On Wednesday 25 March 2015 19:26:16 ??? wrote: > Hi Thiago, Hi Uze
> I?m not sure whether we are communicating correctly, so let me clarify for > your response. > > If you feel there is any discrepancy from my understanding, let me know. > > > So I agree with Uze's proposal: > > You agree with my opinion that any smart connectivity selection logic is > not proper. > > > > 1) Framework has the priority for adaptor and If Application does not > > > specify, then follow the framework policy else follow the specified one. I agree that it may not be enough and I agree that the application may need to specify which adapter to send on. If the application does not make that decision, then the framework will choose a suitable default. The important thing is that the application developer should not have to write anything if the default is fine. > > I don't think we need to do this part: > This is the same idea to your ?priority list? concept, but You don?t > think this is required. > > > > 2) Framework monitors data transmission rate and designate the > > > appropriate > > > adaptor. > > > I don't think we need to monitor transmission and loss rates. That's a job > > for > > the lower below OIC -- the reliable delivery. > > The purpose of monitoring transmission is to decide the appropriate > connectivity in some time frame and not for the reliable delivery. I didn't read these as the same. I don't think we need to monitor transmission rates. The framework should be able to have a priority list without having to monitor anything. We don't need the complexity of monitoring data rates and I don't think that we could write a good one even if we wanted to. -- Thiago Macieira - thiago.macieira (AT) intel.com Software Architect - Intel Open Source Technology Center
