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

28th
Mar 2016
Jason Calabrese
@jasoncalabrese
Mar 28 2016 00:02
maybe a duplicate treatment? the NS IOB doesn't consider temps
Matt Pressnall
@logichammer
Mar 28 2016 00:06
I was wondering that, too...check mongo for the last three hours, right? Thanks!
Jason Calabrese
@jasoncalabrese
Mar 28 2016 00:07
you might see it in the treatments report
Matt Pressnall
@logichammer
Mar 28 2016 00:11

You are a genius...clearly!
!(http://content.screencast.com/users/loki_mdog/folders/Snagit/media/f5373ed6-caf7-4569-b59c-82bd669f702a/03.27.2016-17.10.png)

Also, I should be ashamed to admit I didn' t know about this report...and that I can edit the data here directly without needing to mess with mongo! Thanks!

Actually, there were two of them...got double uploaded it looks like...all should be well now!
Jason Calabrese
@jasoncalabrese
Mar 28 2016 00:15
glad you found it, wonder how it go there, the post is using an upsert now that should prevent dupes
Matt Pressnall
@logichammer
Mar 28 2016 00:16
hmmmm...first I have run into it...If it happens again, I'll see what kind of conditions it was under to make it happen.
Jim Matheson
@jmatheson
Mar 28 2016 00:17
crazy to watch my BG impact from stress
Bill Stackpole
@billstackpole
Mar 28 2016 00:20
@itsn0tm3 - awesome. Look forward to seeing you there!
Matt Pressnall
@logichammer
Mar 28 2016 00:21
@jasoncalabrese, only weird thing we did was have the pi switching between networks from the mobile to the home network. Not sure if that helps for debugging purposes but that's what I know. Thanks again!
Jason Calabrese
@jasoncalabrese
Mar 28 2016 00:22
do you have a check in your loop to prevent uploading when the pump is still bolusing? think that could cause it
https://github.com/jasoncalabrese/indy-e3/blob/master/openaps.ini#L3 && test $(cat monitor/status.json | json bolusing) == false) || (echo Pump bolusing, waiting to pull history; exit 1)"
Matt Pressnall
@logichammer
Mar 28 2016 00:24
No, I don't have the preflight stuff integrated into my code yet but am going to be making some changes (including using the TI stick this week). Nice! Thanks for the code!
Jim Matheson
@jmatheson
Mar 28 2016 00:24
Seems like some people tune every time and some people tune never. The optimal way i would think would be to use some logic to decide if we should try to retune. Is anyone doing something like this?
Jason Calabrese
@jasoncalabrese
Mar 28 2016 00:27
I was thinking about something like only doing mmtune if there's a failure, but what I'd really like is for mmtune to be smarter and keep it's own state, then it could quickly check the last best freq and skip most of the scan
Jim Matheson
@jmatheson
Mar 28 2016 00:28
yeah. that would be good. i am not sure how walking around near a device impacts things. i’d like it to just go for it, if a problem then try to tune, if still an issue than try to reset the port.
also thinking about trying to figure out making it smart enough to figure out which device is plugged in where and then use the right config to talk to the pump based upon what it figured out. basically so only need one openaps.ini file
and possibly just work if you yanked out a TI and plugged in a carelink
Matt Pressnall
@logichammer
Mar 28 2016 00:51
Nice, @bewest ! This stuff is looking super robust and filled out!
Ben West
@bewest
Mar 28 2016 00:52
btw, I think that one includes your PRs to docs
Matt Pressnall
@logichammer
Mar 28 2016 00:55
Great! Glad that they made it in / hope they help other folks. Want to add in something about prereqs / finding a good simple linux / ssh tutorial for folks so they can get started with that side of it even if it takes them awhile to find a pump. Maybe hereish:
http://bewest-openaps-docs.readthedocs.org/en/wip-guides/docs/walkthrough/phase-0/setup.html
And where they could practice...like a $5 a month website from Bluehost.com will give you the ability to ssh into something.
Jim Matheson
@jmatheson
Mar 28 2016 00:59
so AMA pulls carb entries from NS as well as the pump? @scottleibrand
Jason Calabrese
@jasoncalabrese
Mar 28 2016 01:02
There's a field for it, but didn't look like it was working
Jim Matheson
@jmatheson
Mar 28 2016 01:02
gather command in there is nothing like mine.
That file has a report in there to grab it from NS
Dana Lewis
@danamlewis
Mar 28 2016 01:04
Yes @jmatheson pulls from either place if you set it up that way
Jim Matheson
@jmatheson
Mar 28 2016 01:05
thx
Dana Lewis
@danamlewis
Mar 28 2016 01:05
(For example, I do carb entry in DIYPS and goes to NS; from NS to openaps.)
Jason Calabrese
@jasoncalabrese
Mar 28 2016 01:07
I saw that, but didn't see how it was used in the meal tool
Jim Matheson
@jmatheson
Mar 28 2016 01:08
i am struggling a bit here.
hard since everyone’s setup is a bit different
Scott Leibrand
@scottleibrand
Mar 28 2016 01:11
I have it set up to use pumphistory carbs if they exist, and NS carbs if they don't.
Jason Calabrese
@jasoncalabrese
Mar 28 2016 01:11
Every edit of my loop seems to turn into a bunch of typos
Thought there was a merge, but didn't look like that input was used
Kind of a hack.
Aaron Michelson
@Aaybob
Mar 28 2016 01:21

What is wrong to cause this?
$ git config --global user.email "diabeticaaron@gmail.com"
error: failed to write new configuration file /home/pi/.gitconfig.lock

I just installed dependencies and then sudo apt-get install git in the root directory. Do I need to first make a git directory or something?

Jason Calabrese
@jasoncalabrese
Mar 28 2016 01:25
Oh, I see what it's doing now, didn't notice it when looking at the diff
Aaron Michelson
@Aaybob
Mar 28 2016 01:29
ls
Aaron Michelson
@Aaybob
Mar 28 2016 01:40
I'm getting the "no space left on device" error...
the memory formatted with 8GB but when I wrote the raspberry image file to it it showed it was 59.9MB with 40MB available. What is going wrong. I ran the programs several times on old and new microSD cards.
Bill Stackpole
@billstackpole
Mar 28 2016 01:40
@Aaybob - you need to expand your card.
Aaron Michelson
@Aaybob
Mar 28 2016 01:40
oh in raspconfig?
Bill Stackpole
@billstackpole
Mar 28 2016 01:41
in the instructions (both for the raspi in the box and in the git repo instructions) there is an indication - can't remmeber the cmd - to expand the card to make all the space avaialble
Jason Calabrese
@jasoncalabrese
Mar 28 2016 01:41
What does df -hshow
Bill Stackpole
@billstackpole
Mar 28 2016 01:41
Had a similar prob
Aaron Michelson
@Aaybob
Mar 28 2016 01:42
I did do the expand file system in raspi-config
df -h shows /dev/root 3.6 Used 3.4G
Bill Stackpole
@billstackpole
Mar 28 2016 01:43
Aaron Michelson
@Aaybob
Mar 28 2016 01:44
;it has a list of filesstems tmpfs but the first lines shows /dev/root 3.6G 3.4 G Avail 0 use% 100% mounted on /
maybe i never rebooted
Bill Stackpole
@billstackpole
Mar 28 2016 01:46
someone indicated to me that if you are using Jessie, and it has Wolfram installed, that you can recover quite a bit of data by UNinstalling Wolfram. (and a reboot is required to resize the card)
Aaron Michelson
@Aaybob
Mar 28 2016 01:46
yes i forgot to reboot
Bill Stackpole
@billstackpole
Mar 28 2016 01:47
Hope that solves it (easy one - whee!)
Aaron Michelson
@Aaybob
Mar 28 2016 01:49
after reboot I'm at 7.1G 3.4G used 51% used
Bill Stackpole
@billstackpole
Mar 28 2016 01:52
woop!
:+1:
Aaron Michelson
@Aaybob
Mar 28 2016 01:53
thanks for being happy for me
Iain
@iainct
Mar 28 2016 02:29
a simple question, where on earth does the oref0 src/oref0 install on the pi? I can easily find it in home/src on my linux(vm) laptop that I did a preliminary openaps set up on... cheers.
but no dice on the pi.....
Jason Calabrese
@jasoncalabrese
Mar 28 2016 02:32
if you install the master/release version it will be in your node packages
which oref0-determine-basal will show /usr/local/bin/oref0-determine-basal, which is a symlink to /usr/local/lib/node_modules/oref0/bin/oref0-determine-basal.js
Iain
@iainct
Mar 28 2016 02:38
Thanks Jason, I am trying to do this step http://openaps.readthedocs.org/en/latest/docs/walkthrough/phase-2/Configure-Automatic-Sensitivity-Mode.html do I need to execute the git pull etc in the /usr/local/bin/ folder?
Scott Leibrand
@scottleibrand
Mar 28 2016 02:40
You need to install from source instead of packages. Quick-src.sh.
Jason Calabrese
@jasoncalabrese
Mar 28 2016 02:40
Iain
@iainct
Mar 28 2016 02:41
i ran the quick-src.sh install but changed to the loudnate g5 source...
the loudnate branch that works with the g5 is more accurate
everything is up and running i hope i don’t need to pull it apart again......!
Jason Calabrese
@jasoncalabrese
Mar 28 2016 02:46
if you already have the source setup you can go into oref0 and do a git pull to update dev
Iain
@iainct
Mar 28 2016 02:47
but i would need to be in the oref0 source folder? which i cannot find....
Scott Leibrand
@scottleibrand
Mar 28 2016 02:48
quick-src puts that in ~/src/oref0
Iain
@iainct
Mar 28 2016 02:51
i ran a local version of quick-src.sh with the loudnate changes in ~ $ would that mess things up?
the logs show a bunch of stuff went into here /usr/local/lib/node_modules/oref0/ but i get permission denied when i try to look....
Scott Leibrand
@scottleibrand
Mar 28 2016 02:54
Just run unmodified quick-src then switch to his branch and re-do the npm stuff for that one package.
Iain
@iainct
Mar 28 2016 02:56
thanks I will roll with that
Aaron Michelson
@Aaybob
Mar 28 2016 02:56
is there documentation for how to run from the g5? @iainct
re-do the npm stuff for a package is over my head
Iain
@iainct
Mar 28 2016 02:58
@Aaybob not that I could find, and I have managed to get it working twice and both are slightly different. basically pull git clone git://github.com/loudnate/dexcom_reader.git || \ (cd dexcom_reader && git pull) and run the python setups in there
looks like i am going to need to re-do it now so i will let you know....
Scott Leibrand
@scottleibrand
Mar 28 2016 02:59
s/npm/easy_install/ in what I said earlier then. :-)
Aaron Michelson
@Aaybob
Mar 28 2016 03:01
and I don't need g5 now....just restarted artificial pancreas from an old version I had on remote repository. I think I lost all my later work but I have a basic setup running without visualization now for just overnight use....I'm going into a tonne of overtime on my job so some good overnight blood sugars are going to help.
Iain
@iainct
Mar 28 2016 04:26
after all that and following the steps on i've hit this issue
pi@raspberrypi:~ $ cd src/ pi@raspberrypi:~/src $ cd oref0/ pi@raspberrypi:~/src/oref0 $ git pull Already up-to-date. pi@raspberrypi:~/src/oref0 $ git checkout meal-assist error: pathspec 'meal-assist' did not match any file(s) known to git.
pi@raspberrypi:~ $ cd src/
pi@raspberrypi:~/src $ cd oref0/
pi@raspberrypi:~/src/oref0 $ git pull
Already up-to-date.
pi@raspberrypi:~/src/oref0 $ git checkout meal-assist
error: pathspec 'meal-assist' did not match any file(s) known to git.
are the docs up to date....?
Scott Leibrand
@scottleibrand
Mar 28 2016 04:29
Someone just updated dev docs. You want dev branch now instead of meal-assist, since it got merged.
Iain
@iainct
Mar 28 2016 04:36
cool thanks!
ah hang on, dev branch of oref0?
Scott Leibrand
@scottleibrand
Mar 28 2016 04:42
Yes
Iain
@iainct
Mar 28 2016 04:43
pi@raspberrypi:~/src/oref0 $ git pull origin dev
From git://github.com/openaps/oref0
 * branch            dev        -> FETCH_HEAD
is that correct ^^^
Scott Leibrand
@scottleibrand
Mar 28 2016 04:44
Not sure, but just git checkout dev
well, not sure now if that will do the same thing
may have to git branch -d dev first
Iain
@iainct
Mar 28 2016 04:45
damn it!
pi@raspberrypi:~/src/oref0 $ git checkout dev
M    bin/oref0-raw.js
Already on 'dev'
Your branch is up-to-date with 'origin/dev'.
pi@raspberrypi:~/src/oref0 $ git checkout meal-assist
error: pathspec 'meal-assist' did not match any file(s) known to git.
Dana Lewis
@danamlewis
Mar 28 2016 05:07
Meal-assist isn't a branch anymore, it got merged to dev.
See what @scottleibrand said ^^ re delete first . Did that work now?
Stephan
@MosiGitHub
Mar 28 2016 12:52
@danamlewis where is the data of oref0.json used? I didn't find anything in the oref0-determine-basal. I was just wondering how it works. Thx
Scott Leibrand
@scottleibrand
Mar 28 2016 13:01
We need to get that doc updated with a better name for that report.
Oh, or just delete it. That is for openaps-predict, which isn't part of oref0 at all.
No, it uses oref0-determine-basal as the device...
Stephan
@MosiGitHub
Mar 28 2016 13:13
@scottleibrand thanks for the update. I had auto-sens and the meal-assist working, but right now I am fighting with the easy pumphistory.json that is not working anymore :-(
Scott Leibrand
@scottleibrand
Mar 28 2016 13:15
I actually do my 24h pumphistory as a separate report: https://github.com/scottleibrand/openaps-sh/blob/mmeowlink/setup.sh#L155
as the extra data makes it more likely to fail.
We probably should update the docs to do that too.
Stephan
@MosiGitHub
Mar 28 2016 13:20
i have two reports as well, one for uploading to NS with only one hour of data - but the one with 24 hours is failing at the moment- it was working 1 hour ago .... I don`t know why it is now failing, but finding out ;-)
Scott Leibrand
@scottleibrand
Mar 28 2016 13:22
You need 3-4h to loop, but 24h is optional, so I put that in the part of my loop that can fail without breaking anything.
Stephan
@MosiGitHub
Mar 28 2016 13:23
thanks - I have to make it more "fail-safe" as well
by the way ... restarted the Edison and it is working again
and the updated configure-automatic-sensitivity.mode.md makes much more sense .... that is why I was looking what the oref0.json was used for - finally for nothing. as suggested.json it will be enacted
Stephan
@MosiGitHub
Mar 28 2016 13:32
interesting is the following: 70.2U meal bolus remaining, .... my son had a total of 2.4U meal bolus ..... now the system decided to give max. temp basal .... I have to cancel that and look for the source of that .... I think someone has reported a similar problem two days ago
Scott Leibrand
@scottleibrand
Mar 28 2016 13:41
Eew, ya. Is that on dev?
Can you paste the determine-basal output?
Stephan
@MosiGitHub
Mar 28 2016 13:43
that is the old suggested.json ;-) not changed yet...
{"temp":"absolute","bg":106,"tick":"+0","eventualBG":103,"snoozeBG":103,"mealAssist":"Off: Carbs: 40 Boluses: 2.9 Target: 105 Deviation: 27 BGI: -9.16","reason":"70.2U meal bolus remaining, 103-103 in range: no temp required; setting current basal of 23.8 as temp","duration":30,"rate":1.5}
looks like the oref0.json ...
{"temp":"absolute","bg":106,"tick":"+0","eventualBG":104,"snoozeBG":104,"mealAssist":"Off: Carbs: 40 Boluses: 2.9 Target: 105 Deviation: 22 BGI: -7.39","reason":"70.2U meal bolus remaining, 104-104 in range: no temp required; setting current basal of 23.9 as temp","duration":30,"rate":1.5}
with a little difference (because of auto-sens I expect)
will fix this to the updated doc in a minute
Scott Leibrand
@scottleibrand
Mar 28 2016 13:46
I mean the stderr output of determine-basal, from your log file.
There must be something wrong with your profile.json
Stephan
@MosiGitHub
Mar 28 2016 14:02
sorry ... I looked at the messages and syslog file and didn't find anything usefull - which log do you mean?
Scott Leibrand
@scottleibrand
Mar 28 2016 14:03
You should be logging the output of determine-basal somewhere.
Most likely by logging the output of your main loop alias.
Some people send it to logger / syslog. I redirect straight to a loop.log file.
It is probably in your crontab entry.
Stephan
@MosiGitHub
Mar 28 2016 14:06
i normally log the output via cron, but i had cron disabled to find the mistake and was running the loop manually without logging ... normally the output goes in //var/log/messages
Scott Leibrand
@scottleibrand
Mar 28 2016 14:07
Ok. Then I would look at git log -p
start by looking for crazy numbers in profile.json
Stephan
@MosiGitHub
Mar 28 2016 14:07
here is the profile.json by the way: looks normal to me: {"max_iob":2,"type":"current","dia":3,"current_basal":0.4,"max_daily_basal":0.5,"max_basal":1.5,"min_bg":100,"max_bg":110,"sens":130,"carb_ratio":0.6}
ok
that is concerning the suggested.json from the git log:

Date: Mon Mar 28 15:53:02 2016 +0200

openaps-report invoke enact/suggested.json

      TODO: better change descriptions
      /usr/local/bin/openaps-report invoke enact/suggested.json

diff --git a/enact/suggested.json b/enact/suggested.json
index c2da1c8..30a5059 100644
--- a/enact/suggested.json
+++ b/enact/suggested.json
@@ -1 +1 @@
-{"temp":"absolute","bg":106,"tick":"+0","eventualBG":103,"snoozeBG":103,"mealAssist":"Off: Carbs: 40 Boluses: 2.9 Target: 105 Deviation: 27 BGI: -9.16","reason":"70.2U meal bolus remaining
, 103-103 in range: no temp required; setting current basal of 23.8 as temp","duration":30,"rate":1.5}
+{"temp":"absolute","bg":134,"tick":"+8","eventualBG":181,"snoozeBG":153,"mealAssist":"On: 100%, 100%, Carbs: 40 Boluses: 2.4 Target: 95 Deviation: 28 BGI: -3.41","reason":"70.7U meal bolus
remaining, Eventual BG 181>=110, adj. req. rate:25.2 to maxSafeBasal:1.5, no temp, setting 1.5U/hr","duration":30,"rate":1.5}

commit e6b5a7ab5a62d4168818be677d9186f06be6fba0

'''test'''
test

```Date: Mon Mar 28 15:53:02 2016 +0200

openaps-report invoke enact/suggested.json

      TODO: better change descriptions
      /usr/local/bin/openaps-report invoke enact/suggested.json

diff --git a/enact/suggested.json b/enact/suggested.json
index c2da1c8..30a5059 100644
--- a/enact/suggested.json
+++ b/enact/suggested.json
@@ -1 +1 @@
-{"temp":"absolute","bg":106,"tick":"+0","eventualBG":103,"snoozeBG":103,"mealAssist":"Off: Carbs: 40 Boluses: 2.9 Target: 105 Deviation: 27 BGI: -9.16","reason":"70.2U meal bolus remaining
, 103-103 in range: no temp required; setting current basal of 23.8 as temp","duration":30,"rate":1.5}
+{"temp":"absolute","bg":134,"tick":"+8","eventualBG":181,"snoozeBG":153,"mealAssist":"On: 100%, 100%, Carbs: 40 Boluses: 2.4 Target: 95 Deviation: 28 BGI: -3.41","reason":"70.7U meal bolus
remaining, Eventual BG 181>=110, adj. req. rate:25.2 to maxSafeBasal:1.5, no temp, setting 1.5U/hr","duration":30,"rate":1.5}

commit e6b5a7ab5a62d4168818be677d9186f06be6fba0
```

Bill Stackpole
@billstackpole
Mar 28 2016 14:18
Morning. I read and setup the instructions for bitbucket backup. Is there a writeup for the bitbucket restore process?
Scott Leibrand
@scottleibrand
Mar 28 2016 14:23
@MosiGitHub nothing new there. Need to see what the input files contained at the time.
Stephan
@MosiGitHub
Mar 28 2016 14:26
okay... I haven't changed anything, because my son is playing with a friend and I decided to keep the edison with me as long as the prblem isn't fixed. I will start checking the files - do you want me to post one (or more) of them?
Scott Leibrand
@scottleibrand
Mar 28 2016 14:27
You should see in git log -p diffs for the other files at the same time as those ones. Paste just the one-liner diffs for profile.json and anything that looks funny.
Stephan
@MosiGitHub
Mar 28 2016 14:28
okay... thanks for your support :clap:
Iain
@iainct
Mar 28 2016 14:40
@scottleibrand and @danamlewis thanks for clarifying the meal-assist --> dev the git syntax is new to me but i'm getting there, i've only used svn previously :-S I should be able to continue on now :-)
Stephan
@MosiGitHub
Mar 28 2016 14:55
@scottleibrand that I found concerning the profile.json - for me the carb-ratio looks maybe to low 0.6
iff --git a/settings/profile.json b/settings/profile.json
index 8b9b917..2ea4148 100644
--- a/settings/profile.json
+++ b/settings/profile.json
@@ -1 +1 @@
-{"max_iob":2,"type":"current","dia":3,"current_basal":0.35,"max_daily_basal":0.5,"max_basal":1.5,"min_bg":100,"max_bg":110,"sens":130}
+{"max_iob":2,"type":"current","dia":3,"current_basal":0.35,"max_daily_basal":0.5,"max_basal":1.5,"min_bg":100,"max_bg":110,"sens":130,"carb_ratio":0.6}
Scott Leibrand
@scottleibrand
Mar 28 2016 14:56
What is it on the pump?
You should also look at the contents of the report that comes from. Maybe settings.json?
Stephan
@MosiGitHub
Mar 28 2016 14:56
the pump uses 0.6 U of insulin for 10gr of KH - if we use the bolus wizard
Scott Leibrand
@scottleibrand
Mar 28 2016 14:57
Carb ratios are measured in g/U not U/g
unless this is a mmol thing
(i.e. a different standard in locations that use mmol/L, like MPG vs. L/100km)
Stephan
@MosiGitHub
Mar 28 2016 15:02
that is what I think the problem is... I recalculated the carb-ratio for the Careportal by just 1 divided by 0.6 was around 16 gramm per Unit
I think I have to switch the pump setting to the gramm basis instead of the BE basis and then I can use the 16 ... that should work ... KId is back from playing. So I can try that
Scott Leibrand
@scottleibrand
Mar 28 2016 15:05
So you're saying the pump has a mode for units per 10g carbs?
What does the settings report show when it's in that mode? We may need to add a mode check and reject unconverted units like we do for mmol/L.
Stephan
@MosiGitHub
Mar 28 2016 15:07
that is settings report from the lof:
log:
openaps-report invoke settings/settings.json

      TODO: better change descriptions
      /usr/local/bin/openaps-report invoke settings/settings.json
diff --git a/settings/settings.json b/settings/settings.json
new file mode 100644
index 0000000..bfaab1e
--- /dev/null
+++ b/settings/settings.json
@@ -0,0 +1,27 @@
+{
  • "low_reservoir_warn_point": 48,
  • "keypad_lock_status": 0,
  • "maxBasal": 1.5,
  • "temp_basal": {
  • "percent": 100,
  • "type": "Units/hour"
  • },
  • "low_reservoir_warn_type": 1,
  • "insulinConcentration": 100,
  • "audio_bolus_enable": false,
  • "variable_bolus_enable": true,
  • "alarm": {
  • "volume": 2,
  • "mode": 2
  • },
  • "rf_enable": false,
  • "auto_off_duration_hrs": 0,
  • "block_enable": false,
  • "timeformat": 1,
  • "insulin_action_curve": 3,
  • "audio_bolus_size": 0,
  • "selected_pattern": 0,
  • "patterns_enabled": true,
  • "maxBolus": 5.0,
  • "paradigm_enabled": 1
    +}
here is the carb_ratios.json before changing the setting
{
"units": "exchanges",
"raw": "0x02 0x00 0x06 0x0c 0x07 0x12 0x06 0x17 0x06 0x1f 0x06 0x2b 0x06 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00",
"first": 2,
"schedule": [
{
"start": "00:00:00",
"r": 6,
"ratio": 0.6,
"offset": 0,
"i": 0,
"x": 0
},
{
"start": "06:00:00",
"r": 7,
"ratio": 0.7,
"offset": 360,
"i": 12,
"x": 1
},
{
"start": "09:00:00",
"r": 6,
"ratio": 0.6,
"offset": 540,
"i": 18,
"x": 2
},
{
"start": "11:30:00",
"r": 6,
"ratio": 0.6,
"offset": 690,
"i": 23,
"x": 3
},
{
"start": "15:30:00",
"r": 6,
"ratio": 0.6,
"offset": 930,
"i": 31,
"x": 4
},
{
"start": "21:30:00",
"r": 6,
"ratio": 0.6,
"offset": 1290,
"i": 43,
"x": 5
}
]
here after changing:
{
"units": "grams",
"raw": "0x01 0x00 0x0f 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00",
"first": 1,
"schedule": [
{
"start": "00:00:00",
"r": 15,
"ratio": 15,
"offset": 0,
"i": 0,
"x": 0
}
]
interesting is units in both files: first: exchange, thereafter: grams
I was lazy and have only put in one value for the carb_ratio based on grams so far - just for testing
Stephan
@MosiGitHub
Mar 28 2016 15:17
@scottleibrand thanks for your support. Loop is running normally again. So I will switch the cron back one and have an eye on it after the next meal ... :+1:
Scott Leibrand
@scottleibrand
Mar 28 2016 15:45
can you do one more thing for me?: open an issue under oref0 to check the units and disable meal assist if the pump is in exchange units instead of grams
probably also need to disable meal assist if the "meal bolus remaining" is above a threshold too
this is a potential safety issue, so I want to get it fixed ASAP
Stephan
@MosiGitHub
Mar 28 2016 16:25
@scottleibrand i will do...
Jim Matheson
@jmatheson
Mar 28 2016 17:04
Example situation. Wake up 200, zero non basal IOB. 1 unit drips 25 points. Max basal allowed is 6u/hr or 3u/30m. Current basal is 2u/hr or 1u/30m. I need 4 units. Openaps will take 1 hour to give me 4 extra units. So, manual correction gives me 4u in 1m. Thus no choice but to manually correct.
with temp basal being over a period of time seems like there is no automated solution with our current params?
Dana Lewis
@danamlewis
Mar 28 2016 17:10
It is a choice. You happen to choose to manually correct; which makes sense. But if you were sleeping, that's the safest way for the system to bring you down over time. We don't want to enable automation of large doses of insulin for safety reasons.
Stephan
@MosiGitHub
Mar 28 2016 17:11
@scottleibrand : done
Dana Lewis
@danamlewis
Mar 28 2016 17:12
Thx @MosiGitHub
Jim Matheson
@jmatheson
Mar 28 2016 17:24
Yea. I just wanted to verify that corrections for the most part need to be done manually if you want your BG to come down in (what I consider) a reasonable amount of time. It’s a math issue due to the temp basal drip factor and our self imposed safety measures.
with that said, i don’t expect to ever be 200 with zero IOB if the AP is running 24x7
Dana Lewis
@danamlewis
Mar 28 2016 17:26
Yes. Also, it varies depending on what you consider reasonable, etc. For all of this, it's important to decide what you feel comfy staying hands off and letting the loop do and knowing when you want to do something manual (not just talking high correction, also re deciding when a low is a low that needs sugar, etc.) but understanding the parameters is :+1: for figuring that out for yourself
Jim Matheson
@jmatheson
Mar 28 2016 17:26
