These are chat archives for nightscout/beta

30th
Sep 2014
PierreVandevenne
@PierreVandevenne
Sep 30 2014 16:36
@jasoncalabrese - local chart doesn't display on cheap archos tablet - Android 4.2.2 - galaxy note 2 is perfect.
Ross Naylor
@rnpenguin
Sep 30 2014 17:14
Arrr, so it's not showing up at all?
What os version number is tablet?
PierreVandevenne
@PierreVandevenne
Sep 30 2014 17:19
Nope - will test other devices later tonight. Also, but that is only a usability issue, I think an addition page with simply the arrows and the value in a big font could still be useful for monitoring purposes. I will revert the night monitoring device to the old version so I can glance at it from far away.
Ross Naylor
@rnpenguin
Sep 30 2014 17:20
I had issues with chart on 4.4, but fixed those, only tested on 4.3 and up
Probably not an issue for phones, but will probably impact a lot of old cheap uploaders.
Ross Naylor
@rnpenguin
Sep 30 2014 17:32
Okay thanks! Hoping I can fix
PierreVandevenne
@PierreVandevenne
Sep 30 2014 18:21
Just PM me if you need additional testing. That tablet will be running overnight anyway.
Also, some strange behaviour, a bit surprising to see the two points on top of one another on the left and the yellow sequence of points did not appear in the monitoring portal until I hit refresh. Portal is a few days old (I usually download zip files and compile install myself), no fork/clone.
PierreVandevenne
@PierreVandevenne
Sep 30 2014 19:16
and this is the db at at that moment https://dl.dropboxusercontent.com/u/2445259/db.PNG
seems related to the use of the gap sync function.
Jason Calabrese
@jasoncalabrese
Sep 30 2014 19:19
the direction changes, from flat to not computable
maybe we should hold on to the raw dexcom system time, this could be a time conversion issue
PierreVandevenne
@PierreVandevenne
Sep 30 2014 19:34
When the portal wasn't displaying the yellow dots, it was displaying the "-" indicator and the correct glucose value. Dots I marked in yellow and the arrow re-appeared when I manually hit refresh. The previous event, which I didn't notice on the moment, also leads to a somewhat strange db. Can share it if you think it can help.
Ross Naylor
@rnpenguin
Sep 30 2014 21:00
interesting. Don't think it is a time conversion problem. Wierd that it says "NOT COMPUTABLE" and the sgv is 94
PierreVandevenne
@PierreVandevenne
Sep 30 2014 21:19
Actually, the Dexcom itself agrees -
<Glucose InternalTime="2014-10-01 02:15:32" DisplayTime="2014-09-30 20:15:12" Value="83"/>
<Glucose InternalTime="2014-10-01 02:20:32" DisplayTime="2014-09-30 20:20:12" Value="86"/>
<Glucose InternalTime="2014-10-01 02:30:04" DisplayTime="2014-09-30 20:29:45" Value="89"/>
<Glucose InternalTime="2014-10-01 02:30:32" DisplayTime="2014-09-30 20:30:12" Value="94"/>
<Glucose InternalTime="2014-10-01 02:35:32" DisplayTime="2014-09-30 20:35:12" Value="101"/>
<Glucose InternalTime="2014-10-01 02:40:32" DisplayTime="2014-09-30 20:40:12" Value="109"/>
Kevin Lee
@ktind
Sep 30 2014 21:21
maybe I'm reading this wrong.. but that suggests that the system time isn't PST as originally thought?
PierreVandevenne
@PierreVandevenne
Sep 30 2014 21:21
But this is the first time I see this (Dexcom giving value in spacing different than 5 mins) - have the commands sent to the Dex changed a lot? When it works, it's virtually instantaneous now.
I am in Belgium
Kevin Lee
@ktind
Sep 30 2014 21:24
Ross completely re-factored the way we read from the receiver
PierreVandevenne
@PierreVandevenne
Sep 30 2014 21:24
And I say when it works, because it doesn't work all the time after unplugging. With the older uploader you'd have to wait a few minutes, but it never failed reconnecting to that tablet. With this one, once you disconnect (here I did to download the Dex the standard way) the tablet never recovers and never gets the data back from the Dex - I had to reboot it. Doesn't even charge at this point.
Ben West
@bewest
Sep 30 2014 21:25
back/quit doesn't work?
back replug with 10 second delay works
there is a send feedback button
sends log
PierreVandevenne
@PierreVandevenne
Sep 30 2014 21:26
I am going to retry that.
Kevin Lee
@ktind
Sep 30 2014 21:26
there may be a logic error in the way we manage the polling with the disconnect/reconnect
Ben West
@bewest
Sep 30 2014 21:26
I usually have to unplug, back/quit, replug
Kevin Lee
@ktind
Sep 30 2014 21:26
I've squashed at least one of those bugs recently
Ben West
@bewest
Sep 30 2014 21:26
I have one phone with a loose port
where the port on the phone flags up and down
Kevin Lee
@ktind
Sep 30 2014 21:27
I have that issue too on one of my phones
Ben West
@bewest
Sep 30 2014 21:27
can send it to you for stress testing
great for stressing this issue
just wiggle
Kevin Lee
@ktind
Sep 30 2014 21:27
=)
Ben West
@bewest
Sep 30 2014 21:27
happens to many phones, and to dex in later stages of life
"ah, that's the reason this was a cheap used phone"
Ross Naylor
@rnpenguin
Sep 30 2014 21:29
i think i know where the unplug bug is
PierreVandevenne
@PierreVandevenne
Sep 30 2014 21:31
back works - unplug replug did this time - but failed twice before and definitely failed after I hooked the dex to my laptop and reconnected. report sent, at least I sent feedback, no indication that logs were sent.
