Re: Making Implicit C-E work.

On Wed, Apr 30, 2014 at 3:55 AM, Matthew Kerwin <matthew@kerwin.net.au>wrote:

> [snipping]
>
> On Apr 30, 2014 4:44 PM, "Roberto Peon" <grmocg@gmail.com> wrote:
> >
> >> But even so, why do you have to fix it in HTTP/2? And why does it hurt
> h2 to *not* fix it?
> >
> > Compression is an important part of making latency decrease/performance
> increase, and, frankly, there is little practical motivation to deploy
> HTTP/2 if it doesn't succeed in reducing latency/increase performance.
>
> Do the other improvements in HTTP/2 not give those successes? Or are you
> saying that they're not enough without ubiquitous payload compression?
>

Each helps for specific usecases. And I am saying that ubiquitous
compression of the payload is a big deal for many of the browsing cases
where the resources are larger, as is very often the case with HTML/css/js
and other resources.


> >> > The proxy, when forwarding the server's response to the HTTP/1
> client, must ensure that the data is uncompressed when forwarding to the
> HTTP/1 client since the client didn't ask for c-e gzip.
> >>
> >> Cache-Control:no-transform explicitly forbids the proxy from altering
> the representation. It's not allowed to decompress it.
> >
> > In fact what we're doing is offering two representations simultaneously.
> >
>
> It's a very messy way of doing it, though, and it makes me nervous. Too
> many edge cases, too many potential holes. That's ok for a de facto
> standard or something, but not for a formal IETF spec. And it strikes me as
> a big disincentive to adoption.
>
> To get it right would hold back HTTP/2 (unneccessarily, I say), so don't
> rush it in there. Hold off for the next iteration. I'd be happy if HTTP/3
> focused solely on fixing compression. I suspect others would be too.
>
Arguably the t-e stuff is more complicated, so I guess I disagree there.
I'm certainly up for other fixes in HTTP/3, but I don't feel that the
feature which is already implemented is something we should throw out now
when the fix is relatively easy.
-=R