Last week totals review i saw my max BG was 185. pretty awesome :)
Dana Lewis
@danamlewis
Mar 28 2016 17:26
:)
Jim Matheson
@jmatheson
Mar 28 2016 17:28
my wall plug popped out last night so my pi died at 4am. I woke up at 175 and was trying to decide what to do. OpenAPS kicked in at the max as soon as it was back up and running but i did not want to lose the extra 45 minutes for it to drip what i needed to catch up.
in general i have been quick to the draw at overriding the system when BG is rising. will see how things go though today as i flip up auto sens and AMA
i’d prefer to let it run, but i am becoming OCD with my levels
This message was deleted
Stephan
@MosiGitHub
Mar 28 2016 18:26
@scottleibrand to deactivate meal-assist and auto-sens: is it sufficient to stop invoking the auto-sens.json and meal.json? I also switched the pumphistory back to 4 hours ... Maybe I try it after ski vacation again .... By the way: my wife hates the change to "grams" - therefor next time I will generate a static json report based on grams and leave the pump setting "exchange" ;-)
Scott Leibrand
@scottleibrand
Mar 28 2016 18:28
Yeah, and delete the output files for good measure. We should extend the units conversion stuff to convert exchange to grams too.
Stephan
@MosiGitHub
Mar 28 2016 18:40
that wouldn't be to difficult I expect just checking what value "units" has and in case of "exchange" x = 1/value ... deleting the output files makes sense. Tanks again
Ben West
@bewest
Mar 28 2016 18:51
it does
oh exchanges
hmm
amoskane
@amoskane
Mar 28 2016 19:12
hi y’all, is there anyplace where we’ve documented the basics of the T1 stick? I’m looking to get started with one, and realized I don’t know what product it is on the TI site...
;-)
has anybody an idea why my Azure based website is so slow ist is nearly not opening at all?
Jim Matheson
@jmatheson
Mar 28 2016 19:18
how long ago did you update to the most recent of dev NS?
if you do a lot of uploads from openaps it was going pretty slow, but there were some improvements made a few weeks ago which helps with speed quite a bit.
Stephan
@MosiGitHub
Mar 28 2016 19:29
a couple of weeks ago ... I will check wether there is an update - thx for the hint
there was an update - is now merged ... thanks
Jim Matheson
@jmatheson
Mar 28 2016 19:36
nice. quicker now?
Stephan
@MosiGitHub
Mar 28 2016 19:42
not really .. I deleted all records and changed the browser. Firefox is reporting a problem with this: azurewebsi…ublic/js/bundle.js?v=0.9.0-dev:56089 - if I stop this it is working again - strange
now it crashed ... I think I switch to Heroku for the rest of the evening ....
Jim Matheson
@jmatheson
Mar 28 2016 19:44
might want to ask on the NS channel. in general Heroku has worked well for me.
it seems like oref0 is using my target BG range from overnight (I have one set for Midnight-9am and one for the rest of the day). Does it not use time based ranges? and just pull the first one?
Dana Lewis
@danamlewis
Mar 28 2016 19:47
when are you pulling settings?
Jim Matheson
@jmatheson
Mar 28 2016 19:49
i can see both time based values in my bg-targets.json. I am not sure your specific question. Are you asking when I pull my settings.json in the order of my loop process?
it seems like the right settings are in the json, so i think it’d would not matter when i pull settings.
Bill Stackpole
@billstackpole
Mar 28 2016 19:55
backup / restore completed. New RasPi 3 configured from scratch, restored from BitBucket - easy peasy.
Literally two lines once Jessie installed and updated.
'''(line 1: git remote add origin https://<user>@bitbucket.org/<path>/ <reponame_from_bitbucket>.git )
(line 2: git pull https://<user>@bitbucket.org/<path>/<reponame_from_bitbucket>.git master)
Now have two built. Woop.
Only some weeks of debug left to complete <smile>. Tnx for everyone's help so far.
Scott Leibrand
@scottleibrand
Mar 28 2016 20:40
@jmatheson is profile.json updated?
Jim Matheson
@jmatheson
Mar 28 2016 20:41
yes, it’s current but it has the wrong value in there.
Scott Leibrand
@scottleibrand
Mar 28 2016 20:43
So do you have a profile.json report being run regularly to refresh it from the pump inputs?
Jim Matheson
@jmatheson
Mar 28 2016 20:44
yes
the report seems to be pulling the wrong values from my bg-targets.json
looks like it uses a default 90/100 if there is a problem, which also is the values I have set for Midnight to 9am :) so there is a chance it found a problem and used the default?
