On Thursday, 20 April 2017 13:49:45 PDT Gregg Reynolds wrote: > regarding API: are we talking about standardizing a kind of OCF API or just > cleaning up the iotivity API? I'm thinking about DOM, a standard API for a > standard data language (HTML). I can imagine some kind of standard OCF > API, with e.g. registerResource etc. which could be mapped to multiple > impkementation languages.
OCF API is out of scope for IoTivity. That needs to be discussed in OCF itself. IoTivity can influence that greatly by refusing to implement an OCF API that is poorly designed, harmful, insecure, etc. If we do that, we need to have an explanation of why we chose that way. If possible, we should also have a suggestion on how to do it better. Moreover, IoTivity can experiment and make proofs of concept, showing that something can be done and how it can be done. That can be reported back to OCF and lead to standardisation of new functionality. Example[*]: UUIDs always contain exactly 128 bits of data. So if we see a property that transmits UUIDs as 296 bits instead of 128, we should refuse to implement wastefully, implement it instead efficiently, and report back on what and why. [*] unfortunately, this is an example of where we did exactly that and OCF insisted that we implement properties in an inefficient manner. -- Thiago Macieira - thiago.macieira (AT) intel.com Software Architect - Intel Open Source Technology Center
