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

17th
Oct 2017
Scott Leibrand
@scottleibrand
Oct 17 2017 00:02
Screenshot 2017-10-16 17.01.55.png
jaylagorio
@jaylagorio
Oct 17 2017 00:14
I really like the no 0 and no U, and I like the .1. Since the bubbles are so small it's not like you can mistake two .1s close together as 1.1 because that produces a much bigger bubble.
Steve Bernacki
@sbernacki
Oct 17 2017 00:23
Hi, I just upgraded 0.5.2 -> 0.5.5. Question, do I need to run the setup script from scratch or is it OK to run oref0-runagain.sh ?
Scott Leibrand
@scottleibrand
Oct 17 2017 00:24
and the blue bubbles are hard to see against the green line without a label. so I think I'll go with no 0 and no U, but leave all labels on.
fidoman
@efidoman
Oct 17 2017 01:01
@scottleibrand :+1: for no 0 and no U and leave all labels on.
Scott Leibrand
@scottleibrand
Oct 17 2017 01:10
nightscout/cgm-remote-monitor#2926
(which builds on nightscout/cgm-remote-monitor#2919)
if anyone can test that rendering branch and provide feedback on those PRs, it'd be appreciated.
Jake Punshon
@JakePunshon
Oct 17 2017 01:25
@scottleibrand +1 for your most recent screenshot... no leading 0 and no U but show all SMBs
Jacob H
@jdhigh
Oct 17 2017 03:51
I just installed latest dev. I'm getting this message in my logs now: Listening for 40s silence before mmtuning: .No interfering pump comms detected from other rigs (this is a good thing!)
mmtune: KeyError: 'mmtune'
Usage: grep [OPTION]... PATTERN [FILE]...
Try 'grep --help' for more information.
Usage: grep [OPTION]... PATTERN [FILE]...
Try 'grep --help' for more information.
No wait required.
Unsuccessful supermicrobolus pump-loop at Mon Oct 16 23:51:07 EDT 2017
Scott Leibrand
@scottleibrand
Oct 17 2017 03:57
hmm, that's probably a bug, but not sure. Are you using an Edison?
Jacob H
@jdhigh
Oct 17 2017 03:58
yes
Scott Leibrand
@scottleibrand
Oct 17 2017 03:58
try service cron stop; killall -g oref0-pump-loop; cd ~/myopenaps; openaps mmtune; service cron start
and then try grep -v setFreq monitor/mmtune.json | grep -A2 $(json -a setFreq -f monitor/mmtune.json)
Jacob H
@jdhigh
Oct 17 2017 03:59
Ok, well, I just restarted the install npm run global-install. So it's halfway through that.
Scott Leibrand
@scottleibrand
Oct 17 2017 03:59
k
Jacob H
@jdhigh
Oct 17 2017 04:00
Now that I think of it, I don't recall being asked if I wanted to stop the cron, etc which is usual after running it... It might have ended prematurely.
Scott Leibrand
@scottleibrand
Oct 17 2017 04:01
ah, yeah. I was wondering if you somehow had a partial install, or cruft from an old one
maybe try re-running oref0-setup first
Jacob H
@jdhigh
Oct 17 2017 04:03
That's probably it. Setup's running now. I'll let you know how it goes.
This is how setup ended:
libtool: install: /usr/bin/install -c obexd/src/obexd /usr/local/libexec/bluetooth/obexd
/bin/mkdir -p '/usr/local/lib/cups/backend'
/bin/bash ./libtool --mode=install /usr/bin/install -c profiles/cups/bluetooth '/usr/local/lib/cups/backend'
libtool: install: /usr/bin/install -c profiles/cups/bluetooth /usr/local/lib/cups/backend/bluetooth
/bin/mkdir -p '/etc/dbus-1/system.d'
/usr/bin/install -c -m 644 src/bluetooth.conf '/etc/dbus-1/system.d'
/bin/mkdir -p '/usr/local/share/man/man1'
/usr/bin/install -c -m 644 tools/rctest.1 tools/l2ping.1 tools/bccmd.1 tools/btattach.1 tools/hid2hci.1 '/usr/local/share/man/man1'
/bin/mkdir -p '/usr/local/share/man/man8'
/usr/bin/install -c -m 644 src/bluetoothd.8 '/usr/local/share/man/man8'
/bin/mkdir -p '/lib/udev/rules.d'
/usr/bin/install -c -m 644 tools/97-hid2hci.rules '/lib/udev/rules.d'
/bin/mkdir -p '/lib/udev'
/bin/bash ./libtool --mode=install /usr/bin/install -c tools/hid2hci '/lib/udev'
libtool: install: /usr/bin/install -c tools/hid2hci /lib/udev/hid2hci
cp: cannot create regular file ‘/usr/local/bin/bluetoothd’: Text file busy
Couldn't make bluez
Scott Leibrand
@scottleibrand
Oct 17 2017 04:08
ok, apparently our bluez install doesn't work when it's running
one sec
let's try the steps manually with some extra checks
Jacob H
@jdhigh
Oct 17 2017 04:11
ok
Scott Leibrand
@scottleibrand
Oct 17 2017 04:11
cd ~/src/bluez-5.47
sudo cp ./src/bluetoothd /usr/local/bin/
ps aux | grep -v grep | grep bluetoothd
Jacob H
@jdhigh
Oct 17 2017 04:12
cp: cannot create regular file ‘/usr/local/bin/bluetoothd’: Text file busy
root@:~/src/bluez-5.47#
Scott Leibrand
@scottleibrand
Oct 17 2017 04:13
ps aux | grep -v grep | grep bluetoothd
Jacob H
@jdhigh
Oct 17 2017 04:13
root 7884 0.0 0.1 5948 1796 ? S 00:06 0:00 sudo /usr/local/bin/bluetoothd --experimental
root 7901 0.0 0.1 5088 1892 ? S 00:06 0:00 /usr/local/bin/bluetoothd --experimental
Scott Leibrand
@scottleibrand
Oct 17 2017 04:13
killall bluetoothd
then ps aux | grep -v grep | grep bluetoothd again
Jacob H
@jdhigh
Oct 17 2017 04:14
ok
Scott Leibrand
@scottleibrand
Oct 17 2017 04:15
oh, I bet it's a race condition with the oref0-bluetoothup in cron
Jacob H
@jdhigh
Oct 17 2017 04:17
Does it have to do with the BlueZ update? openaps/oref0#724
Scott Leibrand
@scottleibrand
Oct 17 2017 04:19
yeah. I've got a PR for you to test
openaps/oref0#726
Jacob H
@jdhigh
Oct 17 2017 04:22
I hate to admit... how do I test this PR?
Scott Leibrand
@scottleibrand
Oct 17 2017 04:22
I just added a comment to the PR to answer that: refresh. :)
Jacob H
@jdhigh
Oct 17 2017 04:25
do I need to npm run global-install? Or just go to setup?
Scott Leibrand
@scottleibrand
Oct 17 2017 04:25
not if you just did for dev
Jacob H
@jdhigh
Oct 17 2017 04:36
It was able to complete setup. Just fixed my pref file and rebooted. Watching loop log now.... and it's good. Looping as normal.
Thanks for your help!! @scottleibrand
Scott Leibrand
@scottleibrand
Oct 17 2017 04:44
cool, thanks for finding it and helping me get it fixed quickly
Martin Haeberli
@mhaeberli
Oct 17 2017 05:18
bluetooth failed - getting this error on hardwire serial connection:
[  680.208732] wifi_platform_set_power = 0
[  686.924651] 
[  686.924651] Dongle Host Driver, version 1.141.59 (r)
[  686.924651] Compiled in /home/robin/src/edison-linux-helper/edison-bcm43340 on Apr  4 2017 at 22:42:52
[  686.924826] wl_android_wifi_on in
[  686.944370] wifi_platform_set_power = 1
[  687.500691] F1 signature OK, socitype:0x1 chip:0xa94c rev:0x2 pkg:0x0
[  687.510009] DHD: dongle ram size is set to 524288(orig 524288) at 0x0
[  687.519778] dhdsdio_download_firmware: firmware path=/lib/firmware/fw_bcmdhd.bin, nvram path=/lib/firmware/bcmdhd.cal
[  687.656580] sdioh_request_buffer: [1] doing memory copy buf=f50f8000, len=2037
[  687.668810] dhdsdio_write_vars: Download, Upload and compare of NVRAM succeeded.
[  687.838445] dhd_bus_init: enable 0x06, ready 0x06 (waited 0us)
[  687.848431] wifi_platform_get_mac_addr
[  687.848517] wifi_get_mac_addr_intel: unable to open /config/wifi/mac.txt
[  687.868080] Firmware up: op_mode=0x0015, MAC=fc:c2:de:32:f0:19
[  687.898798] Firmware version = wl0: Aug 15 2014 15:09:17 version 6.20.190.2 (r497061) FWID 01-bb9bbd07
[  687.909954] dhd_preinit_ioctls wl ampdu_hostreorder failed -23
look forward to suggestions
Martin Haeberli
@mhaeberli
Oct 17 2017 06:19
help - no bnep interface showing up but bluetoothd installed
Martin Haeberli
@mhaeberli
Oct 17 2017 07:09
ok - reminder to self and all - iOS hotspot sometimes spontaneously turns itself off. If a previously working rig fails to bluetooth, this is the first thing to check
Paul Dickens
@thebookins
Oct 17 2017 08:23
Hey all, noticed tonight COB dropped from 46 to zero in the space of one loop, and oref0 zero-temped in response. The logs showed Warning: setting mealCOB to 0 because currentDeviation is null/undefined. I haven't seen this before - should I expect it to resolve itself?
cameronrenwick
@cameronrenwick
Oct 17 2017 11:11
@jaylagorio totally agree...like no 0, no U but would like all the bolus', regardless of size. Particularly when reading on my phone, .1's (if only the blue bubble) would likely be missed, at least by me
kempera
@kempera
Oct 17 2017 12:19
Dear all, Thanks for your help on the date problem - I was able to fix it. Unfortunately, a new problem crashed APS:
It seems to be related to mmtune and compression, but all my troubleshooting doesn't work so far:

Listening for 30s silence before mmtuning: .No pump comms detected from other rigs
mmtune: '
"868.390", 5, -52 Unsuccessful supermicrobolus pump-loop at Tue Oct 17 14:19:43 CEST 2017

Starting supermicrobolus pump-loop at Tue Oct 17 14:20:02 CEST 2017 with 10 second wait_for_silence:
MDT CGM configured; not waiting
Radio ok. Listening: .No pump comms detected from other rigs
Preflight '
fail. SMB pump-loop failed. grep: monitor/temp_basal.json: No such file or directory

PopcornPam
@PopcornPam_twitter
Oct 17 2017 12:24
Hey all, where can I find info about 'exercise mode'?
kempera
@kempera
Oct 17 2017 12:24
I believe the problem is related to "git gc" but the suggested removal of gc.log didn't work... Any thoughts or suggestions? Thanks
grafik.png
cameronrenwick
@cameronrenwick
Oct 17 2017 12:36
in order to test 2926 I'm thinking I need to use something like cd ~/src/oref0 && git fetch && git checkout make SMBs and predictions fit better upgrade && git pull and re-run oref0-setup / oref0-runagain but just am not 100% sure.. can someone help with this?
garykidd
@garykidd
Oct 17 2017 13:01
@PopcornPam_twitter I think...it's synonymous with temporary (high target) before exercise...like eating soon relates to temporary low target preceding meals...
Diabetic_Dude
@diabetic_dude_twitter
Oct 17 2017 13:07
@cameronrenwick I do not believe that is correct.... what is your objective?
the command you typed above will try to pull a certain branch for example only "cd ~/src/oref0 && git checkout tiny-basil-eSMB && git pull" would pull the tiny-basil-eSMB branch from github
then "npm run global-install" would install it
philipgo
@philipgo
Oct 17 2017 13:11
@cameronrenwick The PR you are talking about is a Nightscout PR. It has nothing to do with oref0 setup
cameronrenwick
@cameronrenwick
Oct 17 2017 13:12
duh.. of course! thx
philipgo
@philipgo
Oct 17 2017 13:13
You would have to merge the PR in your Nightscout repository, but I have not yet done that myself.
Diabetic_Dude
@diabetic_dude_twitter
Oct 17 2017 13:22
nightscout/cgm-remote-monitor#2926
but I am not sure it can be merged yet??? I am trying figure it out, reading this: https://help.github.com/articles/merging-a-pull-request/
Dana Lewis
@danamlewis
Oct 17 2017 13:37
@popcornpam_twitter see the notes in the dev PR about exercise mode
@garykidd it's going to be slightly different for high targets moving forward ^
ejoe132
@ejoe132
Oct 17 2017 14:03
Going to be order a whole new setup, are the breakout boards still run monthly?
Dana Lewis
@danamlewis
Oct 17 2017 14:07
@ejoe132 if you have a solid explorer board edison rig already, I would hold off. Writing a blog post to explain more, but going to recommend people consider waiting a few weeks and going with a Pi hat for next rig
ejoe132
@ejoe132
Oct 17 2017 14:10
@danamlewis Ok, thanks for the heads up
Brian Rabinovitz
@bdr1177
Oct 17 2017 14:20
This is a typical pattern we see after meal boluses, where openaps predicts bg to go lower and zero temps. BG almost always rises steeply, contrary to the prediction, and then we end up with a high temp that sometimes causes a roller coaster. I'm sure that I'm just not optimizing settings, but I feel like bolus snooze used to prevent the initial zero temp. I know bolus snooze is or will be phased out. How would I better deal with this situation. (We're on 0.6.0-dev from two nights ago and don't use SMB because of small basals).
Capture.JPG
Capture2.JPG
Brian Rabinovitz
@bdr1177
Oct 17 2017 14:29
I should add that autotune doesn't seem to have recommendations that would help with this problem. It mostly suggests a much higher ISF and generally lower basals. Its recommendation for CR is to go from 40 to 38, but with the bolus increments on the 722, this will make almost no difference in practice.
Velibor Maric
@vebaba
Oct 17 2017 14:30
Only 0.5U for 23UH?
Brian Rabinovitz
@bdr1177
Oct 17 2017 14:31
@vebaba Yes, this is for my 6 year old daughter. Here insulin doses are low.
It may be that she needs a lower CR, but the data hasn't shown this to be the case. In any event, my question here has primarily to do with the openaps predictions.