oh wait, so if you have something lower than 90/100 it just overrides it?
during the day i’ve got mine set to 90/90. so maybe it is just not allowing me to do that.
Scott Leibrand
@scottleibrand
Mar 28 2016 20:55
yeah, I remember we set some lower bounds about there. I would be fine with a PR to change them to 90/90
but probably not any lower than that
Dana Lewis
@danamlewis
Mar 28 2016 20:56
:+1:
Scott Leibrand
@scottleibrand
Mar 28 2016 20:56
obviously you can set whatever you want locally if you edit your copy of the code
Jim Matheson
@jmatheson
Mar 28 2016 20:56
yeah, i am on the bottom end of what i am comfortable with. and i pretty much am on the extreme aggressive end of that scale.
Scott Leibrand
@scottleibrand
Mar 28 2016 21:15
If no one has any objections to that (openaps/oref0#92, changing lower limit on max_bg from 100 to 90, to match min_bg of 90) I’ll be merging it soonish.
probably tonight or tomorrow
Jason Calabrese
@jasoncalabrese
Mar 28 2016 22:06
looks like they forgot to bolus for the afternoon snack, but it didn't really matter openaps took care of it
there was a check and calibration at the snack time, but no bolus in the pump history, I though it was a sync bug at first
Jim Matheson
@jmatheson
Mar 28 2016 22:09
nice
Jason Calabrese
@jasoncalabrese
Mar 28 2016 22:11
guess I'm going to put the school nurse out of a job
the school district has been sending a nurse 3 times a day for boluses since they don't want the health tech to bolus
Milos Kozak
@MilosKozak
Mar 28 2016 22:14
maybe time to start doing it remotely
Jason Calabrese
@jasoncalabrese
Mar 28 2016 22:25
I'd like to do it on a schedule
for snacks he just has a small bag of goldfish, think I could just put the carbs in the future and it might work out
Scott Leibrand
@scottleibrand
Mar 28 2016 22:36
yeah, that seems like a good application for AMA, since it will wait until it starts to see the carbs take effect, and then high-temp until it's covered all the carbs. Would be even better once we have the ability to set "eating soon" and activity modes by adjusting the BG target temporarily.
Jason Calabrese
@jasoncalabrese
Mar 28 2016 22:36
I was thinking carbs in the future would be eating soon
Scott Leibrand
@scottleibrand
Mar 28 2016 22:37
then you have to be able to edit them somehow...
Jason Calabrese
@jasoncalabrese
Mar 28 2016 22:38
yeah, we can do that now, I was avoiding having NS feed the loop, but seems like the best way
Scott Leibrand
@scottleibrand
Mar 28 2016 22:38
and with the AMA algorithm, carbs that haven't hit yet wouldn't affect things much: it only responds to them to the extent you're rising unexpectedly (vs. BGI from insulin activity)
so we'd have to add some sort of special-case detection of future-dated carbs if we wanted it to do anything...
Jason Calabrese
@jasoncalabrese
Mar 28 2016 22:39
thinking about adding an authorized field, so I can leave the open careportal for school, but flag the treatments that are authorized and only pull those
Scott Leibrand
@scottleibrand
Mar 28 2016 22:40
seems more elegant to allow entry (via NS API or something) of a temporary BG target. then have it decay back to normal target over 2h or whatever
Milos Kozak
@MilosKozak
Mar 28 2016 22:40
i was thinking about to do it from NS but i'll probably use SMS. it looks less dangerous
Scott Leibrand
@scottleibrand
Mar 28 2016 22:41
with temporary BG targets, we could have a predefined range and only allow temp targets within that range
Jason Calabrese
@jasoncalabrese
Mar 28 2016 22:41
sms saved me today, the verizon data to the phone was lost, but an sms got through and he was able to reboot the phone (after he rebooted the edison)
guess we could do both
I was thinking that if you know the size of the meal coming up we could adjust things more
Dana Lewis
@danamlewis
Mar 28 2016 22:43
that’s not safe enough, given varying carb absorption lags from exercise or delayed meal or ends up being different sized, etc.
Jason Calabrese
@jasoncalabrese
Mar 28 2016 22:45
I think most aggressive would be to target 80, but for a smaller snack you might want to be more conservative
Dana Lewis
@danamlewis
Mar 28 2016 22:49
eating soon is immune to meal size. AMA is where meal size (and knowledge of carbs actually ingested) would matter (and help)