These are chat archives for nightscout/intend-to-bolus

23rd
Sep 2016
David
@davidkeddydb
Sep 23 2016 00:06
@scottleibrand I am going to leave CGM, pump as devices and wipe everything else out and just go from Oref0 set up and see what happens. Should I leave tz or delete it? It sounds like Oref0 won't need it
Scott Leibrand
@scottleibrand
Sep 23 2016 00:07
I would just do a new loop directory. No need to delete that way.
You will want to clean up crontab first though.
@mccgm looks like you are getting undefined COB for some reason.
I would run and check inputs and outputs of oref0-meal
Eric Jensen
@elnjensen
Sep 23 2016 00:22
@jasoncalabrese I read what @elnjensen wrote as meaning that the UTC times in his iob.json were off by 4 hours (EDT times with a Z appended). @elnjensen can you clarify?
Yes, that's exactly right. And because there is already a timezone indicator in there, a simple tz rezone doesn't work - it just replaces the 'Z' with '+00:00'.
@scottleibrand @jasoncalabrese To answer your earlier questions about timezones in my setup, my rig is set to EDT (or more precisely, America/New_York which will automatically toggle EDT/EST as needed).
Zoning my pumphistory.json works fine (and indeed was needed to get it to work with NS).
Eric Jensen
@elnjensen
Sep 23 2016 00:28
The issue here is that the timestamps in iob.json are getting tagged as UTC, rather than as the correct timezone, or being left unzoned. If the latter were the case, then I could zone them with the tz device.
But I do think we have an explanation of why changing from oref0 master to dev caused the change in behavior: in master, there are no timestamps in that report, and just a single entry, so NS must just assume it's current at upload. In dev, there are timestamps, so NS uses its default (and I would say consistent) behavior of treating them as UTC.
Well, I guess it should treat them as UTC, since they are tagged that way!
Scott Leibrand
@scottleibrand
Sep 23 2016 01:00
yeah, we need to figure out why those are incorrectly getting Z
Jason Calabrese
@jasoncalabrese
Sep 23 2016 01:15
maybe some env var? how was the time zone set?
Eric Jensen
@elnjensen
Sep 23 2016 01:16
Tried looking through the oref0 code a bit, and I see a bunch of instances of var tz = require('timezone'); but I'm not sure what that implies about where it is getting the info.
pi@raspberrypi:~/openaps_sor $ date
Thu Sep 22 21:19:24 EDT 2016
pi@raspberrypi:~/openaps_sor $ cat /etc/timezone 
America/New_York
Eric Jensen
@elnjensen
Sep 23 2016 01:22
In the meantime, unfortunately I seem to have killed my NS website - possibly uploaded some poorly-formatted data while I was trying to fix this. Now I get a black screen accessing the site, and errors trying to upload...
Jason Calabrese
@jasoncalabrese
Sep 23 2016 01:23
Is it an Edison or pi
Eric Jensen
@elnjensen
Sep 23 2016 01:23
It's an RPi 3.
Jason Calabrese
@jasoncalabrese
Sep 23 2016 01:24
Of try using raspi-config again to set the timezone
Also wondering if you get different results when running with a cron and when running interactively
Eric Jensen
@elnjensen
Sep 23 2016 01:27
Pretty sure it's not a problem with the Pi's timezone - was working fine until I updated oref0. Could rezone pumphistory, times all are fine, etc.
Unless it doesn't like America/New_York as a timezone (though that's valid, and in fact more standard than something like EDT).
Unfortunately I can't test anything until I get my NS site working again. Any troubleshooting suggestions? App restart didn't do it. Possibly need to clear out some bad data...
Jason Calabrese
@jasoncalabrese
Sep 23 2016 01:29
The js code seems to think your in utc
If you can get to the NS admin you could delete any future data
Eric Jensen
@elnjensen
Sep 23 2016 01:32
Where is "NS admin"?
Scott Leibrand
@scottleibrand
Sep 23 2016 01:42
@elnjensen you should be able to troubleshoot locally on the pi. objective is just to get that iob.json to report the correct present and future times, right?
