I’m giving this serious consideration, and will reply more comprehensively within the next couple of days.
-Chris > On 26 Jul 2018, at 19:20, David Lloyd <david.ll...@redhat.com> wrote: > >> On Wed, Jul 25, 2018 at 9:43 AM Chris Hegarty <chris.hega...@oracle.com> >> wrote: >> >> Clearly the request builder `timeout` method can be used to avoid >> extremely long connection timeouts ( as demonstrated below ), but I see >> Bernd's call for more fine grained control over various aspects of the >> request. >> >> I'm not opposed to an "HttpRequest.Builder.connectTimeout` method, but >> this is coming very late in the JDK 11 development cycle. How important >> is this for 11, given that the naked `timeout` can be used, somewhat, to >> mitigate against long connection timeouts? > > FWIW I think this is a design error (one that I have made in multiple > areas in the past) and should be rectified ASAP. It is becoming > increasingly evident (to me at least) that it is important to > distinguish between connection and request timeouts, especially for > protocols like HTTP where connections may be reused. Connection > mutliplexing and reuse means that the first request has a different > overall disposition than subsequent requests. The semantics of a > connection failure may be substantially different from a request > failure, and as was previously pointed out, different architectures > have different tolerances for the two that may not align easily. > > -- > - DML