Dana Lewis
@danamlewis
Oct 17 2017 14:35
@bdr1177 plz also share insulin curve, dia, etc
Velibor Maric
@vebaba
Oct 17 2017 14:38
@bdr1177 OK, I understand
Meghan Rutledge
@mar5477
Oct 17 2017 14:42
Ran into the issue of getting SMB for rescue carbs last night. Just read that I should enter them and high temp. I usually enter rescue carbs via IFTTT, anybody know a way to enter rescue carbs and high temp with minimal button pushes
Ugh. I scrolled down in the docs, nevermind. If you're wondering, it's {"enteredBy": "IFTTT-button", "eventType": "Temporary Target", "reason": "low treatment", "carbs": 5, "targetTop": 120, "targetBottom": 120, "duration": 60, "secret": "your_hashed_api_goes_here!!!"}
garykidd
@garykidd
Oct 17 2017 14:43
@mar5477 it's somehting that's been discussed multiple times but IFTTT appears to be the simplest way to work with---after it's set up that is..
Dana Lewis
@danamlewis
Oct 17 2017 14:45
@mar5477 glad you found it :smile:
Meghan Rutledge
@mar5477
Oct 17 2017 14:46
Another thing though, I under-bolused my pre-bolus for breakfast this morning and got SMB right after to bring me up to what the suggested bolus would have been and before I even started eating (and so there definitely wasn't any carb action happening). I thought SMB was supposed to wait until it saw something happening to enact?
image.png
garykidd
@garykidd
Oct 17 2017 14:48
hmmm I thought the carb entry and pre-bolus trigger.... what you're saying is def. true for UAM though.
Meghan Rutledge
@mar5477
Oct 17 2017 14:49
image.png
Less than 10 minutes after my initial bolus I had an additional .6U with no real change in BS. So it wasn't UAM that caused that?
Scott Leibrand
@scottleibrand
Oct 17 2017 15:01
@thebookins that currentDeviation warning usually means you're missing CGM data, and will self-resolve when you have 15m of good data again.
Diabetic_Dude
@diabetic_dude_twitter
Oct 17 2017 15:01
@scottleibrand could you please help a few people understand how to merge nightscout/cgm-remote-monitor#2926 into our nightscout site for testing purposes???? @cameronrenwick and @garykidd and I would love to do some testing for you but I do not think any of us have gotten too far at all. I have been reading for a bit and think I may not have the authorization but am not sure at all
Scott Leibrand
@scottleibrand
Oct 17 2017 15:05
@diabetic_dude_twitter thx for the early morning troubleshooting help. Just FYI, we never recommend basil for BGs, it doesn't work nearly as well as basals (or even cinnamon) :stuck_out_tongue_winking_eye:
Diabetic_Dude
@diabetic_dude_twitter
Oct 17 2017 15:05
:P
garykidd
@garykidd
Oct 17 2017 15:05
It's awesome on pizza/pasta though.
Scott Leibrand
@scottleibrand
Oct 17 2017 15:12
@bdr1177 can you capture the OpenAPS pill or (preferably) pump-loop.log from when it set the zero temp after meal bolus?
For testing NS branches, I think you just change your target deploy branch from dev in heroku.
Diabetic_Dude
@diabetic_dude_twitter
Oct 17 2017 15:15
but do any of the NS branches have #2926 in them?
Scott Leibrand
@scottleibrand
Oct 17 2017 15:17
@mar5477 that SMB after carb entry behavior in dev is based on the remainingCA predictions, which assume carbs will hit over 3h. It works really well for no-bolus operation. If you end up going low as the meal starts, you might not have to prebolus as early. If that isn't working well, LMK and we can look at the new algorithm with your data.
@diabetic_dude_twitter look at the top of the PR: it says what branch its from.
Diabetic_Dude
@diabetic_dude_twitter
Oct 17 2017 15:18
:( UGH! so samn simple!
(I still cant spell either :fire: !!!)
Meghan Rutledge
@mar5477
Oct 17 2017 15:22
@scottleibrand I'll try tomorrow with less time between prebolus and eating and see how it goes. I'm skeptical of just doing eating soon and entering carbs with no bolus to take care of everything, but is that what's really recommended??
jaylagorio
@jaylagorio
Oct 17 2017 15:23
@mar5477 It feels weird, but it actually works really well.
I've been doing it with dinners for the past couple of weeks and it's going pretty well.
garykidd
@garykidd
Oct 17 2017 15:35
sorry...back to basics here...regarding #2926....after I go to ....nightscout/cgm-remote-monitor#2926
is first step to fork?
Samuchco
@Samuchco
Oct 17 2017 15:49
Screenshot (9).png
Zero temping after bolus
is that normal?
Is taking a whole lot of Temp Target "eating soon" to bring him down
Scott Leibrand
@scottleibrand
Oct 17 2017 15:56
@mar5477 our recommendations are rather fluid at this point as the capabilities change, but we've found using Fiasp that pre-entering carbs and lettings SMBs ramp up works even without a meal bolus. if you're on Novolog/Humalog/Apidra/Novorapid, then a prebolus might still be warranted, which you can either do manually when you enter the carbs, or which you could let SMB do for you (possibly followed by a top-off bolus that takes into account how much SMB has delivered). at this point we're all experimenting with what works best, both in terms of outcomes and ease of use
@Samuchco you're running dev? If so, please update to a version from the last 16h or so: there's a fix for that
@garykidd are you deploying from your own fork or directly from the NS repo?
Samuchco
@Samuchco
Oct 17 2017 15:58
@scottleibrand Thanks. Do I have to do a full reinstall or can I do a shortcut.
Brian Rabinovitz
@bdr1177
Oct 17 2017 15:58
@scottleibrand Can't get the logs right now because I'm at work, but here's what the pill looked like right after the zero temp.
Capture3.JPG
Scott Leibrand
@scottleibrand
Oct 17 2017 15:59
@bdr1177 was that with the ! enableUAM fix?
@Samuchco no need to run oref0-setup again, just the git pull and npm run global-install steps
Brian Rabinovitz
@bdr1177
Oct 17 2017 15:59
This was dev from when we were troubleshooting two nights ago. So if that fix was yesterday, probably not.
Samuchco
@Samuchco
Oct 17 2017 16:00
Thanks. will do it tonight when he comes from school
Scott Leibrand
@scottleibrand
Oct 17 2017 16:00
@bdr1177 I can't keep all the fixes straight, so probably best to just update and try again.
Brian Rabinovitz
@bdr1177
Oct 17 2017 16:00
Okay, I'll try that this evening.
Scott Leibrand
@scottleibrand
Oct 17 2017 16:03
@mar5477 if you do decide to test no-bolus, be aware that bg will likely rise more than with a bolus for a given amount of carb pre-entry. entering the carbs a bit earlier might mitigate that a bit. but what most people find with no-bolus is they no longer have to worry as much about post-meal lows, as it's much more conservative about only bolusing for entered carbs when BG data confirms it's safe to do so. (if you don't eat at all, you'll still go low from the prebolus, but it's safer for meals where the food actually eaten is only 2/3 of the carb estimate, or absorption is delayed for some reason)
and there are some further improvements in the PR pipeline for the UAM logic that supports safe SMBs when carb estimates are off in either direction, which will help a bit more there
cameronrenwick
@cameronrenwick
Oct 17 2017 16:10
@mar5477 I can share that in my experience (Fiasp w using IFTTT buttons for carb entry only - letting SMB do the incremental bolusing) that it works well, but if I go above a certain number of carbs I have to prebolus from my pump. This is good/bad as since using SMB I've let myself get a bit more "carby" with food which is pure laziness. When I eat well, and have sensible amounts of carbs, simply using IFTTT buttons works great. My best advice is to start a 10gm carb amount with IFTTT button and then work your way up.