Eric Jensen
@elnjensen
Sep 23 2016 01:44
Yes, ultimately - but if I can't get anything into Nightscout I can't test it. Nightscout site is currently just a black screen.
Scott Leibrand
@scottleibrand
Sep 23 2016 01:45
you should see that same data in the file locally on the Pi
or are you saying you can't get BGs without NS?
can you plug the CGM in?
Eric Jensen
@elnjensen
Sep 23 2016 01:47
No, sorry I wasn't clear. Just saying that the ultimate goal of fixing that file is to have it appear correctly in NS, and at the moment I wasn't able to get the IOB data into NS. But that appears to have righted itself, so that's all good.
How does the js code for IOB calculation determine what timezone it thinks it's in?
Scott Leibrand
@scottleibrand
Sep 23 2016 01:49
I wonder if it's using the date field off one of your inputs
    t = new Date(clock.getTime() + i*60000);
var clock = new Date(tz(inputs.clock));
jinx!
Looks like clock, yes.
Scott Leibrand
@scottleibrand
Sep 23 2016 01:52
so what's your clock.json have?
Eric Jensen
@elnjensen
Sep 23 2016 01:53
"2016-09-22T21:50:24"
Scott Leibrand
@scottleibrand
Sep 23 2016 01:53
try passing a clock-zoned.json to calculate-iob instead?
Eric Jensen
@elnjensen
Sep 23 2016 01:53
No TZ there, and correct local time.
Scott Leibrand
@scottleibrand
Sep 23 2016 01:53
surprised my code doesn't do that already
Eric Jensen
@elnjensen
Sep 23 2016 01:53
What's the right call to tz rezone if it's not JSON?
Scott Leibrand
@scottleibrand
Sep 23 2016 01:53
or maybe it does. do you have a clock-zoned.json already?
Eric Jensen
@elnjensen
Sep 23 2016 01:53
Not using your code... :smile:
...yet
Well, in fact it's all your (and others') code.
Scott Leibrand
@scottleibrand
Sep 23 2016 01:54
oh, I thought you had this problem after running oref0-setup.sh?
cassknox
@cassknox
Sep 23 2016 01:54
Has anyone ever had an issue setting up where when you run: openAPS use pump model, it takes a long time and then only returns " " ? I'm not sure why it doesn't return "522" as that is the model of the Medtronic pump I'm using. Any ideas?
Scott Leibrand
@scottleibrand
Sep 23 2016 01:54
or was the problem just with the old loop after it upgraded oref0 to dev?
Eric Jensen
@elnjensen
Sep 23 2016 01:54
Exactly.
Scott Leibrand
@scottleibrand
Sep 23 2016 01:55
@cassknox that just means it couldn't get an answer from the pump
Eric Jensen
@elnjensen
Sep 23 2016 01:55
Been fixated on getting the old loop back to working status before continuing with understanding new loop.
Scott Leibrand
@scottleibrand
Sep 23 2016 01:55
@cassknox I would try openaps use pump mmtune if you're using mmeowlink. if you're on Carelink, maybe try reseating the stick or moving it closer to the pump?
@elnjensen ok that makes more sense
openaps report add monitor/iob.json text iob shell monitor/pumphistory-zoned.json settings/profile.json monitor/clock-zoned.json
openaps report add monitor/clock-zoned.json JSON tz clock monitor/clock.json
and add an invoke of clock-zoned.json right after invoking clock.json, something like: openaps alias add monitor-pump "report invoke monitor/clock.json monitor/temp_basal.json monitor/pumphistory.json monitor/pumphistory-zoned.json monitor/clock-zoned.json monitor/iob.json monitor/meal.json monitor/reservoir.json monitor/battery.json monitor/status.json"
guess we need to update oref0 dev branch to complain if clock input is unzoned or something
Eric Jensen
@elnjensen
Sep 23 2016 01:58
Ah, thanks for the syntax. Was trying to figure out tz rezone instead.
