You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
MikeBishop opened this issue
Jun 8, 2017
· 4 comments
Labels
editorialAn issue that does not affect the design of the protocol; does not require consensus.parkedAn issue that we can't immediately address; for future discussion.
Several existing sub-issues around what is version-independent. I'm proposing that the cross-version envelope of QUIC should be a separate final RFC, and therefore a separate document now. If we landed that split, it would resolve several other questions about what, specifically, is supposed to be cross-version.
I would argue that version negotiation and everything that's visible outside the encryption envelope belongs in this separate draft. It's the piece that will ossify, and that we're willing to have ossified. Everything that's version-dependent belongs in the specific version, which is the Transport draft.
However, I'll note that the current split means that packet protection and framing live inside the version. That means there's a lot that will need to be re-stated in a new version -- are we okay with that?
no more documents please - the crossrefs are hard enough to deal with already
Perhaps a short section in transport recapping the invariants used elsewhere in the doc would be helpful as an alternative?
mnot
changed the title
Move version-independent pieces to a "core" document
Standalone version-independent document
Jun 20, 2017
mnot
added
the
editorial
An issue that does not affect the design of the protocol; does not require consensus.
label
Jun 20, 2017
MikeBishop
added
parked
An issue that we can't immediately address; for future discussion.
and removed
proposal-ready
An issue which has a proposal that is believed to be ready for a consensus call.
labels
Aug 15, 2017
janaiyengar
added
proposal-ready
An issue which has a proposal that is believed to be ready for a consensus call.
and removed
proposal-ready
An issue which has a proposal that is believed to be ready for a consensus call.
labels
Aug 15, 2017
From editors meeting: park this issue for now. There is value in separating these invariants, and perhaps something along the lines of mcmanus@'s proposal makes sense. The plan is to do something like that by the Singapore IETF timeline.
editorialAn issue that does not affect the design of the protocol; does not require consensus.parkedAn issue that we can't immediately address; for future discussion.
Several existing sub-issues around what is version-independent. I'm proposing that the cross-version envelope of QUIC should be a separate final RFC, and therefore a separate document now. If we landed that split, it would resolve several other questions about what, specifically, is supposed to be cross-version.
I would argue that version negotiation and everything that's visible outside the encryption envelope belongs in this separate draft. It's the piece that will ossify, and that we're willing to have ossified. Everything that's version-dependent belongs in the specific version, which is the Transport draft.
However, I'll note that the current split means that packet protection and framing live inside the version. That means there's a lot that will need to be re-stated in a new version -- are we okay with that?
See PR #600.
The text was updated successfully, but these errors were encountered: