Open TymHuckin opened 1 month ago
Can confirm, same issue here.
Noticed this with G7 as well.
Let’s all observe closely and gather data on our next sensor changes to provide details and logs so we can help get to the bottom of this.
Observed again today.
The sensor details are stored in a file on your phone, probably in the Trio/monitor
directory. Could you please look that up and post it here?
Definitely, will do when kiddo is home and I can look at his phone - most likely this evening.
For some background, SAGE uploads to Nightscout was implemented by merging this branch:
https://github.com/trixing/freeaps/commits/patch/sage/
After some refactoring, SAGE was later fixed for G6 by this PR:
https://github.com/Artificial-Pancreas/iAPS/pull/212
The fix for G7 is in here somewhere:
https://github.com/nightscout/Trio/commit/cf5d5621785d44369d54f69a5936b0646d000bed
Then apparently SAGE was broken by a refactor after the transition to submodules in Trio, where relevant files were removed:
https://github.com/nightscout/Trio/commit/47fa01786a7457b30e2307a11e02908719fa0aa9
Later, SAGE was fixed for Libre, but not Dexcom: https://github.com/nightscout/Trio/commit/15713b0f1d5d97cec8273df9edfa02315225a2b0
I assume we would need a similar fix for G6 and G7 SAGE to be updated in NS.
I suspect @avouspierre is already looking at this, but wanted to include a tag, as he fixed this for Libre :-)
Looks like Trio sees a sensor from ~3 sensors back (yes, we've had multiple fail early so changed before 10d).
That may be when we swapped from using xDrip4iO5. Or maybe completely unrelated - I can't pinpoint exactly when we switched over.
@aug0211 , only G6 sessions are written to this file, xdrip sessions are not.
If you look at my sleuthing above, you can track down the deleted files that broke this:
FreeAPS/Sources/APS/CGM/DexcomSourceG5.swift FreeAPS/Sources/APS/CGM/DexcomSourceG6.swift FreeAPS/Sources/APS/CGM/PluginSource.swift
Some change likely broke this for G7 at the same time, I have not pinpointed that.
The fix for Libre can probably be used as a model to fix this for Dexcom again.
Ooh interesting, good finds. Was so intent on getting the requested data uploaded I missed your findings!
I don't think it's accurate that only G6 gets added to that log though, otherwise my log should be blank (using G7 exclusively for some time).
Either way it seems like the fix is somewhere in this area you've identified. Nice!
I meant exclusively Dexcom sessions (probably including G7), and not xdrip :-)
Can confirm, using G6 btw, that the 'cgm-state.json' file has not been updated for me for the new sensor. Last sensor showing up is from 04-May-2024. I changed it on May 31st.
+1 here -- for G7. An install of Trio on a new phone June 12, no cgm-state.json file was created. I've tried scanning for new sensor, deleted CGM and added it again, SAGE does not update and cgm-state.json doesn't exist.
I've noticed that the sensor is doesn't always update in Nightscout. I change the sensor yesterday (G6/Anubis) and it's still showing as over 13 days old. Had this before but then suddenly it updated. 0.1.0 56a3f87 updated on my phone 4 June 2024.