Re: http/2 and "extensions"

Certainly extensions that come in the future would be
optional-to-implement, however, the question is directed more towards
the group of things being discussed today as "extensions"... the set
of things we already know about and are discussing. For example,
suppose we decide now that jumbo frames ought to be an extension. We
could choose to define those as either optional-to-implement (in which
case, they likely won't be implemented consistently) or
mandatory-to-implement-but-optional-to-use. I mainly want to make sure
that when people here say things like "Oh, we can do N as an
extension" that doesn't really mean, "Let's kill N because we don't
like it".

On Tue, Jul 15, 2014 at 10:54 AM, Martin Thomson
<martin.thomson@gmail.com> wrote:
> On 15 July 2014 10:41, James M Snell <jasnell@gmail.com> wrote:
>> all extensions going to be optional-to-implement[?]
>
> all extensions ARE going to be optional-to-implement.
>
> You can't rely on other implementations supporting something that they
> don't know about.  The extensibility features of HTTP (header fields,
> content types) are available if you want something end-to-end.

Received on Tuesday, 15 July 2014 18:02:47 UTC