OK, will update reports to get clock-zoned in there then test - hopefully back soon with results.
cassknox
@cassknox
Sep 23 2016 02:03
Thanks @scottleibrand I unplugged the carelink and replugged it into the raspberry PI and moved closer to the rig and it worked! :)
Eric Jensen
@elnjensen
Sep 23 2016 02:11
@scottleibrand No dice:
pi@raspberrypi:~/openaps_sor $ openaps monitor-pump
pump://JSON/read_clock/monitor/clock.json
reporting monitor/clock.json
tz://JSON/clock/monitor/clock-zoned.json
reporting monitor/clock-zoned.json
pump://JSON/read_temp_basal/monitor/temp_basal.json
reporting monitor/temp_basal.json
pump://JSON/iter_pump_hours/monitor/pumphistory.json
reporting monitor/pumphistory.json
tz://JSON/rezone/monitor/pumphistory-zoned.json
reporting monitor/pumphistory-zoned.json
calculate-iob://JSON/shell/monitor/iob.json
reporting monitor/iob.json
pi@raspberrypi:~/openaps_sor $ head monitor/iob.json 
[
  {
    "netbasalinsulin": 1.35, 
    "activity": 0.018, 
    "basaliob": 0.852, 
    "time": "2016-09-23T02:09:46.000Z", 
    "hightempinsulin": 1.4, 
    "bolussnooze": 0.569, 
    "iob": 1.81
  }, 
pi@raspberrypi:~/openaps_sor $ openaps report show monitor/iob.json --cli
openaps use calculate-iob shell monitor/pumphistory-zoned.json settings/profile.json monitor/clock-zoned.json
pi@raspberrypi:~/openaps_sor $ cat monitor/clock-zoned.json 
"2016-09-22T22:10:13-04:00"
Scott Leibrand
@scottleibrand
Sep 23 2016 02:14
@elnjensen just pushed a commit to the oref0-setup branch to check if clock input is zoned and output a warning if not.
Eric Jensen
@elnjensen
Sep 23 2016 02:14
Cool, thanks.
And I spoke too soon about the above - NS IOB looks correct now. WTF?
Oh, never mind.
I was fixated on the "Z" still being there, but didn't notice that it had also shifted the time by four hours.
Eric Jensen
@elnjensen
Sep 23 2016 02:15
Not what I expected since the other reports tend to leave time alone and append the local TZ signifier.
Scott Leibrand
@scottleibrand
Sep 23 2016 02:16
yeah, that's the way javascript does times internally
if you still have a loop using clock.json, want to pull that branch and see if the warning shows up properly in your logs?
cd ~/src/oref0 && git checkout oref0-setup && git pull && npm run global-install should install that branch
Eric Jensen
@elnjensen
Sep 23 2016 02:24
Just made a small comment on the regex in the changes, could go either way on that. Looks like a good addition. I changed my main loop to clock-zoned.json, but tomorrow will pull latest oref0 and test with just clock.json as you suggest (need to sleep now!) .
Thanks as always for your generous help.
With this working, now looking forward to testing out the new loop setup - and running it! Need to read up on AMA a bit more, but looks really useful, looking forward to using it.
Scott Leibrand
@scottleibrand
Sep 23 2016 02:25
heh, CDO alleviation fix added ;-)
(CDO is like OCD, but alphabetized like it should be) :-p
Eric Jensen
@elnjensen
Sep 23 2016 02:28
Thanks also @jasoncalabrese for help in getting this sorted out.
@rkresha If you're still having the mismatch in IOB reported as you said last night, you can read through the long recent thread here, but the upshot is that you need to make sure your clock.json report gets a timezone added, and then pass that clock-zoned.json to the report that calculates IOB.
jmoutria
@jmoutria
Sep 23 2016 02:51
Hello! I have been reading the openAPS info and have almost completed my collection of hardware to begin working on the AP for my daughter. Currently we are using the Dexcom G5 with an iPhones but a lot of attention is given to NightScout. I'm wondering if I am at a disadvantage by using openAPS with Dexcom verses NightScout. Insight appreciated. I've already set up the RPi3 and I'm just waiting to get the CareLink so I'm sure I'll be taping into this forum a bunch in the near future. Thanks for all the hard work...my wife's sleep will appreciate the fruit of your labor!