Diabetic_Dude
@diabetic_dude_twitter
Oct 17 2017 16:10
yup yup yup
I still use my pump for larger meals like pizza for meals with rice
jaylagorio
@jaylagorio
Oct 17 2017 16:12
I share @cameronrenwick's experience although I'm still on Novolog.
cameronrenwick
@cameronrenwick
Oct 17 2017 16:12
^^ me too
Meghan Rutledge
@mar5477
Oct 17 2017 16:23
Okay, can you help me logic through this too? Let's say I eat exactly 30 grams at basically the same time every day, but my IFTTT button is only set for 25, so that's what I do. (Based on what people have said, I think this kind of estimation is actually okay.) Autotune is running nightly and after a couple days of seeing insulin for 25g not be enough because I'm actually eating 30g it changes my CR. Then one day I eat 60g instead and enter 60g, my BS is going to go low because it assumes my CR is different based on the estimation? Where does the line get drawn between CR being off and UAM?
Scott Leibrand
@scottleibrand
Oct 17 2017 16:28
it's a very fuzzy line. at the beginning of the meal, it doses mostly off the carb entry, but without a BG rise will only get up to dosing for about 2/3 to 3/4 of the carbs. when BG starts to rise, it updates its predictions and doses for the rest of it. as carbs decay, it gradually switches over to more and more of its decision being based on UAM instead of the COB predicted BGs. toward the end of the meal it's largely dosing based off the observed BG deviations using the UAM algorithm, and tapering off the effect of the COB predBGs as COB drops toward zero.
cameronrenwick
@cameronrenwick
Oct 17 2017 16:28
@mar5477 I can't answer the math part of that aside from saying I've done some experimenting with about 6 different IFTTT buttons for carbs ranging from 10 - 70. I've eaten these amounts of carbs, and sometimes have purposely eaten more/less to see how things play out. My CR (right or wrong) hasn't changed much (maybe a gram or 1.5 at max) according to autotune, which I run nightly. So, perhaps my situation is unique and others will find wide swings in the CR from autotune... but mine has not changed much
Scott Leibrand
@scottleibrand
Oct 17 2017 16:29
the blending is linear based on the % of carbs remaining as COB, so when 30 of your 60g have decayed, it will be making decisions half on your carb estimate and half on the UAM one
cameronrenwick
@cameronrenwick
Oct 17 2017 16:30
having been doing IFTTT only for last 2 weeks, my CR on my pump is 12; autotune is 12.33... and hasn't moved within that "12" over this time that I've seen
Scott Leibrand
@scottleibrand
Oct 17 2017 16:30
autotune shouldn't change CR much based on one day's data, but @mar5477 is right that if you always get 30g worth of rise from a 25g entry, it will eventually tune CR to be 25/30ths of what it would've been otherwise, which will make a 60g carb estimate behave like a 60*30/25=72g one. given that it doesn't fully trust the carb estimate anyway, that should be fine
cameronrenwick
@cameronrenwick
Oct 17 2017 16:31
I think this still comes down to appropriate carb counting
Scott Leibrand
@scottleibrand
Oct 17 2017 16:32
@danamlewis doesn't do carb "counting" any more so much as "rough carb guesstimating", and it seems to work well :)
carb entry precision levels of 15, 30, 60, 90 seem "good enough"
jaylagorio
@jaylagorio
Oct 17 2017 16:33
@scottleibrand It's a very fuzzy disease
Meghan Rutledge
@mar5477
Oct 17 2017 16:35
I made a google assistant carb entry recipe, so I just have to tell it what to enter and not have a million buttons! I guess I just need to do some more testing..
Scott Leibrand
@scottleibrand
Oct 17 2017 16:36
@jaylagorio true that
jaylagorio
@jaylagorio
Oct 17 2017 16:37
@mar5477 That's stupid cool! Do you plan to write that up somewhere?
Diabetic_Dude
@diabetic_dude_twitter
Oct 17 2017 16:38
just an FYI just this error after updating
Connecting to www.kernel.org (www.kernel.org)|147.75.196.57|:443... connected.
ERROR: The certificate of ‘www.kernel.org’ is not trusted.
ERROR: The certificate of ‘www.kernel.org’ is not yet activated.
The certificate has not yet been activated
Couldn't download bluez
Scott Leibrand
@scottleibrand
Oct 17 2017 16:39
@diabetic_dude_twitter what is your rig date?
Meghan Rutledge
@mar5477
Oct 17 2017 16:39
@jaylagorio I can add it to the docs! It's pretty much the same as any other IFTTT trigger. I just added one for rescue carbs too. Instead of saying "ok google, enter carbs" I say "ok google, enter rescue carbs".
Scott Leibrand
@scottleibrand
Oct 17 2017 16:40
nice. yes, please PR that in :)
cameronrenwick
@cameronrenwick
Oct 17 2017 16:40
@mar5477 ya!! I'd love it if you could share that recipe! @scottleibrand I too have found that the 6 I use are "pretty close" to what I need but when added to the other 5 or 6 buttons I have for temps etc it'd be nice to see the google assistant version (hint hint @mar5477 )
jaylagorio
@jaylagorio
Oct 17 2017 16:40
Yeah!
Diabetic_Dude
@diabetic_dude_twitter
Oct 17 2017 16:42
@scottleibrand Fri Dec 31 18:18:24 CST 1999
Why do you ask what is today's date??? Do you have any basil????
Meghan Rutledge
@mar5477
Oct 17 2017 16:42
image.png
Dang it. Here's what I have for the trigger (the THIS). THAT is the same Webhooks request as all your other buttons
image.png
@jaylagorio @cameronrenwick
jaylagorio
@jaylagorio
Oct 17 2017 16:45
Nice!
Scott Leibrand
@scottleibrand
Oct 17 2017 16:46
@diabetic_dude_twitter nope, but at that time I was at my folks' place, and pretty sure their spice cabinet was fully stocked. ;-)
Diabetic_Dude
@diabetic_dude_twitter
Oct 17 2017 16:50
@scottleibrand tell me this, if I have two different rigs running on two different NS websites but both are attached to the same pump, can i run them at the same time? Or will it cause conflicts?
Scott Leibrand
@scottleibrand
Oct 17 2017 16:51
Only if you enter carbs in one NS but not the other.
or they have different BG data.
Diabetic_Dude
@diabetic_dude_twitter
Oct 17 2017 16:52
ok
thnx
kempera
@kempera
Oct 17 2017 16:53
grafik.png
Sorry, but is someone an expert with mmtune? Strange error message that a file is corrupt... I don't know why and how to fix it. Thanks
Hence, the OpenAPS doesn't work at all now unfortunately....
jaylagorio
@jaylagorio
Oct 17 2017 16:56
@kempera Try running oref0-git-reset in your myopenaps directory
kempera
@kempera
Oct 17 2017 16:57
@jaylagorio Thx, I tried it, but got the response: command not found
anything I have to install?
jaylagorio
@jaylagorio
Oct 17 2017 17:02
No, because it's actually oref0-reset-git and I'm a dumb
Try that
garykidd
@garykidd
Oct 17 2017 17:05
I think I'm on master direct NS repo. @scottleibrand
Scott Leibrand
@scottleibrand
Oct 17 2017 17:06
@garykidd if you're on the direct NS repo, then you can just change the branch, as that's where my branches are too
@kempera if you can't get your git errors figured out and are using / want to use dev, it will get rid of ~/myopenaps/.git for you and use --nogit
kempera
@kempera
Oct 17 2017 17:12
@jaylagorio Thx, this command was executed, but unfortunately didn't solve the problem...
