Nightscout Integration regression

[beached]

I cannot see why, but since 0.11.1 the temp basal data sent to nightscout does not show up in the basal display or basal pill. I think it is related to history upload but cannot see the relevant change in the patches.

I have re-tested with 0.11.0 and 0.10.1 and it works there.

[ps2]

Can you see if the temp basal events are making to your Event History in Loop? And then check to see if they're making it to treatments.json in NS?

[beached]

I will try that later. Also, this looks related too, but it is from 0.11.0

They are coming from different data sources at different times

[ps2]

That's a different issue; NS calculates IOB differently than Loop. I haven't looked into the details, but you should probably disable the IOB plugin, and use what Loop is telling you.

[beached]

The first image is the Basal pill, not IOB. But yeah, I rarely look at IOB anyways.

[ps2]

I can confirm this issue. Looking in to it.

[ps2]

Odd, since Riley is running 0.11.0 + #153, and it's working fine.

[beached]

This is 11.0 and working, going to 11.1 the basal pill never( or rarely, not sure if it does eventually) shows the temp basal data on the basal pill side

[dm61]

I am looking at two Loops, both at 0.10.1, The one with a 523 pump has worked fine, including NS integration. The other, with a 723 pump, worked fine for a few days, and has then lost pump history items on NS. I've planned to upgrade both to 0.11.1 today, but will probably wait some more - looks like there may be some issue with pump history upload to NS even on 0.10.1.

[ps2]

I'm seeing temp basal pairs where the duration is first getting passed to NightscoutUploader.upload(). I'm thinking this change is likely the culprit: LoopKit/LoopKit@v0.6.0...v0.7.0#diff-f15dc05fbeb2566c8342d34c44f52529R644 @loudnate, I think I'll make the uploader code not dependent on order, and just look for paired (in time) records.

[loudnate]

@ps2 that change was paired with
2c62c17#diff-e49b73a75143263adce7ce2d8d0f763aL846
On Mon, Sep 5, 2016 at 1:46 PM Pete Schwamb notifications@github.com
wrote:

I'm seeing temp basal pairs where the duration is first getting passed to
NightscoutUploader.upload(). I'm thinking this change is likely the
culprit: loudnate/LoopKit@v0.6.0...v0.7.0
#diff-f15dc05fbeb2566c8342d34c44f52529R644
LoopKit/LoopKit@v0.6.0...v0.7.0#diff-f15dc05fbeb2566c8342d34c44f52529R644
@loudnate https://github.com/loudnate, I think I'll make the uploader
code not dependent on order, and just look for paired (in time) records.

—
You are receiving this because you were mentioned.

Reply to this email directly, view it on GitHub
#161 (comment), or mute
the thread
https://github.com/notifications/unsubscribe-auth/AAM_jAPXdAaAbzxSgYrmEsX2waBuE-0Iks5qnH-VgaJpZM4J1SGX
.

[loudnate]

@ps2 Is the order of the pairs random? Or consistent?

[loudnate]

Short-term, it might be easier to just revert those commits. Long-term, there's no need to do the double-work of temp basal matching.

On Sep 5, 2016, at 1:46 PM, Pete Schwamb notifications@github.com wrote:

I'm seeing temp basal pairs where the duration is first getting passed to NightscoutUploader.upload(). I'm thinking this change is likely the culprit: loudnate/LoopKit@v0.6.0...v0.7.0#diff-f15dc05fbeb2566c8342d34c44f52529R644 @loudnate, I think I'll make the uploader code not dependent on order, and just look for paired (in time) records.

—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub, or mute the thread.

[ps2]

Right now, on 0.11.1, as new temp basals are enacted and fetched from history, when I put a breakpoint on the call to uploader.upload(), I see that the TempBasalDurationPumpEvent records are consistently before the TempBasalPumpEvent records.

Consistent as in 4 stops in the debugger confirmed this. I did have a history mis-fetch while watching, and when all four events were passed to the upload, one temp basal NS treatment made it up (from pairing a rate with the wrong duration record).

Your two changes (one reversing, and the other backing out a reverse) seems like they would cancel each other out, but I suppose it depends on what happens in the fetch sorting with events that have the same timestamp.

[ps2]

The mis-pairing is another reason to update the ns uploading code to be more resilient, and not to assume one follows the other.

[ps2]

Where those commits fixing some other issue?