Use quic= instead of v= #235
Conversation
draft-ietf-quic-http.md
Outdated
| representation. | ||
| On receipt of an Alt-Svc header indicating HTTP/QUIC support, a client MAY | ||
| attempt to establish a QUIC connection on the indicated port and, if successful, | ||
| send HTTP requests using the mapping described in this document. Servers SHOULD |
There was a problem hiding this comment.
I think that I would rather the SHOULD here be a MUST. (Then the "but" that follows isn't a "but".)
|
Section "Registration of Version Hint Alt-Svc Parameter" still refers to "v" rather than "quic". |
draft-ietf-quic-http.md
Outdated
| establishment failure, in which case the client should gracefully fall back to | ||
| HTTP/2. | ||
| Origins SHOULD list only versions which they support, but MAY omit supported | ||
| versions for any reason. |
There was a problem hiding this comment.
In the host-neutral case the origin is listing versions for the HTTP/QUIC endpoint. I don't think there is a requirement for the advertising host to support any version of QUIC, right?
There was a problem hiding this comment.
The origin has to support those versions at the advertised endpoint, regardless of whether the particular host that sent the header is that endpoint or not.
draft-ietf-quic-http.md
Outdated
| version "Q034" and version 0x00000001, it would specify the following header: | ||
| When multiple versions are supported, the "quic" parameter MAY be repeated | ||
| multiple times in a single Alt-Svc entry. For example, if a server supported | ||
| both version "Q034" and version 0x00000001, it could specify the following |
There was a problem hiding this comment.
Do we have enough context to understand what "Q034" is, aside from knowing that an ASCII/UTF-8 encoding of "Q" is 0x51?
There was a problem hiding this comment.
It's still there primarily because I wanted to be able to show the multiple-entry case. However, it might make more sense to change that to a draft version and leave the ASCII-like versions out entirely.
There was a problem hiding this comment.
Mm. Except that the draft versions won't be in the final QUIC RFC, so there won't be any context for that reference once it's eventually published. Maybe not.
There was a problem hiding this comment.
We could add that context, or we could add a "note to the RFC Editor" to remove Q034 and the related text on publication.
There was a problem hiding this comment.
But then we'd still have to give them something to replace it with, and keeping the "how to do multiple" is worthwhile. I've expanded slightly, to refer to it as the version that would be rendered in ASCII as Q034. Given that we allow pretty free-form experimentation in that region, I don't know that we need to elucidate why a server might support an experimental version.
|
I still have a poor understanding of how @mnot intends to run the process here WRT |
draft-ietf-quic-http.md
Outdated
|
|
||
| Alt-Svc: hq=":443";v="x1";v="cQ034" | ||
| Alt-Svc: hq=":443";quic=1;quic=51303334 |
There was a problem hiding this comment.
I suspect this is obvious, but it might be nice to explain why 0x00000001 is ASCII encoded as "1" and not "00000001". I assume the idea is that leading '0's may be removed from the hex encoded string?
There was a problem hiding this comment.
Basically, yes. I've made that explicit in a9e47f2.
|
I suppose you could reserve a version for documentation purposes. E.g
DEADBEEF
…On 1 Feb 2017 5:56 p.m., "Mike Bishop" ***@***.***> wrote:
***@***.**** commented on this pull request.
------------------------------
In draft-ietf-quic-http.md
<#235>:
> version-number = 1*8HEXDIG; hex-encoded QUIC version
-When multiple versions are supported, the "v" parameter MAY be repeated multiple
-times in a single Alt-Svc entry. For example, if a server supported both
-version "Q034" and version 0x00000001, it would specify the following header:
+When multiple versions are supported, the "quic" parameter MAY be repeated
+multiple times in a single Alt-Svc entry. For example, if a server supported
+both version "Q034" and version 0x00000001, it could specify the following
But then we'd still have to give them something to replace it with, and
keeping the "how to do multiple" is worthwhile. I've expanded slightly, to
refer to it as the version that would be rendered in ASCII as Q034. Given
that we allow pretty free-form experimentation in that region, I don't know
that we need to elucidate *why* a server might support an experimental
version.
—
You are receiving this because you commented.
Reply to this email directly, view it on GitHub
<#235>, or mute the thread
<https://github.com/notifications/unsubscribe-auth/AGRFtQm-YhTz-maPVm_o9JUUXdJpu5Ehks5rYMdlgaJpZM4LuMve>
.
|
|
I think we merge the changes to keep our repo and CL stack sane. The chairs can call consensus and if there's no consensus, we can always roll back. |
|
Yep, I'll press the button so that you can blame me when we realize that don't have consensus. |
|
@martinthomson (too many martins on this repo!) --
I think the label you're looking for is This is all amenable to changes, of course, but let's see if we can all get on the same page first. As a reminder, see also: |
Fixes #229