What happen when data connection is lost

[Andreas29]

Hi,

what happens if the data connection for wind and / or gps is lost?

Does pypilot switch to standby? Or to compass mode? 
How to trigger an alarm? Via the watchdog plugin for datastream or others?

Thanks and best regards

Andreas

[seandepagnier]

pypilot will switch to compass in that particular case, and also will remember and attempt to switch back to the lost mode if the data comes back.

You can trigger a pypilot alarm with the opencpn watchdog plugin for this.

[Andreas29]

Hi,

thanks for the fast response.

Does pypilot react to "hanging" data? If the source always send one value without any changes?

Thanks and best regards

Andreas

[seandepagnier]

why would the source keep sending the same value? I think it's a bug in the source, but so far it would just keep using it.

[Andreas29]

Hi,

a guy in a german sailing forum asked this. He said that it could happen that signalk when loosing the signal on the input then send the last value on the outside.

This is not a problem of pypilot of course, it's a problem of the datasource.
It's just nice to know how pypilot react in case of such faults.

Thanks and best regards

Andreas

[ironman]

A bit more context on this: the signalk Derived Data plugin can provide a value based on values from different sources. When one source dries up but the other does not, the resulting derived value is still updated as if were based on valid data. Which it is not anymore. It's a signalk thing but no one seems to worry about it. An example would be True Wind Angle, which is based on Apparent Wind and Boat Speed. When the wind feed freezes, you still get True Wind updates as the boat speed updates come in - no indication of anything wrong. This might be the concern where the guy on the forum referred to.

I don't expect this issue to affect the workings of pypilot, as pypilot does not seem to require signalk-generated derived data and I expect pypilot to recognize when any source goes stale EDIT: based on the signalk timestamps. For myself, I'll burn that bridge when I get there and worry when the sails go flapping ;-)

[Geronius]

Being the 'guy on the german forum' I will add some context

... I think it's a bug in the source, but so far it would just keep using it.

I completely agree with Sean, it's a bug to be fixed in the data source

... When one source dries up but the other does not, the resulting derived value is still updated as if were based on valid data. Which it is not anymore. It's a signalk thing but no one seems to worry about it. An example would be True Wind Angle, which is based on Apparent Wind and Boat Speed. When the wind feed freezes, you still get True Wind updates as the boat speed updates come in - no indication of anything wrong. This might be the concern where the guy on the forum referred to ...

correct, that's exactly the case - I wasn't talking of sk-derived data but I was testing another plugin that feeds stale TWD and TWA back to SK.
SK per default propagates any inbound 0183 sentences via 10110, so pypilot will get misled
Thanks all for the clarification
Michael

[seandepagnier]

well, it should ignore duplicate timestamps for signalk then which isn't implemented yet


pypilot could maybe ignore duplicate data after enough time, but technically your wind angle may not change at all for a while too, but then switching to compass mode in that case may be "ok" however confusing to the user who thinks their sensor failed.


This plugin feeding stale data is really the problem.