future session start dates when not in session? #209
Description
Related to #191, it's possible for the library to emit incorrect events from the future, notably session start events.
The pull request correctly filters out events created with invalid future dates, but the correct fix would involve testing for whether or not a session is in progress before creating any such invalid records. xdrip does this by testing for 0XFFFFFFFF as a sentinel timestamp meaning that there are no records to produce, which eliminates the need for filtering.
- Some analysis https://github.com/bewest/rag-nightscout-ecosystem-alignment/blob/main/docs/backlogs/future-dated-entries-analysis.md
- Should Nightscout reject future-dated entries? nightscout/cgm-remote-monitor#8453 (comment) - Nightscout issue: Nightscout should not attempt to delete the data as this will cause a retry loop.
- As noted, Trio uses latestReading, it's not clear if updating this eliminates the lastReading problem without any other code changes? In any case, this explains why Loop's behavior is slightly different here than Trio's.
The result is that future SAGE events show up in Nightscout.
(On a personal note, I'm curious if there is any interest in moving towards SPM based architecture, as this would allow me to contribute and test on linux for issues like these.)