You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
It seems to me a bit forced to try to have http2 logic inside the http 1.1 client. I think it might make sense to move shared stuff up one level and separate these into a HTTP1Client and HTTP2Client.
The text was updated successfully, but these errors were encountered:
Hmm, it sounds doable; but how do we handle the fact that we start the negotiation in the same way, it might seem counterintuitive to have to decide whether to use one or the other depending on the ALPN step.
A couple of things can bring some light on this, e.g., if not using TLS we cannot even try an H2 request, or if the flag is disabled, we do not do that.
One thing that can work is to remove the writeH2 and scope it to a mid-term dispatcher; that, if ALPN states to use H2 we simply use it instead, and pass the socket (pretty much as you opened PR, but the control is passed to that dispatcher, and the main dispatcher just passes the handler)
It seems to me a bit forced to try to have http2 logic inside the http 1.1 client. I think it might make sense to move shared stuff up one level and separate these into a HTTP1Client and HTTP2Client.
The text was updated successfully, but these errors were encountered: