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

12th
Dec 2017
Scott Leibrand
@scottleibrand
Dec 12 2017 00:10
@hartvig what exactly do you mean by "connect online"? are you talking about uploading CGM data? pump history? rig decisions?
Jacob H
@jdhigh
Dec 12 2017 00:36
Good evening! Quick question: If I delete meal.json, will it automatically regenerate?
Toby Canning
@TC2013
Dec 12 2017 00:38
Hi all, I haven't been keeping up with OpenAPS, but I heard Intel Edison may in short supply?
Jacob H
@jdhigh
Dec 12 2017 00:38
Intel Edisons are no longer manufactured, but you can still usually find one, even if you have to resort to eBay, etc.
Toby Canning
@TC2013
Dec 12 2017 00:39
Thanks @jdhigh
Dana Lewis
@danamlewis
Dec 12 2017 01:42
Also local computer stores, microcenters etc often have one or two in stock @TC2013
@jdhigh yes, but why are you deleting it?
Toby Canning
@TC2013
Dec 12 2017 02:38
Thanks @danamlewis. I have like 6 of them so I was deciding if I should be giving some away.
Dana Lewis
@danamlewis
Dec 12 2017 02:45
:+1: if they're not being used that would probably be appreciated
jgels44
@jgels44
Dec 12 2017 02:54
Has a suitable alternative to the edison been found yet or are we better off just trying to find an edison?
Scott Leibrand
@scottleibrand
Dec 12 2017 02:55
there will be an Explorer HAT for the Pi, but it won't be available for a couple more months, so still best to try to find an Edison for your first rig
jgels44
@jgels44
Dec 12 2017 02:56
Okay thanks
Jacob H
@jdhigh
Dec 12 2017 06:22
My rig isn't looping. I know it is something to do with the time setting on my rig. How do I force a rig update to get the latest time sync?
Jacob H
@jdhigh
Dec 12 2017 06:30
'''Checking deliverAt: 2017-12-12T06:23:50.148Z is within 1m of current time: Tue Dec 12 01:23:50 EST 2017
and that smb-suggested.json is less than 1m old
enact/smb-suggested.json: {"temp":"absolute","deliverAt":"2017-12-12T06:23:50.148Z","reason":"Warning: currenttemp rate 0 != lastTemp rate 10.8 from pumphistory; setting neutral temp of 2.45.","duration":30,"rate":2.45}
'''
Jacob H
@jdhigh
Dec 12 2017 06:39
I'm also seeing this problem in my logs: CGM data retrieval from pump disrupted, retrying in 5 seconds...
Reattempting to retrieve MDT CGM data
monitor/cgm-mm-glucosedirty.json raised Page size too short
Jacob H
@jdhigh
Dec 12 2017 08:20
While upgrading to Dev, during install of openaps setup, I'm getting a hard stop at this point:
The following NEW packages will be installed:
libjq1 libonig2
The following packages will be upgraded:
bash-completion jq
2 upgraded, 2 newly installed, 0 to remove and 67 not upgraded.
Need to get 514 kB of archives.
After this operation, 582 kB of additional disk space will be used.
WARNING: The following packages cannot be authenticated!
bash-completion libonig2 libjq1 jq
E: There are problems and -y was used without --force-yes
Couldn't install bc etc.
Paul Dickens
@thebookins
Dec 12 2017 10:45
Hi folks,
Looking for some help with kid staying high once he breaches the 10 - 11 mmol/L mark. The loop settings are doing a very good job when he is close to in range, but it seems there is added resistance at a higher bg. When I asked once before about this, the recommendation was to set "adv_target_adjustments" to true in preferences, but I understand that this has been deprecated as of 0.6.x.
Is there anything to try that would give the algorithm more leeway to dose at a higher bg level?
Enda Farrell
@endafarrell
Dec 12 2017 13:31
@thebookins can you share a picture your OpenAPS pill?
Ebgineer
@Ebgineer
Dec 12 2017 14:44
@thebookins I haven't used it yet myself, but it sounds like what resistance_lowers_target is for. http://openaps.readthedocs.io/en/master/docs/While%20You%20Wait%20For%20Gear/preferences-and-safety-settings.html#resistance-lowers-target
Brian Cadre
@DrCadre_twitter
Dec 12 2017 15:35