Kevin Lee
@ktind
Sep 30 2014 21:33
looks like we did get them - Archos archos Archos 90 Neon right?
PierreVandevenne
@PierreVandevenne
Sep 30 2014 21:36
right
PierreVandevenne
@PierreVandevenne
Sep 30 2014 21:48
I am fairly certain it wasn't the cable/connector. When I replugged, no connection, no charging. The tablet is on the ground and wasn't touched or moved. I rebooted it and as soon as the reboot screen appeared, the Dex began charging. Everything was flat, nothing had moved
Ross Naylor
@rnpenguin
Sep 30 2014 21:58
do you have root enabled in the prefs?
Kevin Lee
@ktind
Sep 30 2014 21:58
and are you rooted?
heh nm the signup sheet answers the rooted question
PierreVandevenne
@PierreVandevenne
Sep 30 2014 22:02
that tablet isn't rooted - have a rooted phone at hand if you want a test with it
PierreVandevenne
@PierreVandevenne
Sep 30 2014 22:07
that internal dexcom time stamp that's not space by 5 mins has me a bit worried. Have roughly six months of data and the dex roughly shifts by one second every 100 measures - never seen it report a non time aligned measure
Jason Calabrese
@jasoncalabrese
Sep 30 2014 22:11
it's almost like the dex created an entry at our request
Ross Naylor
@rnpenguin
Sep 30 2014 22:12
we don't write anything explicity to the dex
only the read command
Kevin Lee
@ktind
Sep 30 2014 22:12
I've seen it occasionally in my testing. I always thought it was the result of a calibration
Ross Naylor
@rnpenguin
Sep 30 2014 22:12
recent testing, or testing for the last year?
Kevin Lee
@ktind
Sep 30 2014 22:13
last year
meterbg contains the actual entry but I thought it was the adjusted value
PierreVandevenne
@PierreVandevenne
Sep 30 2014 22:19
Not linked to calibration here - it was earlier <Meter InternalTime="2014-09-30 13:01:33" DisplayTime="2014-09-30 07:01:13" Value="134"/>
<Meter InternalTime="2014-10-01 01:20:27" DisplayTime="2014-09-30 19:20:08" Value="95"/>
<Glucose InternalTime="2014-10-01 01:15:32" DisplayTime="2014-09-30 19:15:12" Value="91"/>
<Glucose InternalTime="2014-10-01 01:20:32" DisplayTime="2014-09-30 19:20:12" Value="92"/>
<Glucose InternalTime="2014-10-01 01:25:32" DisplayTime="2014-09-30 19:25:12" Value="96"/>
Jason Calabrese
@jasoncalabrese
Sep 30 2014 22:28
I wonder if the read prevented the dex from writing, looks like the 2014-10-01 02:25 reading was skipped, maybe just logged late?
locking?
Kevin Lee
@ktind
Sep 30 2014 22:30
maybe
in BGScout I always had a ~3 second delay after the 5 minute mark
I thought we had something like that in this code
Doesn't appear to be the case, FIVE_MINUTES_MS is set to 5 minutes exactly
Ross Naylor
@rnpenguin
Sep 30 2014 22:34
we put 5 minutes and 3 secs
that thought makes sense
Jason Calabrese
@jasoncalabrese
Sep 30 2014 22:34
I think @PierreVandevenne said this was in response to to clicking gap sync
Kevin Lee
@ktind
Sep 30 2014 22:35
when I had exactly 5 minutes I would frequently see the reading before the dex even went off.. Ben mentioned noticing the same thing
PierreVandevenne
@PierreVandevenne
Sep 30 2014 22:36
It seemed so, but at the moment I wasn't really paying attention to the exact second.
Kevin Lee
@ktind
Sep 30 2014 22:37
The gap sync does the same thing a standard sync does - only it grabs 20 pages instead of 1
PierreVandevenne
@PierreVandevenne
Sep 30 2014 22:37
maybe lasts longer?
Kevin Lee
@ktind
Sep 30 2014 22:37
yeah - it takes longer to pull in all of those records
if it happened right on the mark when the sensor fired
PierreVandevenne
@PierreVandevenne
Sep 30 2014 23:01
Probably not the fault of this version. Couldn't find anything visually so I wrote a small program to double check
Starting Dexcom Import in mgdl
Dexcom Import Complete: Imported Data Points: 40650 --- Imported Calibration Points: 435
Start 2014-03-30 15:45:27
index: 33410
2014-09-01 15:47:01
2014-09-01 15:53:08
2014-09-01 15:57:01
2014-09-01 16:02:01
index: 40382
2014-09-29 22:30:31
2014-09-29 22:35:32
2014-09-29 22:40:15
2014-09-29 22:45:15
index: 40620
2014-09-30 20:20:12
2014-09-30 20:29:45
2014-09-30 20:30:12
2014-09-30 20:35:12
3 events - all of them in September