[Libwebsockets] http2 support

Joel Winarske joel.winarske at gmail.com
Tue Feb 7 01:28:52 CET 2017


Bummer.  Thanks for the status.

As an aside I was doing some bench marking today (site load time) against
our in-house C++ solution - libwebsockets outperforms it :)  Nice work!


Joel


On Mon, Feb 6, 2017 at 3:34 PM, Andy Green <andy at warmcat.com> wrote:

>
>
> On February 7, 2017 6:04:54 AM GMT+08:00, Joel Winarske <
> joel.winarske at gmail.com> wrote:
> >Hi Andy,
> >
> >What's the state of http2 support in master?
>
> "Experimental"... a couple of years ago I got it to the point it could
> serve one h2 stream on a connection over ssl/alpn, via curl or so.  Since
> this was a huge amount of work - it's nothing like http/1.x, just passing
> headers requires all kinds of specialized adaptive binary encodings - I
> wrote up how to reproduce it in the README and here, and nobody was
> interested.
>
> Also you should know despite my arguing for it, the h2 guys seem to regard
> websockets as declasse and discussion about it to stink up their pure and
> beautiful h2 list.  That was the case a couple of years ago when they were
> coming to the end of defining the standard, they defined no way to carry
> ws[s] on h2 and ignored arguments for it.  And it's the case a couple of
> months ago when it came up again, they tried to push discussion on a shut
> down Hybi list as if scraping shit off their shoe.  So there is no
> standardized way to carry ws[s] on h2, the official line is do it on
> http/1, or more accurately, "let them eat http/1".
>
> For those reasons, zero interest until recently, zero help, and no way to
> tie it to ws, it's stuck at about 85%.
>
> -Andy
>
> >
> >Thanks,
> >Joel
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://libwebsockets.org/pipermail/libwebsockets/attachments/20170206/12dffc1b/attachment-0002.html>


More information about the Libwebsockets mailing list