Hello, all. Last night, we successfully looped for the first time! Treatments were being suggested, temp basals set, etc. However, in NS, all the data appeared to be in a vertical line just off screen where I couldn’t scroll to it and it never displayed a BG value at the top LHS of the screen. It did display IOB, COB. I also noticed while watching the logs that I was getting a repeated date error (see below). I have already deleted all the documents and there never were any future treatments available to delete. The date and time on the pump did eventually get reset by the system. Any advice on how to fix would be much appreciated. Below an excerpt from the loop log:

Starting oref0-pump-loop at Tue Dec 12 04:33:02 CST 2017 with 15 second wait_for_silence:
MDT CGM configured; not waiting
Radio ok. Listening: .No interfering pump comms detected from other rigs (this is a good thing!)
Preflight OK.
Attempting to retrieve MDT CGM data from pump
MDT CGM data retrieved
No new MDT CGM data to reformat

Profile less than 60m old; Profile valid. Refreshed pumphistoryWarning: clock input Invalid Date is unzoned; please pass clock-zoned.json instead
and meal.json
Checking pump clock: "2017-12-12T04:33:54-06:00" is within 1m of current time: Tue Dec 12 04:33:56 CST 2017
Temp refreshed

No bolus needed (yet). jq: error: null and number cannot be divided
date: invalid date ‘@’
Settings less than 15 minutes old. jq: error: null and number cannot be divided
date: invalid date ‘@’
Edison battery at 67% is charged (>= 98%) or likely charging (60-70%). pumphistory-24h refreshed
jq: error: null and number cannot be divided
date: invalid date ‘@’
Your instance of oref0 [0.6.0, master] is up-to-date.
Completed oref0-pump-loop at Tue Dec 12 04:34:30 CST 2017

Starting oref0-pump-loop at Tue Dec 12 04:34:32 CST 2017 with 6 second wait_for_silence:
MDT CGM configured; not waiting
Radio ok. Listening: .No interfering pump comms detected from other rigs (this is a good thing!)
Preflight OK.
Attempting to retrieve MDT CGM data from pump
MDT CGM data retrieved
MDT New cgm data reformat

No bolus needed (yet). jq: error: null and number cannot be divided
date: invalid date ‘@’
Settings less than 15 minutes old. jq: error: null and number cannot be divided
date: invalid date ‘@’
Edison battery at 67% is charged (>= 98%) or likely charging (60-70%). Pumphistory-24 < 15m old
jq: error: null and number cannot be divided
date: invalid date ‘@’
Pump profile refreshed; Settings refreshed; Refresh

Dana Lewis
@danamlewis
Dec 12 2017 15:53
Did you try OpenAPS first-upload?
Dave Acklam
@dcacklam
Dec 12 2017 16:39

Oh, on the subject of hardware...
Ran into this today:
https://shop.technexion.com/system-on-modules/pico/pico-modules.html/
Manufacturer claims they are Edison pin-compatible... Not IA-32 like the edison (ARM, ala RPi)... Cheaper than an Edison...

Not going to try and use one myself... But thought someone might be interested. It would obviously require lots of tinkering/setup to make work - but once you boot it into Linux, it theoretically should be able to run OpenAPS & oref0... Obviously not a great idea for the new-to-the-game folks - but a possible 'use' for existing explorer boards that otherwise might not have CPUs to go with them...

