- From: Rahul Gupta <cxres@protonmail.com>
- Date: Wed, 22 Oct 2025 16:23:29 +0000
- To: Kazuho Oku <kazuhooku@gmail.com>
- Cc: Martin Thomson <mt@lowentropy.net>, HTTP Working Group <ietf-http-wg@w3.org>
- Message-ID: <Jl-EL09boy-YwHVLRNtPZd-RAPuWL-um-3OgSafvSmtM2QfRLsdSauOlPr6n_EWV5A9UN6RWJEzC_6P>
On Wednesday, October 22nd, 2025 at 8:36 PM, Kazuho Oku <kazuhooku@gmail.com> wrote: > > > > 2025/10/22 19:06、Rahul Gupta <cxres@protonmail.com>のメール: > > > > > > > On Wednesday, October 22nd, 2025 at 3:19 PM, Martin Thomson <mt@lowentropy.net> wrote: > > > > > > > On Wed, Oct 22, 2025, at 18:07, Rahul Gupta wrote: > > > > > > > > > > > > > > > The intermediary that does not understand > > > > > > buffer-delay as a key can still ignore it. > > > > > > > > > > > > > > > > > > > > > > > But if that is the only thing in the field, it then has to ignore the entire thing. At least with ?1 there is something there: the basic functions. > > > > > > I am not sure if that follows from <https://datatracker.ietf.org/doc/html/rfc9651#section-3.2-1> and <https://datatracker.ietf.org/doc/html/rfc9651#section-3.2-3>. It might look odd, but you could have an empty Dictionary. The header itself can imply the basic function. > > > IIUC, those sections state that, when a list or a dictionary is empty, not only the field value but the entire field is omitted. > Section 4.1 explicitly says: If the structure is a Dictionary or List and its value is empty (i.e., it has no members), do not serialize the field at all (i.e., omit both the field-name and field-value). > > https://datatracker.ietf.org/doc/html/rfc9651#section-4.1-2.1.1 When responding to Martin's example, I was only thinking about parsing the header with zero known keys. In that alternate universe, the intermediary would serialize the resulting empty header as Incremental: value=?1 for forward transmission. (and "value" will then be the only defined key in this draft, as described in my previous post). In any case, the only benefit of using a Dictionary (apart from sometimes saving a few bytes) would be that key-values can themselves have parameters (whereas parameters cannot). And I recognize that I could argue from your side that the requirement of such extensibility is possibly contrived.
Attachments
- application/pgp-keys attachment: publickey_-_cxres_protonmail.com_-_0x0CEC7748.asc
Received on Wednesday, 22 October 2025 16:23:39 UTC