2020-08-18, 09:15 PM
(2020-08-17, 07:20 PM)seandepagnier Wrote: Is there actually a standard defined for signalk messages to command autopilots that is sensible?
The "first draft" defines multiple redundant messages while ignoring many useful parameters available in pypilot. This would just take more logic and code (maybe more bugs) to implement, and my suggestions which were ignored it seems.
The autopilot stuff in the current Signal K specification was not the result of a great design process and I don't have any particular love for it. The only consideration is backwards compatibility, but if we got something wrong that is irredeemable we can have another go at it.
Never attribute to wilful ignorance something that can be explained by a finite amount of time and energy
So, what would a perfect autopilot protocol/interface/api look like in abstract? I can then see about shoehorning it into the current SK or if not possible figure out a way to move beyond the current spec.