https://logs.xmpp.org/council/2021-02-10?p=h#2021-02-10-f8b1c3d52716488e
1) Roll Call Present: Daniel, Zash, Jonas, Dave Apologies: Georg 2) Agenda Bashing Silence. 3) Editor's Update * New XEPs: - XEP-0455: Service Outage Status 4) Items for voting None. 5) Pending Votes Jonas notes the lack of discussion on PR #1032 (Data Forms Clarifications), which he did intend to start, but his eventful life had other ideas. Dave votes on Implicit WebSocket Endpoints: -1 (veto without prejudice, on the assumption that a different proposal will emerge; it implies a server default of listening on unencrypted sessions, and devalues XEP-0156; expecting something closer to a 'suggested defaults' with both test and production recommendations.) Jonas notes a small PR to update the XEP [1]. Daniel is glad Dave is doing the vetoing, so he doesn't have to - Dave suggests he still should, in case of other interesting reasoning. Daniel votes: -1 (a XEP that defines a recommendation for default ports and paths is interesting, but this one goes way beyond and tries to influence client behaviour, which might lead to poor choices with regard to XEP-0156; agree with Dave that starting a fresh XEP is preferable to heavily editing this one.) Jonas votes: -1 (let's not repeat the A/AAAA fallback mistake when there's no evidence it's needed for practical interoperability concerns.) Kev isn't convinced that A/AAAA fallback was a mistake at the time. Jonas expects '.well-known' will require some IANA interaction. Dave can see that defaults are nice for ease of testing and deployment, and recommendations for production deployment are great too, but that seems like a fundamentally different XEP than this one. Kev wonders why production deployment recommendations are so great - if a client is doing lookups, isn't that entirely the deployment's choice, in which case why is one recommendation better than another? Dave thinks that suggesting websockets run on port 443 over TLS is useful advice, and materially better than 5280 without TLS; Jonas suggests that would be an Informational XEP - Zash and Daniel support that idea. Kev agrees that TLS is great advice, though is less sure on port 443, but can see an argument for it (and any other port recommendation would be a bad thing). Kev thinks an XEP saying "always use TLS, and port 443 might be a good choice" would likely be less harmful than other possibilities - Zash thinks there might already be RFCs saying "always use TLS!" Jonas asks whether anyone is willing and able to take on the work of making this happen - Dave was assuming Flow (the author) or Sonny would want to; Jonas thought Sonny looked interested in moving this towards a default recommendation. Zash votes: -1 (fallbacks bad, TLS good, what everyone else said; try again with a new proposal.) Jonas assumes PR #1032 will be discussed on-list. Daniel doesn't see how it wouldn't require a namespace bump - Zash prays to a deity - Dave thought it seemed okay - Jonas thinks it's worth going the whole way and also killing the 'reported' element at the same time. Daniel thinks the usage of "clarifying" has been pretty liberal in the past - Jonas would prefer to forbid that word from defiling any XEP PRs again. Dave presumes that either people do care about the ordering, or they don't; a fix would be to say "SHOULD send in order" and "MUST accept in any order" - Jonas and Daniel could live with that; Daniel queries why anyone would send in order if it will always be accepted in any order. Dave accepts this needs further discussion. Jonas sent a reply on-list [2]. 6) Date of Next 2021-02-17 1600 UTC 7) AOB Jonas has a pleasant conversation with his echo, over kaffee und kekse. 8) Close Thanks all for one, and one for all. [1] https://github.com/xsf/xeps/pull/1035 [2] https://mail.jabber.org/pipermail/standards/2021-February/038148.html
_______________________________________________ Standards mailing list Info: https://mail.jabber.org/mailman/listinfo/standards Unsubscribe: standards-unsubscr...@xmpp.org _______________________________________________
