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

11th
Feb 2016
dougtakeuti
@dougtakeuti
Feb 11 2016 00:00
I see an offset on 2 hours when uploading status: However, what I see on NS is correct. Is this a concern?
reporting monitor/upload-status.json
[{"pump":{"battery":{"status":"normal","voltage":1.31},"status":{"status":"normal","timestamp":"2016-02-10T23:56:21.000Z","bolusing":false,"suspended":false},"reservoir":118.3,"clock":"2016-02-10T21:54:52-02:00"},"openaps":{"suggested":{"bg":206,"temp":"absolute","snoozeBG":149,"timestamp":"2016-02-10T23:55:59.000Z","reason":"Eventual BG 149>130, temp 1.625 >~ req 1.13U/hr","eventualBG":149,"tick":-2},"iob":{"timestamp":"2016-02-10T23:5
Jason Calabrese
@jasoncalabrese
Feb 11 2016 00:01
is the timezone on your pi correct
dougtakeuti
@dougtakeuti
Feb 11 2016 00:02
@jasoncalabrese yes it is.
Jason Calabrese
@jasoncalabrese
Feb 11 2016 00:02
there's a mix of zoned and unzoned timestamps there
dougtakeuti
@dougtakeuti
Feb 11 2016 00:02
yep
Jason Calabrese
@jasoncalabrese
Feb 11 2016 00:02
something we should fix, but they work fine
the Z is got utc/zulu time
dougtakeuti
@dougtakeuti
Feb 11 2016 00:04
I'll double check on that then. Thanks.
Jason Calabrese
@jasoncalabrese
Feb 11 2016 00:05
it's nothing you did wrong, for ns-status I took the timestamp of some of the files that gets combined and added them to the json
but didn't zone them
when it gets processed in NS we use moment to parse/convert the dates so it's handled correctly zoned or not
Jim Matheson
@jmatheson
Feb 11 2016 00:07
left the house for a few hours. no openaps again. getting frustrating. this time the G5 app decided to not send BG to NS. no idea why.
guess it’s just flaky
dougtakeuti
@dougtakeuti
Feb 11 2016 00:08
Ok then. Now something else: I'd like to start working with an mmcommander now, but using a separate sd card so I don't risk messing up with a working carelink setup. Is there an easy way to just clone this card?
Jason Calabrese
@jasoncalabrese
Feb 11 2016 00:10
yeah, I thought it was in the docs too, but I can't find it are you using mac? windows?
dougtakeuti
@dougtakeuti
Feb 11 2016 00:10
windows
was sure that was in the docs
Pete Schwamb
@ps2
Feb 11 2016 01:24
Is there a way to get to the dev branch of the openaps docs in gitbooks?
Dana Lewis
@danamlewis
Feb 11 2016 01:26
heh. I've stopped recommending gitbooks for awhile..would recommend just looking on Github at the docs, very easy to switch branches there. (I know, not the question you asked!)
Jason Calabrese
@jasoncalabrese
Feb 11 2016 01:27
dev is the new master
Pete Schwamb
@ps2
Feb 11 2016 01:27
Ok, the gitbooks view is nice, but yeah, I’ll just use dev.
Dana Lewis
@danamlewis
Feb 11 2016 01:27
:thumbsup:
I need some eyeballs to review the NS vis stuff to merge…or enough :+1: to just say forget it & merge even if it's still not right...
*to merge dev to master
Pete Schwamb
@ps2
Feb 11 2016 01:30
The NS vis stuff on master left me a bit confused (couldn’t figure out how to gen clock-zoned.json). It took me a while to figure out that that was already documented in dev.
Dana Lewis
@danamlewis
Feb 11 2016 01:31
have you been working through it @ps2 ? If you think the dev is better than master, take a look at #80 from @logichammer and see what you think? I’d love to fold that in and then merge dev to master...
Pete Schwamb
@ps2
Feb 11 2016 01:31
Not all the way through, but so far seems like dev is better.
Dana Lewis
@danamlewis
Feb 11 2016 01:32
Pete Schwamb
@ps2
Feb 11 2016 01:32
I don’t see #80
Dana Lewis
@danamlewis
Feb 11 2016 01:32
at this point I'd take even a single +1 from someone to get it in, and get dev then merged to master and we can fix from there
^^just posted the link
Pete Schwamb
@ps2
Feb 11 2016 01:35
looking, thanks. :)
Jason Calabrese
@jasoncalabrese
Feb 11 2016 01:37
I got the mmtune PR merged to dev, so when you setup ns-status add 1 more field named mmtune and point to the report
Then you'll get the freq and rssi on the OpenAPS pill in NS
Pete Schwamb
@ps2
Feb 11 2016 01:38
Nice!
Jason Calabrese
@jasoncalabrese
Feb 11 2016 01:40
Screen Shot 2016-02-09 at 10.43.22 PM.png
Pete Schwamb
@ps2
Feb 11 2016 01:41
Was wondering about suggesting echo -n "mynonhashedapisecret" | shasum instead of entering your secret into a third party website
Scott Leibrand
@scottleibrand
Feb 11 2016 01:41
yes, pls PR. :)
Pete Schwamb
@ps2
Feb 11 2016 01:41
Won’t work for everyone. I don’t have a rpi install with me to test if it works on that ubuntu; won’t work on yocto. Works on mac.
Jason Calabrese
@jasoncalabrese
Feb 11 2016 01:42
the ns-upload tool will generate the hash too
Ben West
@bewest
Feb 11 2016 01:46
same thing, yeah
just add --config
highly recommend using the --config mode of these tools
Scott Leibrand
@scottleibrand
Feb 11 2016 01:49
Would be helpful for someone to update the docs to use --config correctly in a way that works with everything else.
dougtakeuti
@dougtakeuti
Feb 11 2016 01:49
Thanks @jasoncalabrese !
Ben West
@bewest
Feb 11 2016 01:51
try it out
it prints a nice help message and will help get your api secret correct
 ns-upload --config aaaa bbb
openaps device add ns-upload process --require "type report" ns-upload "ns-upload-NIGHTSCOUT" "ns-upload-APIKEY"
sed -i -e "s/ns-upload-NIGHTSCOUT/aaaa/g" ns-upload.ini
sed -i -e "s/ns-upload-APIKEY/5cb138284d431abd6a053a56625ec088bfb88912/g" ns-upload.ini
openaps report add  text ns-upload shell "" ""
bewest@bewest-MacBookPro:~/Documents/openaps$
so my plain api secret bbb is 5cb138284d431abd6a053a56625ec088bfb88912
should probably change it to print json to be used by openaps import
Jason Calabrese
@jasoncalabrese
Feb 11 2016 01:53
I think when I was first setting things up it didn't fully work and I just editing the ini
I edit the ini a lot
makes it hard to document
I think thats why the NS vis part of the docs are the way they are, as I helped a few people get it setup I just pointed them to my ini
Ben West
@bewest
Feb 11 2016 01:57
should try doing bunch of show --json commands, export your json
mine isn't particularly elegant either, need to set up a nice one, then simply export it
then create a simple script to replace few of the values
Jason Calabrese
@jasoncalabrese
Feb 11 2016 01:57
some of the aliases are more like recipes, that should be shared
Ben West
@bewest
Feb 11 2016 01:58
yeah
same for getting data cleaned up...
Jason Calabrese
@jasoncalabrese
Feb 11 2016 01:58
using the same names for things will help too
Ben West
@bewest
Feb 11 2016 01:58
if getting data from medtronic, it needs zoning,, units etc
now we can use json to standardize/templates profiles of reports/loops
like just choose the shape
Jason Calabrese
@jasoncalabrese
Feb 11 2016 02:02
figured I'd get it cleaned up when we move to a web app for config
Jason Calabrese
@jasoncalabrese
Feb 11 2016 03:09
some how SAGE had the wrong sensor start time, so I missed getting a 2nd receiver going now, I have a 2 hours without a loop
need to start using raw asap
Pete Schwamb
@ps2
Feb 11 2016 03:46
Most of the typo/fixup changes in @logichammer’s PR look ok. The changes to visualization.md seem to be out of sync with dev, and are referencing aliases not mentioned elsewhere in the docs. I think I prefer the dev version of that file. I can’t comment on the changes to reference-design.md
But dev is definitely better than master, imo.
for the vis section
Dana Lewis
@danamlewis
Feb 11 2016 03:58
Ok, thanks for reviewing that @ps2 and providing feedback
Pete Schwamb
@ps2
Feb 11 2016 04:00
Overall, the docs are working well for me. There were a few things left out (battery.json, reservoir.json & status.json), but those are easy and give the user something to figure out on their own.
Dev cycle on this is nice. edison + rl are upstairs in the kids room, and I’m ssh’d in. It’s a nice break from working on app-based code.
Ali Mazaheri
@amazaheri
Feb 11 2016 04:06
:+1:
Matthias Granberry
@mgranberry
Feb 11 2016 04:08
@jasoncalabrese there are session start/end records in the receiver logs.
I need to write the Android BLE part of my uploader so I can have it do daily metadata dumps. Maybe I could tie it in to xDrip after NC stretches and make the start records automatic there without the battery hit.
Jason Calabrese
@jasoncalabrese
Feb 11 2016 04:37
That would be nice, manual entry sucks
Working on adding support for raw, most of it is really simple, but there should be some limits and not sure what those are
Like never let raw go over 250? 200?
Limit deltas?
Mainly want some help avoiding lows when Dexcom is in the way
Pete Schwamb
@ps2
Feb 11 2016 04:41
Is there any way to get profile data uploaded from pump/openaps into NS?
Jason Calabrese
@jasoncalabrese
Feb 11 2016 04:42
Yeah, there's a tool for it, but I never got around to setting it up
Was swaping pumping at first too
Pete Schwamb
@ps2
Feb 11 2016 04:43
I’m not swapping (yet), just lazy. :)
Ben West
@bewest
Feb 11 2016 04:45
yes
mm-format-ns-profile
and friends
mm-format-ns-*
SETTINGS=${1-monitor/settings.json}
CARBS=${2-monitor/carb-ratios.json}
BASALRATES=${3-monitor/active-basal-profile.json}
SENSITIVITIES=${4-monitor/insulin-sensitivities.json}
TARGETS=${5-monitor/bg-targets.json}
OUTPUT=${6-settings/ns-profile.json}
you can ns-upload the result to NS profile.json
Jason Calabrese
@jasoncalabrese
Feb 11 2016 05:03
It's it creating a new profile each time or doing an update
There's a bunch of settings around profiles
To be able to look at it historically
Jim Matheson
@jmatheson
Feb 11 2016 05:05
damn, can’t ssh into my edison again. still online and running though.
Dana Lewis
@danamlewis
Feb 11 2016 05:08
ok. merged dev to master so everyone can more easily get to the better version of visualization page (https://github.com/openaps/docs/blob/master/docs/Automate-system/vizualization.md), although please note we know it still needs improvements, so please continue to PR fixes into it if you've found gaps in the docs around that area.