Scott Leibrand
@scottleibrand
Oct 17 2017 17:13
if you don't want to use dev, you can manually rm -rf ~/myopenaps/.git and re-run oref0-setup
kempera
@kempera
Oct 17 2017 17:13
@scottleibrand I tried to enter --nogit but it didn't work
garykidd
@garykidd
Oct 17 2017 17:13
I'm looking for ZT branch then?
Scott Leibrand
@scottleibrand
Oct 17 2017 17:13
@garykidd I would use rendering
@kempera --nogit is a backend thing, not something you enter: to get rid of git, just follow the docs instructions for installing dev
kempera
@kempera
Oct 17 2017 17:16
@scottleibrand Ok Scott, thanks! That's a helpful comment and sorry for not being up to speed on dev... still learning
garykidd
@garykidd
Oct 17 2017 17:16
I take that back---maybe I am on a already forked version...
cameronrenwick
@cameronrenwick
Oct 17 2017 17:26
I think we're all around this... trying to get the NS repo working too... but it's saying review req'd, 2 checks failed... any luck with this?
garykidd
@garykidd
Oct 17 2017 17:27
gotta put it down for afew mins...be back shorttly
Diabetic_Dude
@diabetic_dude_twitter
Oct 17 2017 17:29
@mar5477 should the webhooks body {"enteredBy": "IFTTT-button", "reason": "low treatment", "carbs": #, "secret": "your_hashed_api_goes_here!!!"} look like this? will the # work or does a number need to be added
kempera
@kempera
Oct 17 2017 17:31
@scottleibrand Dear Scott, I deleted the .git folder and reinstalled it, as you suggested - and it works again!!! A big thank you to you & Dana for your great achievements and your ongoing support - it's unbelievable how you and the whole community encourage us in this challenging time!
Meghan Rutledge
@mar5477
Oct 17 2017 17:37
@diabetic_dude_twitter yes, I forgot about that! You just use "#" and it will sub in whatever number you said
Diabetic_Dude
@diabetic_dude_twitter
Oct 17 2017 17:42
@mar5477 is yours showing up on your NS site? mine is not, but if I just substitute the # for an actual number it will work
philipgo
@philipgo
Oct 17 2017 17:43
@cameronrenwick The "checks required" don't matter, the branch works fine
Meghan Rutledge
@mar5477
Oct 17 2017 18:06
@diabetic_dude_twitter you need to add an ingredient
image.png
put your cursor next to the colon, click "Add Ingredient" at the bottom of the Body box, then "NumberField"
Diabetic_Dude
@diabetic_dude_twitter
Oct 17 2017 18:09
@mar5477 awesome that worked like a charm!
make that PR that will be some valuable stuff for others
Meghan Rutledge
@mar5477
Oct 17 2017 18:10
Awesome! I'm adding to the docs right now, so if there's anything else that wasn't straightforward let me know!
Meghan Rutledge
@mar5477
Oct 17 2017 18:16
You can use text and number ingredients for google assistant, so I'm wondering if you could even use it to say "ok google, set 'Site Change'" or some other event type?
Diabetic_Dude
@diabetic_dude_twitter
Oct 17 2017 18:17
oh stop it! now your making me want to buy an android phone....
Meghan Rutledge
@mar5477
Oct 17 2017 18:20
Siri isn't supported yet, but there are some workarounds.
Diabetic_Dude
@diabetic_dude_twitter
Oct 17 2017 18:24
@mar5477 FYI yes it works just tested by adding a new pump site change command and it worked
did not need to change a thing in the webhooks just needed to set up the google assistant part :sparkles:
Meghan Rutledge
@mar5477
Oct 17 2017 18:26
@diabetic_dude_twitter I'm going to DM you to brainstorm some other uses!
Scott Leibrand
@scottleibrand
Oct 17 2017 18:58
@diabetic_dude_twitter @mar5477 you can use "OK, Google" on iOS with the Google App
and if you want a completely hands-free solution, you can do "Hey Siri, launch Google" first ;-)
garykidd
@garykidd
Oct 17 2017 18:59
Siri, google said you're stupid...but Alexa said something I cannot repeat here...
garykidd
@garykidd
Oct 17 2017 19:09
@scottleibrand are you sure the pi0 will be robust enough of a platform to handle oref0/1?
after invoking whatever optimizations that group can think of?
Diabetic_Dude
@diabetic_dude_twitter
Oct 17 2017 19:09
@scottleibrand I know just playing around. I can tell you though the google assistant features are pretty nice
garykidd
@garykidd
Oct 17 2017 19:10
I think @scottleibrand was toally siri..ous...I was joking though
Scott Leibrand
@scottleibrand
Oct 17 2017 19:12
@garykidd we'll do the work required to make it work. there is a ton of inefficiency in our python + node invocations: just a question of how much optimization we have to do to get it fast enough.
garykidd
@garykidd
Oct 17 2017 19:13
that's really cool...rigs are going to get pretty cheap then...
Scott Leibrand
@scottleibrand
Oct 17 2017 19:13
I did test the old basal-only pump-loop with the Pi0W and was able to get it to loop, so it's mostly just a matter of getting it fast enough to support the SMB pump-loop
Craig Brenner
@cbrenner
Oct 17 2017 19:13

Asking again here to see if I can get some guidance on this "' AttributeError: 'NoneType' object has no attribute 'iter_records' " I reported a few days ago. It happened again last night and trying to resolve other than a hard reboot of the rig...

Here is my original request with OpenAPS pill, pump log and NS log: https://gitter.im/nightscout/intend-to-bolus?at=59e151045c40c1ba799a9fc0

The error which continues to randomly shows ups is "AttributeError: 'NoneType' object has no attribute 'iter_records' " . The log info below shows how it began last night but the OpenAPS pill, pump log and NS log from the other day is identical to last night sans the date/time:

Oct 16 22:49:06 AttributeError: 'NoneType' object has no attribute 'iter_records'
Oct 16 22:49:06 brenner_rig cgm-loop.log: 188 2017-10-16T22:41:42-07:00
Oct 16 22:50:02 brenner_rig cgm-loop.log: Mon Oct 16 22:50:02 PDT 2017
Oct 16 22:50:02 brenner_rig network.log: Starting oref0-online.
Oct 16 22:50:03 brenner_rig network.log: At Mon Oct 16 22:50:02 PDT 2017 my local IP is: 10.0.0.30
Oct 16 22:50:03 brenner_rig autosens-loop.log: Mon Oct 16 22:50:03 PDT 2017
Oct 16 22:50:03 brenner_rig autosens-loop.log: settings/autosens.json

Scott Leibrand
@scottleibrand
Oct 17 2017 19:13
@garykidd the Explorer HAT + Pi0W will probably be a little cheaper than Edison + Explorer Block, but not too much
we'll have to see where it comes out with manufacturing and assembly
@cbrenner what command is giving that error?
garykidd
@garykidd
Oct 17 2017 19:14
OK, saw some others using alternate radio board that was really cheap.
Scott Leibrand
@scottleibrand
Oct 17 2017 19:14
@cbrenner if that's coming from the cgm-loop.log, are you actually using a plugged-in CGM? or getting CGM data from NS?
Craig Brenner
@cbrenner
Oct 17 2017 19:15
This message was deleted
Scott Leibrand
@scottleibrand
Oct 17 2017 19:15
if you don't have a CGM plugged in, that error is totally normal and expected, and you can completely ignore cgm-loop.log
Craig Brenner
@cbrenner
Oct 17 2017 19:16
@scottleibrand Well, as soon as that error starts the rig goes into waiting mode and I have to reboot. See my original post https://gitter.im/nightscout/intend-to-bolus?at=59e151045c40c1ba799a9fc0
Scott Leibrand
@scottleibrand
Oct 17 2017 19:18
I suspect that cgm-loop.log error is always occurring, and you only notice it when something else is broken. your real problem seems to be an error getting CGM data from NS. are you running oref0 master or dev?
Craig Brenner
@cbrenner
Oct 17 2017 19:18
oref0 master (0.5.5)
Nalla306
@Nalla306
Oct 17 2017 19:18
I'm having real issues updating night scout ton ightscout/cgm-remote-monitor#2919. I'm connected to Nalla306/cgm-remote-monitor in heroku and have changed the manual deploy a GitHub branch to dev and deployed but don't see any of the changes. I'm probably doing something simple wrong but any tips appreciated
Scott Leibrand
@scottleibrand
Oct 17 2017 19:19
@cbrenner ok. there is an improved oref0-ns-loop in dev now, with better logging, but let's look at what you have first
Craig Brenner
@cbrenner
Oct 17 2017 19:19
@scottleibrand BTW, this only happened after updating to 0.5.5
Scott Leibrand
@scottleibrand
Oct 17 2017 19:20
@Nalla306 you probably want the rendering branch (from nightscout/cgm-remote-monitor#2926) rather than dev. if you want to actually test nightscout/cgm-remote-monitor#2919 that would be the ZT branch
@cbrenner what does your ns-loop.log have during that time?
Craig Brenner
@cbrenner
Oct 17 2017 19:21
@scottleibrand here you go
NS Loop and Pump Log.pdf
Scott Leibrand
@scottleibrand
Oct 17 2017 19:23
and does network.log show a public IP during that timeframe?
Nalla306
@Nalla306
Oct 17 2017 19:24
I don't see it though, only options are master, dev and wip/azurepush wip/node and wip/null .....im doing this in heroku, I look at the page on the link nightscout/cgm-remote-monitor#2919 but see no options to do anything with it. Apologies for the questions
Scott Leibrand
@scottleibrand
Oct 17 2017 19:25
it looks like the ns-loop.log got stuck on one of its downloads from 15:36:20 until it finally completed (or timed out) at 15:53:06
@Nalla306 you're probably using your own fork rather than the nightscout repo.
Nalla306
@Nalla306
Oct 17 2017 19:26
I thought that but can't figure out how to change....will have a google
garykidd
@garykidd
Oct 17 2017 19:27
@scottleibrand #2919 works well...thanks
Scott Leibrand
@scottleibrand
Oct 17 2017 19:27
@Nalla306 or you can PR in the change to your repo
garykidd
@garykidd
Oct 17 2017 19:28
Love the 2 hour feature too! Good idea.
if you open and then merge that PR, you can then re-deploy from your own dev branch
@garykidd on my local copy I changed 3H to 4H as well, so it's 2H 4H 6H 12H 24H. I doubt most NS users would appreciate that level of cheese-moving though
garykidd
@garykidd
Oct 17 2017 19:30
I pushed it out as Master...probably not best form but assuming its rather arbitrary...?
Scott Leibrand
@scottleibrand
Oct 17 2017 19:31
@garykidd you can do whatever you want to your own repo: just a question of what's easiest for you
and you can always delete your repo and re-fork if needed
(if you did that you could deploy straight from your new local rendering branch)
Craig Brenner
@cbrenner
Oct 17 2017 19:33
Oct 13 15.pdf
garykidd
@garykidd
Oct 17 2017 19:34
yup. I think the part for me that was the most foreign (especially since it's been so long since I've done PR) was wasnt completely sure about which 'thing' I was pulling from / ...and therefor 'to'
Scott Leibrand
@scottleibrand
Oct 17 2017 19:34
@cbrenner so you had a network outage that caused it to give up on wifi and try bluetooth at 15:40
and it looks like the ns-loop.log was in the middle of something and didn't give up on it at that time, but rather took until 15:53 (10m after connectivity was restored) to time out.
Nalla306
@Nalla306
Oct 17 2017 19:35
Thanks, think I've got it now! Cheers
garykidd
@garykidd
Oct 17 2017 19:35
@scottleibrand was you're 2/4/8/12 under 'ZT' branch by chance?
Scott Leibrand
@scottleibrand
Oct 17 2017 19:36
@garykidd it's not pushed to a NS branch currently
garykidd
@garykidd
Oct 17 2017 19:36
just locally then..ok
Scott Leibrand
@scottleibrand
Oct 17 2017 19:36
and I'm doing 2/4/6/12/24, not 8
garykidd
@garykidd
Oct 17 2017 19:37
cool. go t the concept though
Craig Brenner
@cbrenner
Oct 17 2017 19:37
@scottleibrand Ok, the network recovered in 1 minute but OpenAPS was still in waiting mode for 30mins (see original OpenAPS pill) so I had to switch to our second rig and reboot the primary rig
Scott Leibrand
@scottleibrand
Oct 17 2017 19:37
technically, it was committed but reverted before I PR'd. you can revert the revert commit e60ff3aa9d6c6cb18fe8b414d95cef09e76350b6 to get the change
@cbrenner 30m? hmm. looking...
looks like ns-loop.log had new glucose data at 15:54:13
garykidd
@garykidd
Oct 17 2017 19:39
wow, just when something starts to seem easy
Scott Leibrand
@scottleibrand
Oct 17 2017 19:39
@garykidd or you can convince me to push yet another branch :)
@cbrenner which happens to be the same minute that you said you switched to backup
garykidd
@garykidd
Oct 17 2017 19:40
good point...will have to throw something crazy out there for you to fold in first...don't want to waste your time until it's worth it.
Craig Brenner
@cbrenner
Oct 17 2017 19:40
@scottleibrand Yup, I switched rigs at the point
Scott Leibrand
@scottleibrand
Oct 17 2017 19:40
@cbrenner but all of those entries say brenner_rig, so I think what you're showing is that the same rig recovered then too
although the mmtunes show -99 and -95, so apparently it wasn't able to talk to the pump again because you'd moved it
so the question is how we get ns-loop to recover faster when it gets into that state
Craig Brenner
@cbrenner
Oct 17 2017 19:43
@scottleibrand Yes, indeed. So, last night the exact same thing happened and it did not recover for 25+mins (and don't know when it would have) as I switched rigs again.
Craig Brenner
@cbrenner
Oct 17 2017 19:46
@scottleibrand when do you anticipate this moving to master?
Scott Leibrand
@scottleibrand
Oct 17 2017 19:47
if it's well tested and there's demand, we could patch release it, but it can be to dev much quicker
openaps/oref0#728
if you want to upgrade one of your rigs to the ns-get-timeout branch, you can test it right now. it'll come with the rest of dev, though. if you'd prefer to test a version based on master I can push a branch for that too
actually, easier to just do that than talk about it: https://github.com/openaps/oref0/tree/ns-get-timeout-master
Craig Brenner
@cbrenner
Oct 17 2017 19:50
@scottleibrand Thanks, prefer to test a branch based on master.
Scott Leibrand
@scottleibrand
Oct 17 2017 19:50
so you can just test with cd ~/src/oref0 && git fetch && git checkout ns-get-timeout-master && git pull && npm run global-install
Craig Brenner
@cbrenner
Oct 17 2017 19:51
@scottleibrand "then"
Scott Leibrand
@scottleibrand
Oct 17 2017 19:53
edited to clarify
"then" was outside the backticks as the last word of my sentence
Craig Brenner
@cbrenner
Oct 17 2017 19:54
@scottleibrand Ok, thanks. Will test later tonight when son is home
cameronrenwick
@cameronrenwick
Oct 17 2017 20:05
question about IFTTT and carb buttons. If my pump has a target top of 5.2 and target bottom of 5.2, when creating a recipe for carb buttons, is it necessary to have a temp target as well?
Scott Leibrand
@scottleibrand
Oct 17 2017 20:24
the reason for temp targets on rescue carbs is to disable SMBs while they take effect. if you set a temp target above 100 mg/dL (5.6 mmol/L or higher) that will turn off SMB activity until the temp target expires.
garykidd
@garykidd
Oct 17 2017 20:26
@scottleibrand not appreciating why you can't make that the default assumption...say when BG < X, treat new carbs as rescue carbs..otherwise treat them normally? At a certain point rescue carbs assumption seems entirely logical.
Scott Leibrand
@scottleibrand
Oct 17 2017 20:27
and if BG is 120 but heading downward, and you eat carbs, are those rescue carbs or meal carbs?
cameronrenwick
@cameronrenwick
Oct 17 2017 20:27
thx @scottleibrand .. that makes a ton of sense. FTR, my IFTTT buttons for carbs is working well as are SMB (aside from the need for a bit more aggressive SMB at first but we've chatted about that - fingers crossed that's a feature that's being worked on....)
Scott Leibrand
@scottleibrand
Oct 17 2017 20:28
@garykidd I'm focused instead of making the algorithm work better at making SMBs safe even in a rescue carb situation, as not everyone marks them every time, and sometimes you're doing rescue carbs but then meal carbs, and don't want SMB disabled very long, etc. etc.
cameronrenwick
@cameronrenwick
Oct 17 2017 20:28
fortunately I don't need to use rescue carbs hardly ever so no trouble with turning SMB off etc
Craig Brenner
@cbrenner
Oct 17 2017 20:52

@scottleibrand so this is new... Just ran cd ~/src/oref0 && git fetch && git checkout ns-get-timeout-master && git pull && npm run global-install and re ran setup and now keep getting these broadcast messages....

Broadcast message from root@brenner_rig (Tue 2017-10-17 13:49:01 PDT):
Rebooting to fix radio errors!
The system is going down for reboot at Tue 2017-10-17 13:51:01 PDT!

never got this before
garykidd
@garykidd
Oct 17 2017 21:03
@scottleibrand in reading your last response---thinking---I think that's what my suggestion would do. realistically if you sat down to eat 100 carbs meal when your BG was 50, for example, ...I still think a portion of that carbs entry should be considered rescue carbs...think # of carbs that are needed to get you back in acceptable BG range...with the balance being non-rescue carbs. this way if you didn't log them--a portion would be treated as rescue carbs regardless. Just a thought.
if BG is 120 but heading down, is a different question as to whether any ingested carbs should be treated as rescue or not...still like the concept that would assume a portion of carbs that are needed to bring you within acceptable BG range should be attributed to rescue type. again, just an idea.
Scott Leibrand
@scottleibrand
Oct 17 2017 21:09
@garykidd the "carbs needed to get you back up to target" part is already accounted for. the question is whether reacting to the "extra" carbs like you normally would (with SMBs) is likely to be a problem, for example because there's activity, or a bad CR, or something else going on that is causing BG to drop more than expected.
plus there's the fact that straight sugar can spike BG in a way that looks like it's going to be a big sustained rise, well beyond the entered carb count, but that then ceases very suddenly.
the original UAM algorithm got tricked badly by that. the newer versions of UAM (particularly in openaps/oref0#714) are much better about not overreacting to that
@cbrenner radio errors like that can be caused by having two processes trying to access the radio at once, which can happen at oref0-setup time. if it clears on its own, it'll cancel the reboot, and if not the reboot will fix it.
garykidd
@garykidd
Oct 17 2017 21:13
which, one ma y think, warrants a more conservative approach with delivering insulin when you're low..even though the logic already compensates for your BG is...I'm saying, perhaps it should be curtailed moreso until min BG is achieved...its the inverse of the comment I made last night about more aggressive response when BG exceeds a certain level...(again, understanding that by definiition, the repsonse is always more...but I'm saying perhaps even more so---at least initially until visibility into acceptable range is highly certain.
garykidd
@garykidd
Oct 17 2017 21:19
It might be OK to go a little higher than you would otherwise to thwart a re-peat low but being a little extra careful when you're low....conversely, it might also beneficial (although likely somewhat less important) to treat high BGs quicker and even harder than normal when BG is very high). Clearly, you've thought about this stuff a ton...and way further than I have...Still..I'm just trying to think through some thoughts that keep entering my mind. Thanks, Scott
 Loving the latest 'rendering' SMB' visualization....even if it isn't MSB....  thanks!
Scott Leibrand
@scottleibrand
Oct 17 2017 21:22
we do have a minGuardBG safety check in place that will prevent SMBs if any predicted BG value is below threshold (which is 70 for a target of 100). the overreaction can come after the rescue carbs start to kick in, when BG has actually risen back to normal levels, and the question is how long it'll continue rising, and how fast to give insulin to cover it
I've recently added a feature to autosens that will add negative deviations whenever you're running a high temp target, which should help with making OpenAPS more aware of sensitivity that develops during the day when there's too much carb activity for it to detect it through normal autosens methods using basal deviations...
that's in openaps/oref0#705
also, the openaps/oref0#714 changes do address the "treat high BGs quicker and even harder than normal when BG is very high" situation, by allowing it to super bolus for high BGs as long as it has plenty of runway to prevent a crash by zero temping
garykidd
@garykidd
Oct 17 2017 21:26
is there somewhere you can point me to that would explain all the output for l that shows on 0.6.0. I think I understand most of the output---but suspect that I can learn more
Scott Leibrand
@scottleibrand
Oct 17 2017 21:27
aside from the inside of my brain, the most thorough documentation is the code itself. we still need some help to write up some more user-friendly summaries, so feel free to ask questions about specific sections, and take notes to help us update the docs...
garykidd
@garykidd
Oct 17 2017 21:28
happy to help as well..but just have to make sure I have it correct before I waste everyones time/effort.