Jon Cluck
@cluckj
Dec 12 2017 17:09
cool!
Brian Cadre
@DrCadre_twitter
Dec 12 2017 17:13
@danamlewis What is OpenAPS first-upload?
Scott Leibrand
@scottleibrand
Dec 12 2017 17:15
You can run openaps first-upload to trigger upload of BGs to a NS site that doesn’t have any CGM data yet.
garykidd
@garykidd
Dec 12 2017 18:08
what's latest on pi zero pi hat?
Dana Lewis
@danamlewis
Dec 12 2017 18:29
Same as before. ETA Late January or early feb
Brian Cadre
@DrCadre_twitter
Dec 12 2017 18:36
@scottleibrand thanks. I will try it tonight.
garykidd
@garykidd
Dec 12 2017 18:42
thx...what kinda optimizations need to be figured out still?
Dana Lewis
@danamlewis
Dec 12 2017 19:09
there's Pi0 improvements to be made still, more stuff for setup ease, etc.
garykidd
@garykidd
Dec 12 2017 19:28
ok, thx. I was really wondering what seems to be bogging the system down---I guess after thinking about it---its a little surprising to me to learn about the issue because I hadn't thought the process was very demanding from a processor standpoint.
Scott Leibrand
@scottleibrand
Dec 12 2017 19:54
We got it down to less than 5m per loop already. Further improvements will be possible by reducing the Python overhead, such as by using @ecc1’s Go code to talk to the pump.
There are a few processor intensive things left in oref0, including COB calculation, due to the level of recursion required. But actual oref0 calculation only represents about 10% of the loop time now.
garykidd
@garykidd
Dec 12 2017 19:57
does each loop need to be shorter than 5 mins?
Jon Cluck
@cluckj
Dec 12 2017 20:12
for it to effectively run, yes. but right now the loops are taking a little bit too long and running the CPU at full blast all the time :)
meaning less battery life, extra heat, slow responsiveness in the UI, etc.
garykidd
@garykidd
Dec 12 2017 20:49
@scottleibrand @ecc1 Go code to talk to MM will provide better performance [than OpenAPS?] --is it a lighter setup?
@cluckj thx
Scott Leibrand
@scottleibrand
Dec 12 2017 20:50
@garykidd it has less overhead and can talk to the pump much more quickly (and uses less CPU)
garykidd
@garykidd
Dec 12 2017 20:51
cool. sounds like it may help considerably..is it otherwise compatible in term sof functions it provides?
Scott Leibrand
@scottleibrand
Dec 12 2017 20:52
dunno yet. lots of work to do to integrate it and see where the gaps are.
garykidd
@garykidd
Dec 12 2017 20:52
k. cool. thx
Dave Acklam
@dcacklam
Dec 12 2017 21:01
@cluckj Python (which is what almost-everything in OpenAPS that isn't a shell-script is written in) is not exactly the fastest language out there...
Scott Leibrand
@scottleibrand
Dec 12 2017 21:02
oref0 is almost all Javascript (which isn't all that fast either), whereas the underlying openaps toolkit is Python
Dave Acklam
@dcacklam
Dec 12 2017 21:09
Ahh yes, forgot about the node.js part of things (haven't done much-if-anything to that)...

@garykidd
The shorter the loops, the higher the 'resolution' (the faster it recalculates and re-enacts basal levels).

A 5min , or 8min, or 10min 'slice' of time very easily contains changes in BG/etc that would influence a theoretical 'realtime' algoritihm, which are not reacted to...

Kind of like the difference between loan interest compounded monthly vs continuous-compounding...

garykidd
@garykidd
Dec 12 2017 21:38
right. Still I wonder if looping every 5 mins would really create any significant issues vs. more frequent loop---I dont think I'd be surprised to find it works well enough...but maybe others have done this and I am wrong...
Jon Cluck
@cluckj
Dec 12 2017 22:12
in my experience, the "slow" looping doesn't do a whole lot to treatment
the explorer-hat as it is does three loops in fifteen minutes, so the impact is not bad (except on the CPU and battery)
Diadon81
@Diadon81
Dec 12 2017 23:37
Dears, after update my dev branch, my rig looped with following error:
Old settings: ls: cannot access settings/profile.json: No such file or directory Profile invalid: ls: cannot access settings/profile.json: No such file or directory Couldn't refresh_old_profile
after that rig starts mmtune and everything repeat
also I found profile.json.new in settings with zero size.