The modified text (and the history of the current text being removed) risks giving the false impression that an extension cannot add a new pseudo-header field when we have an RFC which demonstrates they can, at least for HTTP/2. As such, I’m hesitant to take even that change. The current text is not wrong.
Perhaps it would be clearer to say “MUST NOT…; however, an extension could negotiate a modification of this restriction.” From: Lucas Pardue <[email protected]> Sent: Friday, October 16, 2020 10:08 PM To: Martin Duke <[email protected]> Cc: Mark Nottingham <[email protected]>; Lars Eggert <[email protected]>; WG Chairs <[email protected]>; Mike Bishop <[email protected]>; IETF QUIC WG <[email protected]>; Magnus Westerlund <[email protected]> Subject: Re: HTTP/3 Nits Thanks for the clarification Mark. I don't think we need to open this can. I've made https://github.com/quicwg/base-drafts/pull/4238/files to shiffy things along. Martin, any chance you can create an issue for this? Cheers Lucas