David
@davidkeddydb
Sep 23 2016 02:55
@scottleibrand @elnjensen @jasoncalabrese Took some doing, but when through oref0 set up, made some changes and no more time zone issues and running again open loop. Have to get my cron in order then hooking up. Thanks to all of you and other for all the help. If I can the cron figured I will start looping tomorrow morning.
Eric Jensen
@elnjensen
Sep 23 2016 02:59
@davidkeddydb Great! Glad you got it working. :thumbsup:
@jmoutria G5 has been working well with OpenAPS for me. You'll probably still want to send the CGM data to Nightscout, too (pretty easy with iPhone -> Share -> NS), but I like being able to pull it straight from the receiver. NS is useful for visualization.
David
@davidkeddydb
Sep 23 2016 03:04
Quick nightscout question on uploading status to it. The status shows when uploading, but then say 10 secs later it blanks on when last uploaded:
Screenshot 2016-09-22 23.02.57.png
Screenshot 2016-09-22 23.02.57.png
Screenshot 2016-09-22 23.03.06.png
any idea why the openaps and pump information doesn't stay up?
I can hover and see it, but would be great if it was just there
Matthias
@ceben80
Sep 23 2016 04:26
@scottleibrand reducing the count to 288 in ns-glucose.ini....no issues so far with Nightscout in terms of 24 h limit :+1:
....and my loop was running great last night :star2: :tada:
 [device "ns-glucose"]
fields = 
cmd = bash -c "curl -m 30 -s $NIGHTSCOUT_HOST/api/v1/entries/sgv.json?count=288 | json -e \"this.glucose = this.sgv\""
args =
Scott Leibrand
@scottleibrand
Sep 23 2016 04:28
:+1: want to PR in that change?
Jason Calabrese
@jasoncalabrese
Sep 23 2016 04:34
@davidkeddydb you might not have DEVICESTATUS_ADVANCED=true set
Matthias
@ceben80
Sep 23 2016 04:39
@scottleibrand damn....unexpected work :smile: :smile: :smile: .....certainly, I will do that at the weekend
Scott Leibrand
@scottleibrand
Sep 23 2016 04:39
;-)
Ali Mazaheri
@amazaheri
Sep 23 2016 05:02
Hi everyone I do have some extra stuff in case anyone need one let me know. 1 Edison donated... 1 left. 1 505 g4 receiver, 1 g4 share receiver 2 Care link sticks
Hristina Dimova
@xpucuto
Sep 23 2016 05:09
@amazaheri I am interested in the share receiver . I found one and it will arrive soon but I would like to have a spare one for just in case . I am willing to pay for it just and for the shipping . Outside US they are hard to find .
David
@davidkeddydb
Sep 23 2016 05:09
@jasoncalabrese I will check that right now. I am looping!! So cool seeing it be spot on with the basal adjustments. Neat seeing it when it is actually connected.
David
@davidkeddydb
Sep 23 2016 05:15
@jasoncalabrese When I put it in originally like a month ago, I put advance not advanced. Should I be seeing graphical representation of insulin delivery? I know some people have had screen shots showing that. Did I miss something?
Screenshot 2016-09-23 01.16.13.png
Jason Calabrese
@jasoncalabrese
Sep 23 2016 05:21
the X means Not Enacted, whats the tooltip show?
might be some errors in your logs too
also make sure you have basal rendering enable in the settings drawer, there's an env var to make it stay on by default
Tim Street
@tim2000s
Sep 23 2016 07:56
Is AMA documented somewhere? I was googling but didn't find much.
Colin Lennon
@colinlennon
Sep 23 2016 08:20
@tim2000s are you on dev or master?
Tim Street
@tim2000s
Sep 23 2016 08:52
Dev
Tim Street
@tim2000s
Sep 23 2016 08:59
Had an "interesting" experience last night with the Medtronic 640G uploaded. It started asking for calibration at 5 am so I hit snooze. Throughout the two hours it was snoozing, so was I and I didn't see the glucose readings that it was producing at 42.8mmol/l till I reviewed this morning due to an unexpected hypo.
oref0 had looked at these numbers and as you'd expect treated them. So I wondered how I might implement something that double checks glucose values and doesn't act if there is something clearly erroneous, such as this sort of random high or a compression low? What do you think? Or is this already in place.
Colin Lennon
@colinlennon
Sep 23 2016 09:10
@tim2000s If you're on dev, then it should just be a case of following these instructions - https://openaps.readthedocs.io/en/dev/docs/walkthrough/phase-6/advanced-features.html
I know that refers to meal-assist, but I think the setup for AMA is exactly the same and so if you're on dev, you'll have AMA as it was merged to dev a few months back
Maybe @danamlewis or @scottleibrand could confirm if that's correct?
Tim Street
@tim2000s
Sep 23 2016 09:15
I've got it set up @colinlennon , and checking the logs can see what it seems to be doing. I just wondered if this was documented!
Colin Lennon
@colinlennon
Sep 23 2016 09:25
Have a read through this post if you haven't seen it already - https://diyps.org/2016/07/11/picture-this-how-to-do-eating-soon-mode/
I initially made the mistake of setting up AMA but not using it as intended - which, is by "correcting to 80" up to 1 hour before a meal. From my experience, that part is key
Tim Street
@tim2000s
Sep 23 2016 10:10
Ah, I hadn't read that. It makes sense. Is there a way to tell openaps that's what you're doing?
Or does it just detect it?
Matthias
@ceben80
Sep 23 2016 10:23
@colinlennon thanks for the link :+1:
Colin Lennon
@colinlennon
Sep 23 2016 10:29
I think for the pre-bolus, you can just manually bolus from the pump 30 minutes to 1 hour before the meal, then the AMA stuff will kick in when you enter the carbs details at mealtime via the pump bolus wizard. I set a temporary target for 1 hour via my Pebble watch rather than doing the pre-bolus, which works well, but the manual pre-bolus is probably more effective. I'm just lazy and like the ease of doing it on my watch :smile:
Entering carbs accurately is vital too, AMA needs to know how carbs are in the meal
Tim Street
@tim2000s
Sep 23 2016 10:31
Right. So following the blog, you Pre-bolus, then you do your normal boils wizard entry. Do you configure bolus to 0 when you submit or do the bolus anyway?
Boils=bolus
Colin Lennon
@colinlennon
Sep 23 2016 10:37
You still bolus, but about 2/3rds of what your bolus would be otherwise. AMA usually takes care of the rest, but after an hour or so, you can take the remainder, or some of the remainder, if your BG has risen more than expected
I've tweaked my carb ratios now based on the assumption that I'm using AMA for all meals, so that the bolus wizard recommendations are around 2/3rds of what they'd usually be
I'm fairly new to AMA and still fine-tuning my carb ratios etc. to suit it, hopefully others with some more experience of it will chime in with some advice too!
Andrew Warrington
@andrew-warrington
Sep 23 2016 10:54
Hi all, potential meet up opportunity. We will be at the Luxembourg Science Centre on the morning of Saturday, October 8th, to see the "Beyond the Lab" exposition which features DIY diabetes input from @timomer. Anyone interested in meeting up, reply here, tagging me. If there is enough interest I will organise it more formally.
jmoutria
@jmoutria
Sep 23 2016 11:00
@amazaheri I am interested in the Care Link sticks. Can we work out a deal? Thanks!
Tim Street
@tim2000s
Sep 23 2016 11:03
Thanks @colinlennon - That explains it.
Tim Street
@tim2000s
Sep 23 2016 11:59
Giving it a try!
John Sjolund
@sjolundjohn
Sep 23 2016 12:44
@danamlewis I am using a Non-US pump. Model 522. V2.2A 1.1 OB OB
Tim Street
@tim2000s
Sep 23 2016 13:19
Okay @colinlennon thanks for your help this morning. Loving AMA.