I had a thought…toying with the idea of forking the docs, stubbing out the new pages based on the outline we've been discussing, then giving perms to everyone here to contribute directly to that fork (aka merge your own PRs) for while I'm offline next week, and see what can get done with everyone putting in quick and dirty versions that's better than the stubs, doesn't have to be perfect.
Then, when I get back online, we can do a PR into a dev fork on openaps docs.
But that might help for anyone who wants to give contributing to the docs a try, in a low risk environment. Really can't hurt anything on the fork. (Same is true for openaps forks, but hopefully people would be even more comfortable filling in blank pages on my fork?)
Jim Matheson
@jmatheson
Feb 11 2016 05:18
Huge help adding visualization! i remember getting prett stuck at that point.
Dana Lewis
@danamlewis
Feb 11 2016 05:18
Yea. Needs more work though, people are still struggling through that process, so please review with fresh eyes and keep adding to it :)
Jim Matheson
@jmatheson
Feb 11 2016 05:18
i like the fork idea..
Dana Lewis
@danamlewis
Feb 11 2016 05:19
:fork_and_knife: cool
Pete Schwamb
@ps2
Feb 11 2016 05:19
Thanks @danamlewis for getting that merged. Will certainly help others.
Jason Calabrese
@jasoncalabrese
Feb 11 2016 05:24
@danamlewis, that sounds like the xdrip-e repo, should ask them how it's worked out
Stephan
@MosiGitHub
Feb 11 2016 05:29
Tnx @danamlewis - this part is really important for me and I was wondering how to set it up . That helped me a lot.
Dana Lewis
@danamlewis
Feb 11 2016 05:30
@MosiGitHub glad to hear! would love feedback on whether you got all the way through to being set up with visualizations on NS with the revised docs, or where you're still running into problems.
I'm not sure what % rate of success people are having getting their visualizations up, so would love to know from folks who made it through as well as the areas that are major blockers to improve
Stephan
@MosiGitHub
Feb 11 2016 05:31
I will definitely let you know...
Dana Lewis
@danamlewis
Feb 11 2016 05:32
:+1:
Stephan
@MosiGitHub
Feb 11 2016 05:34
I am using NS quite often and it is so helpful that I don't want to miss it. Careportal and basal are already implemented . So my concern was , when I change the setup for openaps to be without NS for a while
Jason Calabrese
@jasoncalabrese
Feb 11 2016 05:36
Just setup 2 NS sites
They can share the same db
marius a. eriksen
@mariusae
Feb 11 2016 05:38
@scottleibrand @jasoncalabrese where did you guys get the activity/iob curves in oref0?
Jason Calabrese
@jasoncalabrese
Feb 11 2016 05:39
Think @scottleibrand came up with it using excel
Stephan
@MosiGitHub
Feb 11 2016 05:41
Could be an idea... My concern was if I flash the mmcommander with the subg_rfspy firmware this won't work with the uploader. Maybe I take an old firmware of the Mmcommander with Tx enabled
So I can switch the hardware
marius a. eriksen
@mariusae
Feb 11 2016 05:42
@jasoncalabrese cool thanks. i’ve pretty much been doing the same by eyeballing the humalog charts
Dana Lewis
@danamlewis
Feb 11 2016 05:43
@mariusae yes. reverse engineering with handy dandy excel
marius a. eriksen
@mariusae
Feb 11 2016 05:43
:+1:
(eqn 18)
the curves look the same :) i wonder if they also just eyeballed things
Dana Lewis
@danamlewis
Feb 11 2016 05:45
phew. the equations on page 2 already look way more complicated than necessary ;)
marius a. eriksen
@mariusae
Feb 11 2016 05:46
@danamlewis so it looks like you used excel to do a second degree polyonmial regression on the data before the peak, and another on the data after?
Jim Matheson
@jmatheson
Feb 11 2016 05:46
that guy on page 9 has an original nintendo controller attached to his arm
marius a. eriksen
@mariusae
Feb 11 2016 05:48
@danamlewis and, do you recall, was this from Humalog or Novolog? (or some combination?)
Dana Lewis
@danamlewis
Feb 11 2016 05:49
@mariusae no regression. We assumed curve should be 0 at 0 minutes; peak at some number (75 minutes because my DIA was 3, 90 when I was DIA of 4), and then 0 by end DIA.
I use Humalog
Jim Matheson
@jmatheson
Feb 11 2016 05:51
are there published curve differences between humalog vs novalog?
Scott Leibrand
@scottleibrand
Feb 11 2016 05:51
what I actually ended up doing was modeling the insulin activity "curve" as two straight lines, from zero to a peak at 75/90m and back down to zero at DIA. I compared that to the empirically derived curve in some paper (I forget which), and noticed that the error bars were wide enough that those straight lines actually fit inside them. :)
marius a. eriksen
@mariusae
Feb 11 2016 05:52
@scottleibrand but they’re quadradic no?
Scott Leibrand
@scottleibrand
Feb 11 2016 05:52
the excel work was mostly just to integrate that linear insulin activity curve to get two quadratic IOB curves, before and after peak
marius a. eriksen
@mariusae
Feb 11 2016 05:53
oh i see ok, that makes sense.
Scott Leibrand
@scottleibrand
Feb 11 2016 05:53
since I couldn't remember enough calculus from high school to solve the integral symbolically, I did it with Excel curve fit. :)
Dana Lewis
@danamlewis
Feb 11 2016 05:53
simple, but it works.
marius a. eriksen
@mariusae
Feb 11 2016 05:53
hey, that’s the best.
Dana Lewis
@danamlewis
Feb 11 2016 05:54
:+1: :chart_with_upwards_trend: :chart_with_downwards_trend:
Scott Leibrand
@scottleibrand
Feb 11 2016 05:54
lol
Jim Matheson
@jmatheson
Feb 11 2016 05:55
@mariusae have you seen curve differences published between humalog and novalog?
marius a. eriksen
@mariusae
Feb 11 2016 05:55
@jmatheson yeah
@jmatheson novolog peaks a little later
@jmatheson let me see if ic an dig it up
Jim Matheson
@jmatheson
Feb 11 2016 05:55
thanks
Scott Leibrand
@scottleibrand
Feb 11 2016 05:56
the other thing I did to "check" our IOB curve was eyeball it against the ones MDT uses. it looks pretty similar. :)
Jim Matheson
@jmatheson
Feb 11 2016 05:56
i am on novalog. i found that it worked better for me. but it was 10 years ago and i just remember i seemed less resistent.
marius a. eriksen
@mariusae
Feb 11 2016 05:56
well, peak is a little lower, has slightly stronger tail
Screen Shot 2015-09-13 at 10.09.28 PM.png
that’s been my goto humalog curve
marius a. eriksen
@mariusae
Feb 11 2016 06:01
" Fasting insulin lispro levels reached 50% of peak concentration at 20 ± 1 min and aspart at 30 ± 3 min (P = 0.02) (Fig. 2). The decrease of free insulin concentration from peak concentration to 50% of the maximum concentration was found at 113 ± 10 min during insulin lispro and 154 ± 14 min during insulin aspart (P = 0.02) (Fig. 2)"
Jim Matheson
@jmatheson
Feb 11 2016 06:03
good stuff
i dont miss the days of Humalog R
marius a. eriksen
@mariusae
Feb 11 2016 06:04
now if we could only find a way to administer afrezza automatically..
Dana Lewis
@danamlewis
Feb 11 2016 06:04
HA. and measure it's on-board activity...
*and prevent all the brand trolls from joining this conversation, or any other conversation where that product is mentioned
marius a. eriksen
@mariusae
Feb 11 2016 06:05
we’ll just have to come up with a codeword
insulin “A”
Dana Lewis
@danamlewis
Feb 11 2016 06:05
How about "Floo powder" (any harry potter fans here?)
marius a. eriksen
@mariusae
Feb 11 2016 06:05
haha
Jim Matheson
@jmatheson
Feb 11 2016 06:05
i am still waiting for all the #murseless trolls to show up
Dana Lewis
@danamlewis
Feb 11 2016 06:06
@jmatheson I'm very serious about the trolls. be aware if you want to talk about that product in a public forum :(
marius a. eriksen
@mariusae
Feb 11 2016 06:06
what is up with that? is it some sort of gueriilla marketing?
(i have noticed that; especially on facebook)
Jim Matheson
@jmatheson
Feb 11 2016 06:07
as a marketer you want to pay attention to where your brand is being talked about.
so people made tools to help discover that.
marius a. eriksen
@mariusae
Feb 11 2016 06:11
do you guys notice changes in the insulin response curve as pump sites age? i’m pretty sure we see this with our daughter. (even after a single day)
Dana Lewis
@danamlewis
Feb 11 2016 06:11
@jmatheson pharma has hands tied though in some of what they do because of AE reporting, etc. so it's not a given that they'd be watching everything like other brands might, but surely by now they're aware of this type of online activity. anyway. just wanted to point that out if you hadn't seen it so you were aware.
@mariusae not in first days, but definitely day 3-4 very different
marius a. eriksen
@mariusae
Feb 11 2016 06:12
may just be that a 3-year old has a much faster immune response to sites…
Dana Lewis
@danamlewis
Feb 11 2016 06:12
possibly, because less fat & stuff surrounding the site?
marius a. eriksen
@mariusae
Feb 11 2016 06:13
maybe that too—but was just thinking that young kids repair wounds like it’s nothing
Jim Matheson
@jmatheson
Feb 11 2016 06:13
i am not sure how much for me is age vs location/angle.
Dana Lewis
@danamlewis
Feb 11 2016 06:13
@mariusae good point.
Jim Matheson
@jmatheson
Feb 11 2016 06:14
interesting..
marius a. eriksen
@mariusae
Feb 11 2016 06:14
which, oh! i also learned something interesting about dexcom. probably you guys already knew this.
Dana Lewis
@danamlewis
Feb 11 2016 06:14
@jmatheson I notice for location, too, some difference. but even a good site starts to look like a less-good site after day 3 or so
marius a. eriksen
@mariusae
Feb 11 2016 06:14
but the reason there’s a 3 hour warmup is that your body has an inflammatory response to insertion
which can affect readings
Jim Matheson
@jmatheson
Feb 11 2016 06:15
yeah, also when my site stars falling out…no so good :)
Dana Lewis
@danamlewis
Feb 11 2016 06:15
@mariusae have you tried inserting a sensor, putting an old transmitter on it for a day, and thus waiting for a full day before putting the good transmitter on and calibrating?
I've started doing that, and the day of "pre-soak" before starting it smooths out a lot of the first-day jumpy readings
marius a. eriksen
@mariusae
Feb 11 2016 06:15
@danamlewis i have not, but i saw @scottleibrand discuss this technique in slack.. intriguing.
Jim Matheson
@jmatheson
Feb 11 2016 06:16
by old transmitter you mean one that is dead?
Dana Lewis
@danamlewis
Feb 11 2016 06:16
doesn't always work (if you rip out a sensor by accident), but if you're planning to change a sensor, worth a try to see if it helps. some people don't see first-day jumps, but makes quite a difference for me.
@jmatheson yea. dead battery transmitter that you put on it just to protect the sensor
very easy to pop it off and put the working one on when you're ready to swap it.
(and then once you do that, obviously pull out the old sensor)
Jim Matheson
@jmatheson
Feb 11 2016 06:17
thanks, worth noting. my only old transmitter i have left is busted in 10 pieces after i tried to open it to replace tha battery :)
Dana Lewis
@danamlewis
Feb 11 2016 06:17
lol
yea. don't use that one then ;)
Jim Matheson
@jmatheson
Feb 11 2016 06:17
but i will have a new old one in a few months
Dana Lewis
@danamlewis
Feb 11 2016 06:17
:+1:
Scott Leibrand
@scottleibrand
Feb 11 2016 06:17
for awhile we had two transmitters, which was even better: we could see which one was less noisy and decide when to switch. but if you don't want to burn two transmitters' batteries at once, this method works well too. way better than using a brand new sensor to make insulin dosing decisions. (oref0 does pretty well, but manual dosing decisions are challenging on a jumpy sensor)
Jim Matheson
@jmatheson
Feb 11 2016 06:18
i was pretty frustrated waiting for dexcom to send me a new one so figured had nothing to lose
Scott Leibrand
@scottleibrand
Feb 11 2016 06:20
the other thing we often do is only restart one of our two receivers when we first swap the transmitter to the newly-soaked sensor. the non-restarted one usually picks up after about an hour of ???, and is often fairly accurate when the 2h is up and we calibrate the restarted one. at that point we can restart the one we left running, to purge the "history" from the old sensor's calibrations without having a 2h gap.
marius a. eriksen
@mariusae
Feb 11 2016 06:20
so i guess you guys are on the G4?
Dana Lewis
@danamlewis
Feb 11 2016 06:20
^that means there's only about one hour without data, which is better than two
yea. G4 with share receivers
Scott Leibrand
@scottleibrand
Feb 11 2016 06:20
that also helps a week later when we want to have one sensor still running when we restart the other at the end of its 7d life
marius a. eriksen
@mariusae
Feb 11 2016 06:21
i wonder how long they’ll keep the G4 around ...
Scott Leibrand
@scottleibrand
Feb 11 2016 06:21
quite a while. all the integrated pumps use it.
marius a. eriksen
@mariusae
Feb 11 2016 06:21
ahh
Dana Lewis
@danamlewis
Feb 11 2016 06:21
it's integrated with many pumps, so it'll be a while
marius a. eriksen
@mariusae
Feb 11 2016 06:21
cool
Dana Lewis
@danamlewis
Feb 11 2016 06:21
we'll just need to apply some elbow grease to get it ordered vs being upsold to g5
marius a. eriksen
@mariusae
Feb 11 2016 06:21
yeah, they automatically upgraded us
(or maybe they asked? i’m not sure.)
Dana Lewis
@danamlewis
Feb 11 2016 06:21
heh
Scott Leibrand
@scottleibrand
Feb 11 2016 06:22
at some point I think the G5 (with custom BLE integration with OpenAPS and other similar apps) may be better than the G4, but not until we build some tools to fill the gaps left by Dexcom's app
Dana Lewis
@danamlewis
Feb 11 2016 06:22
I also hope they realize the limitations of g5 and won't push too hard, given the integration of the g4s, and the g6 comes out sooner and will hopefully be better than g5 and worth switching to
*I, as the patient, am not convinced the G5 is better.
Jim Matheson
@jmatheson
Feb 11 2016 06:22
alright, you guys talked me into opening my G5 receiver.
Dana Lewis
@danamlewis
Feb 11 2016 06:23
seen too many examples of the g5 not getting data to a phone that's right next to the person :(
marius a. eriksen
@mariusae
Feb 11 2016 06:23
yeah, this is now the #1 reason we have to wake up at night
Jim Matheson
@jmatheson
Feb 11 2016 06:23
mine is a little spotty but their app getting the data to share is 10x spottier. speaking of 10x, the box is 10x the size it needs to be to fit this receiver in it.
Dana Lewis
@danamlewis
Feb 11 2016 06:23
lol @jmatheson
Jim Matheson
@jmatheson
Feb 11 2016 06:24
nice! free extra micro usb cord. already more value than i expected
Dana Lewis
@danamlewis
Feb 11 2016 06:24
as @scottleibrand was saying, if others build on the good work @loudnate is doing to read directly from the transmitter vs. us having to rely on the Dex app, I think that will work
marius a. eriksen
@mariusae
Feb 11 2016 06:24
yeah, this is on my todo list for tinyAP
Dana Lewis
@danamlewis
Feb 11 2016 06:24
:)
I love that name, btw
marius a. eriksen
@mariusae
Feb 11 2016 06:25
thanks :) i was trying to figure out what its distinguishing feature was… (or at least, what its goals are); and that about nails it
Jim Matheson
@jmatheson
Feb 11 2016 06:25
sounds like i have some competition
marius a. eriksen
@mariusae
Feb 11 2016 06:26
mine is currently 0.97x1.89 inch (24.71x48.01 mm)
excluding batteries!
Dana Lewis
@danamlewis
Feb 11 2016 06:26
I bet @mariusae's 3yo wouldn't mind a purse/murse
Jim Matheson
@jmatheson
Feb 11 2016 06:26
i meant for naming…but oh….it’s on!
Dana Lewis
@danamlewis
Feb 11 2016 06:27
LOL
marius a. eriksen
@mariusae
Feb 11 2016 06:27
haha
Jim Matheson
@jmatheson
Feb 11 2016 06:27
what about the 3rd dimension?
marius a. eriksen
@mariusae
Feb 11 2016 06:27
@danamlewis oh man, she absolutely loves anhythiung to do with purses. but we mostly use sewed-in pockets on an undershirt. works great.
