- From: Willy Tarreau <w@1wt.eu>
- Date: Fri, 7 Oct 2016 08:05:31 +0200
- To: Julian Reschke <julian.reschke@gmx.de>
- Cc: HTTP Working Group <ietf-http-wg@w3.org>
Hi Julian, On Thu, Oct 06, 2016 at 03:16:34PM +0200, Julian Reschke wrote: > Now we can't have these specs rely on an abandoned Internet Draft, right? No I disagree with you here. Drafts are drafts. They always have early adopters (otherwise they don't make progress), but early adopters know what risks they are taking. Otherwise you end up censoring yourself when writing a draft just in case someone would implement it, which is not acceptable at all. > So what do we do? > > a) "finish" draft-ietf-httpbis-jfv, documenting known issues, and publish > this as informational or experimental? It could be an option at least not to lose all the work done on this and to serve as input for future designs (specifically the known issues). But I don't know if that requires lots of work on your side or not. It could end up like those few RFCs saying "foobar considered harmful", documenting a design choice and its impacts and side effects. > b) conclude work on this draft, and let me continues to finish my pre-WG > draft (with the same goals as above) > > c) give up, and let the users of the draft figure out a solution? Well these two ones are not exclusive of the first one. If it is possible to "flush the pipe" with option a) without spending too much effort on it, the lowest the effort, the highest c) becomes prevalent. And in all cases, b) definitely is up to you (I think). Maybe the draft should be renamed to something like "The limitations of using JSON for header field values" or something like this, explain the design, the fact that there are known adopters, and enumerate the known issues without too much details (just to discourage future adopters unless they're fine with these limitations). Just my two cents, Willy
Received on Friday, 7 October 2016 06:05:57 UTC