Colin Lennon
@colinlennon
Sep 23 2016 13:20
@tim2000s Glad to hear it! :smile:
David
@davidkeddydb
Sep 23 2016 13:26
Ok I was running and then my cron got this error: raspberrypi size (0) is less than 64 and not 15, which may cause an error.
David
@davidkeddydb
Sep 23 2016 13:33
This would be because I thought I had set everything to work of ns without being connected. Apparently I had not. Plugged in Dex and back. Do most people have some language in their cron to reset the usb port for the carelink stick so often or based on failure to communicate?
Tim Street
@tim2000s
Sep 23 2016 13:41
Yes, I do a USB reset before each loop cycle.
oref0-usb-reset
Or maybe the other way around.
oref0-reset-usb
I have to run it as sudo
David
@davidkeddydb
Sep 23 2016 13:44
ok, so do you run two crons or do you run it with usb-reset and then your loop on the same cron line? Sorry, complete linux newbie
David
@davidkeddydb
Sep 23 2016 13:51
@tim2000s Never mind. Think I answered my own question. You likely want to do the usb reset as a separate cron in order to give it time to cycle through I would think. Unless you can run the reset on the same line as the cron for the loop and it will wait until that process finishes before it goes to run the loop.
Dana Lewis
@danamlewis
Sep 23 2016 14:38
@tim2000s "eating soon mode" you can do separate from AMA, but AMA or not makes a big difference in meal spikes. And using temp targets in NS is an easy way for the loop to know that's what you're doing so it doesn't counter correct
Also re compression low weirdness, yes
Tim Street
@tim2000s
Sep 23 2016 14:40
@davidkeddydb I run it as the first thing in my loop alias.
Matthias
@ceben80
Sep 23 2016 15:36
@scottleibrand this PR has been created :wink:
openaps/oref0#223
Scott Leibrand
@scottleibrand
Sep 23 2016 15:49
@tim2000s the Dexcom is smart enough not to do that. It maxes out at 400mg/dL (vs. your 770) and will go to ??? instead of displaying a number if sensor readings change that abruptly.
So we haven't had to add any of our own logic to catch those kinds of things. Sounds like you'll need to though.
Tim Street
@tim2000s
Sep 23 2016 16:07
Yes. It wasn't something I was expecting to see, to be honest. I've a plan though @scottleibrand so it's not a huge issue!
Eric Jensen
@elnjensen
Sep 23 2016 17:01
Thanks @tim2000s @colinlennon for the discussion of AMA. Planning to implement it this weekend, so would love to hear your or any others' comments on tips and tricks, pointers to other docs, etc. I think I have the resources to get started but there's no substitute for real experience shared by other users.
Eric Jensen
@elnjensen
Sep 23 2016 17:13
Related to temp targets in Nightscout, are there any docs on that? Not sure if I may need to update the code base of my Nightscout installation. I'm on 0.9.0 b3 but think I may need to update to latest dev to get some of these features. (And I assume these instructions are what I need to do to redeploy that new code base to my site - already synced my fork with latest changes.)
Lynne Butts
@flutelynne_twitter
Sep 23 2016 17:24
This message was deleted
Scott Leibrand
@scottleibrand
Sep 23 2016 17:29
@elnjensen version 0.9.0-beta3 (head 7fa33b8) is what we're using at the moment, so that should be fine
re: the earlier discussion about adjusting your carb ratio to bolus for fewer carbs up front when using AMA: there is an advanced setting in preferences.json to support that: carbratio_adjustmentratio. If you set that, it will cause oref0 to use a different carb ratio than your pump bolus wizard does, so you can set it up to have the pump bolus wizard bolus for 2/3 or 3/4 of the carbs, but make sure oref0 knows it needs to take care of the rest at some point. @jasoncalabrese uses that, so can probably explain how it works best for him. (we don't use that, because @danamlewis uses the easy bolus buttons, not the bolus wizard)
if you're just starting out on AMA, though, I would stick with your current settings, do eating soon (either manual bolus or temp target) when you can, and see how that works first
Dana Lewis
@danamlewis
Sep 23 2016 17:34
and @elnjensen you are running into a lot of the good stuff that needs documentation - AMA, temp targets, etc. so would be awesome if you could copy and paste - even in super rough form - anything you get/find in gitter and add it into the docs, so the next person has something to start from!
