On Thu, 2015-02-19 at 03:39 -0800, Brien Colwell wrote: > Hi Oleg, > > For normal requests where there is one response per request, could each > connection just be a pipeline on an async channel, where the pipeline could > have a blocking interface? > >
It is certainly a possibility. We might move core transport components entirely to NIO and provide a simple blocking interface on top of async core. It is an option. > > I get that sever push in the general sense is outside the HttpClient > design as is. > Without a reasonable support for multiple streams the HTTP/2 support would simply be useless. I am still trying to understand what subset of features HC could support initially and what features could be added incrementally over time. It any case it is going to be a slow and painful process. Oleg > Brien > > > On Thursday, February 19, 2015, Oleg Kalnichevski <[email protected]> wrote: > > > On Thu, 2015-02-19 at 03:16 -0800, Brien Colwell wrote: > > > Hi all, > > > > > > Is anyone working on HTTP/2 support for HttpClient? > > > > > > Seems like HttpRequestExecutor could be reworked to support the new > > > protocol. > > > > > > Best, > > > Brien > > > > > > > I am going through the latest draft at the moment and still trying to > > assess the impact HTTP/2 is going to make on the existing design. It > > looks like the impact is going to be massive. I am not even sure HTTP/2 > > can even be reasonably implemented with the classic (blocking) I/O. > > > > Oleg > > > > > > > > --------------------------------------------------------------------- > > To unsubscribe, e-mail: [email protected] <javascript:;> > > For additional commands, e-mail: [email protected] <javascript:;> > > > > --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
