- From: Joris Dobbelsteen <joris.dobbelsteen@mail.com>
- Date: Fri, 7 Dec 2001 16:51:42 +0100
- To: "'Jeffrey Mogul'" <mogul@pa.dec.com>
- Cc: "WWW WG \(E-mail\)" <http-wg@cuckoo.hpl.hp.com>
This is IMHO rather about the fear that we get a HTTP Header Registry with numberous of useless headers, obstructing the creation of new useful headers. Using RFCs or a registry provides some advantages and disadvantages. The use of RFCs means that headers are generated with a useful description how they operate. It takes a lot of time for new headers to be documented, but I think this might be worth the effort, but only in some cases. On the other hand, creating a registry makes it possible to get a long list of mostly useless headers, you don't know where to use them for. If you want to set up a registry, ensure that you make some good RULES for adding headers to the list. Make sure the list doesn't get poluted: all the useless names. A domain name registries (e.g. the .nl domain) runs out of name-space, resulting in strange, stupid and funnie names. Even sentences without spaces are registered. Most companies just use there own headers for there products and I don't think they conflict with others. A registry ensures this doesn't happen. Some examples could be: - header names are reviewed. They must be supplied with a name and (generalized) description (of working). If agreed to the header, it's put on the list. - naming rules: add e.g. company or product name in the header General purpose headers must be published in RFCs, so others can use them. These don't follow the restrictions for the registry. I suppose you could combine both, in order to benefit from the little time needed to register a header on the registry and RFCs for general purpose headers, usable for multiple applications. This combines both strengths, but requires a registration policy... - Joris >-----Original Message----- >From: Jeffrey Mogul [mailto:mogul@pa.dec.com] >Sent: Friday, 7 December 2001 1:28 >To: LMM@acm.org >Cc: 'Mark Nottingham'; 'Graham Klyne'; 'Mark Baker'; mogul@pa.dec.com; >http-wg@hplb.hpl.hp.com; xml-dist-app@w3.org >Subject: Re: SOAP IANA considerations > > > I don't think that a "registry" of HTTP headers is appropriate, > Rather, additional HTTP headers should be documented in IETF > standards-track documents, if they are to be considered extensions > to the HTTP protocol defined by the IETF. > > It is useful to have an index of headers for implementers > to know where various headers are defined (as, say, an update > to RFC 2076), but such an index is not a registry. > >I'm confused. First of all, about where the original message in this >thread appeared (apparently not on HTTP-WG, but then I don't know how >to find the original message). > >More particularly: I don't understand the conflict between the desire >(on the part of at least one of you, apparently) to have an IANA "HTTP >Headers Registry", and Larry's desire to see HTTP headers documented in >IETF standards-track documents. > >As far as I can tell from reading RFC 2434, there is no conflict; the >document that creates an IANA registry normally specifies some sort >of review mechanism, and one "suggested" wording is: > > Specification Required - Values and their meaning must be > documented in an RFC or other permanent and readily available > reference, in sufficient detail so that interoperability > between independent implementations is possible. > >I agree that it would be Darned Good Idea to have an HTTP Headers >Registry administered by IANA, because otherwise I suspect we will >end up with a chaotic situation as more RFCs generate more header >names. > >So I think it would be appropriate to create an "IANA Considerations" >section (logically part of the HTTP specification, but presumably in >a separate document) that says something like: > > IANA Considerations > > The Internet Assigned Numbers Authority (IANA) administers the > name space for HTTP header field-name values. Values and their > meaning must be documented in an RFC, in sufficient detail > so that interoperability between independent implementations is > possible. Subject to these constraints, name assignments are > First Come, First Served (see RFC2434). > > This registry initially consists of those header field-name > values specified in RFC 2616, RFC 2617, [other RFCs TBD]. > > Future RFCs that add values to this registry MUST provide an > explicit list of such values in an "IANA Considerations" section, > for the convenience of IANA in maintaining the registry. > >The list of "other RFCs" could be potentially fairly large, since >there are a bunch of Experimental, Informational, or Proposed >Standard RFCs listed on the HTTP-WG home page (in addition to the >small set of Draft Standard RFCs). > >And, I suppose, some poor soul(s) would have to volunteer to help >IANA glean the list of header names from all of these RFCs. But we >might as well start now, rather than later. > >One more thing: I infer from the Subject line of this thread of >messages that the intent is to create this registry as a section of >the SOAP document. This is insane; it couples progress on >establishing the registry to the progress of a much more complex >technical design, and it hides the registry specification behind a >title that most people would never suspect. (OK, how many people >could guess the name of the RFC that defines the HTTP Status Code >Registry, and no fair peeking at www.iana.org to find the >back-pointer?) > >I'd suggest that we create a separate document called > > IANA Header Field-name Registry for HTTP > >that contains just the text above (and other necessary boilerplate) >and be done with it. > >-Jeff > >
Received on Friday, 7 December 2001 15:54:26 UTC