Scott Leibrand
@scottleibrand
Oct 17 2017 21:29
yep, happy to explain things as long as someone's willing to help capture that for others. :)
Craig Brenner
@cbrenner
Oct 17 2017 21:29
@scottleibrand ok. looks like it resolved on its own
garykidd
@garykidd
Oct 17 2017 21:29
sure. cool. thanks.
Diabetic_Dude
@diabetic_dude_twitter
Oct 17 2017 21:29
@scottleibrand the documentation in the code itself is mind blowing
Scott Leibrand
@scottleibrand
Oct 17 2017 21:30
I've found if I don't comment well I can't figure out what I did later. :)
Diabetic_Dude
@diabetic_dude_twitter
Oct 17 2017 21:30
meaning I have never seen it laid out so well and understandable in any other github project before
you do that well... I have been trying to learn javascript for a bit to learn what is going on before I looked at the code... Then I looked at the code and was like, dang the class is practically right there in the code.
Scott Leibrand
@scottleibrand
Oct 17 2017 21:31
lol nice
it might help that I had to learn JS to do all of this too
so I tend to use the fairly simple and straightforward code features that are easier to understand (if less compact and elegant)
Diabetic_Dude
@diabetic_dude_twitter
Oct 17 2017 21:33
and you keep it pretty clean
(which makes it way easier to understand )
Scott Leibrand
@scottleibrand
Oct 17 2017 21:33
there are certainly some elements that are convoluted and could use some refactoring to simplify, though
Diabetic_Dude
@diabetic_dude_twitter
Oct 17 2017 21:35
everything could be better, when you quit learning, you quit living. same goes for improving
Scott Leibrand
@scottleibrand
Oct 17 2017 21:39
:+1:
Craig Brenner
@cbrenner
Oct 17 2017 21:40
@scottleibrand Do anticipate 'openaps/oref0#705' and 'openaps/oref0#714' and other current pull requests to be part of 0.6.0 master release or might they go into a future 0.5.x master release?
Scott Leibrand
@scottleibrand
Oct 17 2017 21:43
any algorithm changes like that will definitely be on 0.6.x. the only thing that might go into any future 0.5.x releases (if we decide they're necessary) would be bug fixes like openaps/oref0#728
the goal right now is to get everything merged to dev and properly tested so we can start prepping 0.6.0 for release
PieterGit
@PieterGit
Oct 17 2017 22:11
what is the difference between fullNewISF, adjustedISF and newISF, e.g.
Limiting adjusted ISF of 31.23 to 75.00 (which is pump ISF of 90 / 1.2 )
p50deviation: 1.9 p50BGI -2.7 p50ratios: 0.347 Old ISF: 90 fullNewISF: 31.23 adjustedISF: 75 newISF: 87

Autotune pump profile recommendations:
---------------------------------------------------------
Recommendations Log File: /root/myopenaps/autotune/autotune_recommendations.log

Parameter      | Pump     | Autotune
-------------------------------------
ISF [mg/dL/U]  | 90.000   | 87.000
Carb Ratio[g/U]| 12.000   | 12.055
the various ISF's seem to be quite of of each others range
Scott Leibrand
@scottleibrand
Oct 17 2017 22:20
fullNewISF is what the day's data says. that is adjusted, if applicable, and then a percentage of that adjustment is applied to the current ISF to get the newISF
PieterGit
@PieterGit
Oct 17 2017 22:22
so yesterday was an ISF of 31.23?
Scott Leibrand
@scottleibrand
Oct 17 2017 22:26
the limited subset of yesterday's data suitable for calculating ISF would've come up with an ISF of 31, yes
you'll find the fullNewISF is very noisy, which is why we limit/adjust it and then only adjust thew autotuned ISF slightly
PieterGit
@PieterGit
Oct 17 2017 22:28
we lowered the carb ratio because we we're getting quite some highs, i dind't suspect such a major impact on the isf values
we'll see how it progresses.
is the current autotune resistent to missing pump and cgm data. didn't dare to use it when offline for quite some hours a day, and not sure about good internet connectivity
Scott Leibrand
@scottleibrand
Oct 17 2017 22:31
it should avoid changing things it has no data for. worst case is it'll have just enough data to make an adjustment, but the adjustment won't be fully informed, so might not be in the right direction. it won't change much, though, and future days' data should fix it
PieterGit
@PieterGit
Oct 17 2017 22:34
ok, didn't know if i had the possibility of internet connection abroad, and didn't want to risk it. on hindsight keeping autotune running would have been better probably . good lesson
Scott Leibrand
@scottleibrand
Oct 17 2017 22:35
yeah, hard to know in advance which would be better
you can try a manual autotune run on those days and see what it comes up with
PieterGit
@PieterGit
Oct 17 2017 22:37
i'll try to do some analysis, but currently days with too many lows have probably an undererstimation of carbs because rescue carbs are often not entered :worried:
hypo's and rescue carbs cause lack of proper carb entries (easy to say for non-diabetic of course)
i hope that i can PR some of the docs of our findings of the last weeks. but first want to be sure that i make the right assumptions about proper advice that should be documented. i'm out, keep up the good work. I still have quite some small and medium pr's that I would like to finish in the coming week (e.g. pebble integration)
CyclingChic
@CyclingChic
Oct 17 2017 22:47
Hi everyone. Very new to this site. My husband is the computer guru, but he's making dinner right now ;)~. I've been using OpenAPS for a few weeks now, it's awesome! Forever grateful to Dana and Scott for all you've done. It's hard to explain how I've felt these last few weeks. My mind has had the privilege of taking a break (if that makes sense?) with all the thoughts/worries/forethought/afterthought etc that comes with living with diabetes. Seriously - forever grateful to you both and to everyone else who's given my husband advice on fine tuning my settings. Quick question for the group: AIOB - I've historically placed this setting at 3 hours. I'm finding now at times esp. when exercising in the later afternoon (I've figured out how to use IFTTT) that I'm still going low because my basal rates were too high prior to exercising. Not sure if the recommendation from the group would be to set my IFTTT target higher than 140 or figure out a way to simply set this target ahead of time (how long before?), or should I just try setting my overall AIOB setting from 3 hours to 4 hours. I mean let's face it, Humalog really is in the body for longer than 3 hours. It's just not fast enough... Last question - has anyone tried using Fiasp in their rig yet? Thanks in advance.
Scott Leibrand
@scottleibrand
Oct 17 2017 22:51
@CyclingChic sounds like you're about ready to try the dev branch. it has exponential insulin curves that do way better than the old curve (and allow you to properly set DIA out to 5h+ without affecting peak time). it also has a new exercise mode that adjusts IOB calculations based on your temp target, which should do a much better job with the way you're using it now.
and yes, lots of people are looping with Fiasp: there's an ultra-rapid curve for that. we've even found you can get away with not meal bolusing using Fiasp and 0.6.0-dev: https://diyps.org/2017/09/29/not-bolusing-for-meals-fiasp-0-6-0-algorithm-in-oref0-dev-branch-and-more/
sbernacki @sbernacki thinks he'll be upgrading the rig to dev tonight 😂
Scott Leibrand
@scottleibrand
Oct 17 2017 22:53
more info about exercise mode: openaps/oref0#684
and the dev PR with documentation on lots more: openaps/oref0#707
CyclingChic
@CyclingChic
Oct 17 2017 22:55
Thank you sosooo much Scott! Will look into this and let you know how it goes.
Scott Leibrand
@scottleibrand
Oct 17 2017 22:55
:+1: