Re: WebSocket over HTTP2 RFC6455 conformance

On Thu, Feb 20, 2014 at 2:52 PM, Takeshi Yoshino <tyoshino@google.com>wrote:

> On Thu, Feb 20, 2014 at 2:26 PM, Martin Thomson <martin.thomson@gmail.com>wrote:
>
>> On 19 February 2014 21:16, Takeshi Yoshino <tyoshino@google.com> wrote:
>> > So, you're also not interested in preserving RSV1-3 as-is even as
>> > per-message basis?
>>
>> I'm more interested in the semantics (what I described), than the
>> protocol itself.  The protocol in RFC 6455 is just one expression of
>> those semantics.  I believe that an alternative expression of the
>> semantics can be found that is more suited to it's environment (in
>> this case HTTP/2).
>>
>> This doesn't mean that what RSV1-3 represent (extensibility) is not
>> expressed.  But the idea that you need to meticulously preserve the
>> protocol, while perhaps intuitive to some, to me is a little
>> repugnant.
>>
>
> Sorry to be unclear. "As-is" in my comment was bad. I'm also talking about
> semantics. The fact "there're 3 booleans associated with each message"
> surfaces what kind of semantics can be layered over the protocol. So people
> at HyBi would be worried about that and ask for "preserve RSV1-3". It's
> possible some of them are missing "RSV1-3 as 1 bit occupation in baseband
> for the same efficiency as RFC6455", but me and the rest are just talking
> about semantics I think.
>

Also, reusing the identifier "RSV1-3" we now have in RFC6455 for those
booleans (not pointing syntax. not pointing 1 bit serialization. but
semantics) eases writing extensions that work for both RFC6455 and
WS/HTTP/2.0. This is not-so-important but another concern Yutaka and HyBi
people should have.

Anyway, your point that we should be careful when using a term "semantics"
totally makes sense.