Dana Lewis
@danamlewis
Feb 11 2016 06:28
:)
marius a. eriksen
@mariusae
Feb 11 2016 06:28
@jmatheson depends on battery size… but thin. basically PCB+battery
Jim Matheson
@jmatheson
Feb 11 2016 06:28
you gotta include batteries. cheating without them.
marius a. eriksen
@mariusae
Feb 11 2016 06:29
true that
Jim Matheson
@jmatheson
Feb 11 2016 06:29
i think you are ahead of me though. i still have a usb TI stick
IMG_2622.jpg
Scott Leibrand
@scottleibrand
Feb 11 2016 06:31
1"x2" is a pretty good size. that's what the RL is, and Edison with SparkFun board. Intel board is slightly longer, but no big deal. I think biggest differentiator at that size is thickness.
edison+battery+RL is about 1" thick
Jim Matheson
@jmatheson
Feb 11 2016 06:31
yeah. this is where i need to think through things before i make any jokes.
Scott Leibrand
@scottleibrand
Feb 11 2016 06:32
:-p
Jim Matheson
@jmatheson
Feb 11 2016 06:33
i think i am going to prefer the Intel board. Thinner is going to be better i think. will see in a few days
marius a. eriksen
@mariusae
Feb 11 2016 06:33
yeah, the edison form factor makes me re-think going fully embedded a little.
Jason Calabrese
@jasoncalabrese
Feb 11 2016 06:34
first pass at adding support for using raw openaps/oref0#69
Scott Leibrand
@scottleibrand
Feb 11 2016 06:34
edison is an awesome mix of great form factor with full ability to tweak software really easily (even from my phone) :-)
Jason Calabrese
@jasoncalabrese
Feb 11 2016 06:34
curious what everyone will think of it, the idea is you'd run an other report and then feed the enhanced glucose data to determine basal as normal
Pete Schwamb
@ps2
Feb 11 2016 06:35
I’m loving working on the edison. :)
marius a. eriksen
@mariusae
Feb 11 2016 06:35
what battery sizes do you have to work with in practice? for all-day operations?
Pete Schwamb
@ps2
Feb 11 2016 06:36
heh, I’m not mobile with it yet. It’s on a 25000 mah battery. Could probably run for a few months. ;)
Jim Matheson
@jmatheson
Feb 11 2016 06:36
i think 2000mAh
marius a. eriksen
@mariusae
Feb 11 2016 06:36
oh that’s not bad.
Jason Calabrese
@jasoncalabrese
Feb 11 2016 06:36
an 850 gets through the school day
with the RL and edison
Jim Matheson
@jmatheson
Feb 11 2016 06:37
my battery is toast so waiting for a new 2k before confirming that
@mariusae are you on a TI stick or ERF?
i have battery envy
marius a. eriksen
@mariusae
Feb 11 2016 06:37
@jmatheson custom board with cc1110 and an nRF51 (now through the Simblee module)
Scott Leibrand
@scottleibrand
Feb 11 2016 06:39
we got 8.5h with our 850mAh battery. upgraded to a 2000mAh one, and haven't tested its limits yet.
Jim Matheson
@jmatheson
Feb 11 2016 06:39
ok, i am going to concede my loss now
Pete Schwamb
@ps2
Feb 11 2016 06:40
I see you switched to the johanson balun? Good choice.
marius a. eriksen
@mariusae
Feb 11 2016 06:40
yeah
i also went 24mhz after our last conversation :)
Pete Schwamb
@ps2
Feb 11 2016 06:40
:+1:
Jason Calabrese
@jasoncalabrese
Feb 11 2016 06:40
what the size difference between 850 and 2000, any side by side pictures?
Scott Leibrand
@scottleibrand
Feb 11 2016 06:41
we were using a supposedly 2000mAh one from the PS4 controller, but didn't find it lasted any longer than the smaller 850. the 2000mAh one we're running now is a proper silver-and-yellow one whose rating looks more legitimate
@jasoncalabrese lemme snap one
marius a. eriksen
@mariusae
Feb 11 2016 06:41
@ps2 i think i can probably get rid of most of the decoupling caps on the cc1110. i used the reference design but it seems way overkill. i’m going to try a board and only populate 4 of the caps. might save some space/complexity.
also i’m going to get rid of that pesky jst connector and just solder the battery directly to teh board
Jim Matheson
@jmatheson
Feb 11 2016 06:42
@mariusae are you in SF?
Pete Schwamb
@ps2
Feb 11 2016 06:43
Yeah, that thought had occurred to me as well (about the caps); You can only make them so close, and going a little bigger and sharing I can’t see how that could not be sufficient.
marius a. eriksen
@mariusae
Feb 11 2016 06:43
@jmatheson yep. most days. though i live in Menlo Park.
@ps2 if you peruse the TI forums, there are huge flamewars about this very topic :-)
Jim Matheson
@jmatheson
Feb 11 2016 06:43
nice. would love to catch up at the next meetup
marius a. eriksen
@mariusae
Feb 11 2016 06:43
@jmatheson :thumbsup:
Pete Schwamb
@ps2
Feb 11 2016 06:44
I do use the jst connector, though, changing batteries is easy that way.
Jason Calabrese
@jasoncalabrese
Feb 11 2016 06:44
thanks @scottleibrand
still small, but I might need to find a different case
hardly sticks out behind the Edison though
Jason Calabrese
@jasoncalabrese
Feb 11 2016 06:45
@sulkaharo is going to be in SF for a conf, I'm thinking about going up for a night (from SD) maybe a meetup on March 13th?
marius a. eriksen
@mariusae
Feb 11 2016 06:45
sgtm!