Scott Leibrand
@scottleibrand
Sep 23 2016 17:34
specifically to the wip/ama-docs branch of openaps/docs (which is going to get merged to master when the dev branch of oref0 does)
John Sjolund
@sjolundjohn
Sep 23 2016 21:11

Seems after the new setup, I am getting this error quite a bit when looking at logs:

Listening: ...No comms detected
Old pumphistory: 
Error, retrying

After this loop doesn't recover

Scott Leibrand
@scottleibrand
Sep 23 2016 21:13
to troubleshoot, you'll need to pause the loop and run some commands manually. I usually do killall -g openaps; openaps use pump model and killall -g openaps; openaps use pump mmtune (if you're on mmeowlink)
you can also comment out the pump-loop line in the crontab while you're troubleshooting
John Sjolund
@sjolundjohn
Sep 23 2016 21:24

everything else works fine, it just seems to hang occasionally.

I have had the updated loop running for +/- 12h max

John Benjamin
@Bender1061
Sep 23 2016 21:31
@sjolundjohn I get that a lot too, seems to be if there not the best signal between pump and rig. Like If I have my cell phone between the two devices. or I'm standing right beside a large transformer.
Brett
@burgertime14
Sep 23 2016 21:43
I used the set up script and forgot to add a new cron job for oref0-fix-git, got the same error as @sjolundjohn until I wised up and figured out that I had some corrupt files, running really well since, getting a lot more completed loops
R. Dennis Steed
@rdsteed
Sep 23 2016 21:46
@scottleibrand @tim2000s Regarding that 42.8 mmol/l reading. Not sure about the rest of the world, but in the US, Enlite sensors read from 40 to 400 mg/dl (2.2 to 22.2 mmol/l). Could it be that there was an inadvertent units problem?
Tim Street
@tim2000s
Sep 23 2016 21:48
@rdsteed it's not a real reading. It's something that the 640G throws out when it is waiting to be calibrated and has stopped providing any data.
Scott Leibrand
@scottleibrand
Sep 23 2016 21:50
ah. so dexcom uses values < 30 for that
you might want to update the oref0 code to treat anything > 400 as an error the same way it does anything < 30
Tim Street
@tim2000s
Sep 23 2016 21:51
Where's that @scottleibrand? Happy to do so!
I think all the others are in the various aliases, which you're bypassing
Tim Street
@tim2000s
Sep 23 2016 22:09
Grepping for that, I don't seem to have it on my machine
Scott Leibrand
@scottleibrand
Sep 23 2016 22:10
grepping for what? don't have which?
Tim Street
@tim2000s
Sep 23 2016 22:11
the determine-basal.js#L69
I thought I'd pulled it as part of the git fetch I did.
Scott Leibrand
@scottleibrand
Sep 23 2016 22:19
what branch of oref0 are you on?
Tim Street
@tim2000s
Sep 23 2016 22:19
dev
Scott Leibrand
@scottleibrand
Sep 23 2016 22:20
and what does line 69 of your determine-basal.js look like?
Tim Street
@tim2000s
Sep 23 2016 22:29
On github it's the same as that. On my machine, I can't find it.
Scott Leibrand
@scottleibrand
Sep 23 2016 22:39
should be in ~/src/oref0/lib/
Tim Street
@tim2000s
Sep 23 2016 22:48
@scottleibrand Got them both.
Scott Leibrand
@scottleibrand
Sep 23 2016 22:49
cool. you should be able to commit those to your fork and submit a PR back to oref0 dev
Tim Street
@tim2000s
Sep 23 2016 23:49
I'm not sure it's worth committing it back to the main branch. The use of the 640 and CNL for feeding data to oref0 feels really flakey
The connection breaks a lot (as in it'll simply stop functioning for 15 mins at a time)