suitable ChargeTo header. > > > [... Request Header fields] > > *1.7.8 ChargeTo:* > > This line if present contains account information for the costs of the > application of the method requested. The > > format is TBS. The format of this field must be in extensible form. The > first word starts with a specification of the > > namespace in which the account is . (This is similar to extensible URL > definition.) No namespaces are currently > > defined. Namespaces will be registered with the registration authority . > > The format of the rest of the line is a function of the charging system, > but it is recommended that this include a > > maximum cost whose payment is authorized by the client for this > transaction, and a cost unit. > > > [...Response Header fields] > > *2.4.13 Cost: TBS* > > The cost of retrieving the object is given. This is the cost of access of > a copyright work. Format of units to be > > specified. Currently refers to an unspecified charging scheme to be agreed > out of band between parties. > > > [...Registry] > > The HTTP Registration Authority is responsible for maintaining lists of: > > - Charge account namespaces (see ChargeTo field above) > === > > I have a distinct memory of a hallway discussion, somewhere, > in which the two people who attempted to implement 402 agreed > that the definition was wrong, couldn't agree on why, wouldn't accept > each other's fix, but would be horribly upset if we were to simply delete > the status code (like we did everything else that wasn't interoperating). > > Anyway, that's why 402 is reserved. It was in the pre-IETF spec > but did not have two interoperable implementations, yet there was > general agreement on what it means, semantically, which made > it safe to send even if the client has no idea how to make a payment. > > I was told to wait until they figured it out. I'm still waiting. > Tim Berners-Lee said something to me on a call years ago, words to the effect of: "we thought 402 was important at the time, which is why we put it before 403." That seems consistent with the pre-IETF definition Roy quoted. He suggested the W3C Web Payments Community Group as the place to work it out, though that group has since gone dormant. It would be possible to reopen it if there is interest. I agree with Martin and Nico that 402 sits naturally alongside 401: both tell the client that something further is required before the resource can be accessed. In that sense it is closer to 401 than to 403, which expresses refusal rather than a challenge. To add to Mark's list of deployments, there have been many uses of 402 over the years, including one of my own for Solid [1]. It returns a 402 with deposit instructions in the body, together with Cost and Balance response headers. That is not far from the ChargeTo/Cost shape Roy quoted. I'd be happy to compile an ecosystem review of who is sending 402 today, what semantics they attach to it, and what they put in the response, if that would be useful input to Mark's draft. My expectation is that this space will increasingly be shaped by pay-as-you-go agents. That fits Roy's "figure out who wins" framing and, to me, argues for the minimal semantics in Mark's draft rather than trying to define a payment framework. [1] https://jss.live/docs/features/payments > > I don't have an opinion on whether "the IETF" should weigh in on > the current proposals, other than to suggest that this code was > reserved by the IETF to define for those semantics (eventually). > > I think they are welcome to submit a draft (or two) when ready. > Or they can all just implement their own versions and eventually > figure out who wins (or who breaks the least), and then maybe > try to convince the IANA registry expert to change the spec pointer. > > ....Roy > >
Received on Friday, 11 September 2026 13:06:09 UTC