Jason Calabrese
@jasoncalabrese
Feb 11 2016 06:46
oh, maybe it will fit in my case, ordering anyway
Jim Matheson
@jmatheson
Feb 11 2016 06:47
i am generally available for meetups (barring powder days).
Scott Leibrand
@scottleibrand
Feb 11 2016 06:47
@jasoncalabrese if EGV is 140 but raw shows 135, will it change glucose to be 135?
Jim Matheson
@jmatheson
Feb 11 2016 06:47
we had a solid crew at the last one
Scott Leibrand
@scottleibrand
Feb 11 2016 06:49
Jason Calabrese
@jasoncalabrese
Feb 11 2016 06:49
should only set glucose if < 40 or missing and < maxRaw
Scott Leibrand
@scottleibrand
Feb 11 2016 06:50
yeah, question is where the () go in that statement. :)
should only set glucose if (egv < 40 or missing) and (raw < maxRaw)
Jason Calabrese
@jasoncalabrese
Feb 11 2016 06:50
if (entry.raw && (entry.glucose < 40 || !entry.glucose) && entry.raw < maxRaw)
Scott Leibrand
@scottleibrand
Feb 11 2016 06:51
yeah, that's what I found. code is unambiguous, just the english I was confused about at first.
Jason Calabrese
@jasoncalabrese
Feb 11 2016 06:51
yeah, thats why I added some tests
Scott Leibrand
@scottleibrand
Feb 11 2016 06:51
I agree those are good conditions
Jason Calabrese
@jasoncalabrese
Feb 11 2016 06:51
does maxRaw seem like the right place to limit?
Scott Leibrand
@scottleibrand
Feb 11 2016 06:51
so if raw > 150 it just won't add a glucose at all, right?
Jason Calabrese
@jasoncalabrese
Feb 11 2016 06:52
right
Scott Leibrand
@scottleibrand
Feb 11 2016 06:52
it doesn't trim it to 150
Jason Calabrese
@jasoncalabrese
Feb 11 2016 06:52
no
just like now
so for the case where waters get in between sensor and contacts we wouldn't high temp some crazy amount
Scott Leibrand
@scottleibrand
Feb 11 2016 06:53
still thinking through what we'd want to do with noisy deltas
I haven't looked at raw in awhile, but I wonder if 15m avgdelta would be sufficient to deal with it
Jason Calabrese
@jasoncalabrese
Feb 11 2016 06:53
yeah, we don't need to rush it
I was thinking of some max delta
Scott Leibrand
@scottleibrand
Feb 11 2016 06:54
or we could use a 30m avgdelta as well
Jason Calabrese
@jasoncalabrese
Feb 11 2016 06:54
or skipping the previous like we did in NS for awhile
I wa thinking about only setting the most recent glucose
so delta would be 0
Scott Leibrand
@scottleibrand
Feb 11 2016 06:55
if the 15m avgdelta is greater than the 30m avgdelta (indicating 30m-ago data point is in between 15m ago and now) then use the 30m-ago delta as if it were 15m ago
the challenge with delta-0 is you lose a lot of the safety checks around canceling/reducing high-temps when falling etc.
Jason Calabrese
@jasoncalabrese
Feb 11 2016 06:56
that would make sense, I set a flag fromRaw so we could do that in glucoseGetLast
most of the time raw is good too
might be something we don't really put in the main docs
Scott Leibrand
@scottleibrand
Feb 11 2016 06:58
I'm trying to think through the scenarios where having raw data would allow us to make better decisions. what are yours?
Jason Calabrese
@jasoncalabrese
Feb 11 2016 06:58
going low in 2 hour warmup
or ??? at night
when that happens we're now offline
and even worse don't cancel a high temp
Scott Leibrand
@scottleibrand
Feb 11 2016 06:59
We don't get ??? much.
Did we ever add the "cancel high temp on missing data" thing?
Jason Calabrese
@jasoncalabrese
Feb 11 2016 07:00
doesn't look like it
Scott Leibrand
@scottleibrand
Feb 11 2016 07:00
see also strategy above for turning 2h warmup into 1h :-)
would be interesting to see how raw looks in that first hour on a pre-soaked sensor.
Can you open an issue for cancel on missing?
Jason Calabrese
@jasoncalabrese
Feb 11 2016 07:01
I usually get a 2nd sensor going before or swap receivers
Scott Leibrand
@scottleibrand
Feb 11 2016 07:02
need to check if I fixed that in meal-assist branch too
ya, we presoak new sensor for 24h before swapping the transmitter to it.
Jason Calabrese
@jasoncalabrese
Feb 11 2016 07:03
openaps/oref0#70
Scott Leibrand
@scottleibrand
Feb 11 2016 07:03
Thx
Jason Calabrese
@jasoncalabrese
Feb 11 2016 07:03
I forgot to enter the last sensor change, and did it a couple days later, and must have picked the wrong day
thought we had till tomorrow
was going to put a new sensor on, but no one heard the dex beeps, was a surprise
shouldn't happen with all this montioring
Scott Leibrand
@scottleibrand
Feb 11 2016 07:05
Our receivers are always staggered, so we still have one going when the first dies.
Jason Calabrese
@jasoncalabrese
Feb 11 2016 07:05
Screen Shot 2016-02-10 at 11.05.06 PM.png
but worked out ok
still the thought of being down was enough to get me writing some code for next time
Scott Leibrand
@scottleibrand
Feb 11 2016 07:06
:-)
Jason Calabrese
@jasoncalabrese
Feb 11 2016 07:06
will help with a wixel too
Scott Leibrand
@scottleibrand
Feb 11 2016 07:07
i almost always change something during 2h restarts. Usually related to where it pulls data from and how it fails over when data is missing.
Jason Calabrese
@jasoncalabrese
Feb 11 2016 07:07
I'd like to do the golden receiver approach
where we calibrate with dex morning/night and use those with the raw
Scott Leibrand
@scottleibrand
Feb 11 2016 07:07
NS vs. openxshareble vs. wired CGM
wonder what we should do when we have EGV from one receiver but raw from the other.
Jason Calabrese
@jasoncalabrese
Feb 11 2016 07:08
could still use the dex, but edison would work without it
evg first I think
Scott Leibrand
@scottleibrand
Feb 11 2016 07:08
That failover logic gets complicated.
Jason Calabrese
@jasoncalabrese
Feb 11 2016 07:09
need to consider time too
old NS data shouldn't block new raw
Scott Leibrand
@scottleibrand
Feb 11 2016 07:09
problem is it's done earlier, at the data gathering stage, before you know what the data says.
Jason Calabrese
@jasoncalabrese
Feb 11 2016 07:09
need a merge
Scott Leibrand
@scottleibrand
Feb 11 2016 07:09
Last restart I added some js one-liner stuff to fail over on old NS data
Jason Calabrese
@jasoncalabrese
Feb 11 2016 07:10
I've removed get get-bg
edit file to keep it going
just since I could and wasn't much time left
Scott Leibrand
@scottleibrand
Feb 11 2016 07:11
This is all new stuff, no changes to existing, right?
Jason Calabrese
@jasoncalabrese
Feb 11 2016 07:11
a little refactoring
I haven't tested it for real yet
wanted to see what everyone thought of the limits
Scott Leibrand
@scottleibrand
Feb 11 2016 07:12
k, you planning to test in wip or dev?
seem reasonable for now.
Jason Calabrese
@jasoncalabrese
Feb 11 2016 07:12
since it takes setup to use could test in dev
Scott Leibrand
@scottleibrand
Feb 11 2016 07:13
K. Merge?
Jason Calabrese
@jasoncalabrese
Feb 11 2016 07:13
sure
I'll get it setup tomorrow, need to move stuff to a replacement phone before I get charged for not returning the old one
such a pain moving google authenticator
Scott Leibrand
@scottleibrand
Feb 11 2016 07:16
Sometimes a backup will move that, at least on iOS.
Jason Calabrese
@jasoncalabrese
Feb 11 2016 07:19
not android, at least haven't see it
I know I could do other low level backups
but seems like a pain too
Aaron Michelson
@Aaybob
Feb 11 2016 15:17
@danamlewis Yes? how did you format logs and get date/time?

My log looks like this: And it runs if I keep carelink dongle in my pocket next to pump on a ten foot extension.

PREFLIGHT OK
GATHER OK
{"temp":"absolute","bg":91,"tick":-1,"eventualBG":89,"snoozeBG":89,"reason":"Eve
ntual BG 89<90, no temp, setting -0.48U/hr","duration":30,"rate":0} pumpbuilding://JSON/set_temp_basal/enact/enacted.json reporting enact/enacted.json STATUS UPLOADED PREFLIGHT OK GATHER OK {"temp":"absolute","bg":94,"tick":"+3","eventualBG":97,"snoozeBG":97,"reason":"E ventual BG 97>90 but Avg. Delta -0.67 < Exp. Delta 0.2; cancel","duration":0,"ra
te":0}
pumpbuilding://JSON/set_temp_basal/enact/enacted.json
reporting enact/enacted.json
STATUS UPLOADED
PREFLIGHT OK
GATHER OK
{"temp":"absolute","bg":95,"tick":"+1","eventualBG":101,"snoozeBG":101,"reason":
"101 is in range. No temp required"}
No Action Required
STATUS UPLOADED
PREFLIGHT OK
GATHER OK
{"temp":"absolute","bg":96,"tick":"+1","eventualBG":104,"snoozeBG":104,"reason":
"104 is in range. No temp required"}
No Action Required
STATUS UPLOADED
PREFLIGHT OK
GATHER OK
{"temp":"absolute","bg":97,"tick":"+1","eventualBG":103,"snoozeBG":103,"reason":
"103 is in range. No temp required"}
No Action Required
STATUS UPLOADED
PREFLIGHT OK
GATHER OK
{"temp":"absolute","bg":98,"tick":"+1","eventualBG":104,"snoozeBG":104,"reason":
"104 is in range. No temp required"}
No Action Required
STATUS UPLOADED

Aaron Michelson
@Aaybob
Feb 11 2016 15:58
Today is first pacific crest trail hike with AP. :-)
Scott Leibrand
@scottleibrand
Feb 11 2016 16:10
Wow, nice. PCT already in Feb?
marius a. eriksen
@mariusae
Feb 11 2016 16:20
woah, jelly.
Dana Lewis
@danamlewis
Feb 11 2016 16:21
@Aaybob to answer your question as before…I was pulling data from DIYPS, it's not pulling from OpenAPS, so I don't have a how-to in order to do that from the OpenAPS formatted data. It's something I hope we'll build into Nightscout reports, and I'm happy to explain the idea for someone to go implement more quickly than I may get to it.
also @Aaybob re: hiking, if you haven't, you might want to consider upping your targets on your pump while you're active, rather than using your tight overnight targets.
Scott Leibrand
@scottleibrand
Feb 11 2016 16:28
@Aaybob if you just want to see a timestamp in your logs, just add a date command somewhere to one of your aliases or to the cron job itself.
Jason Calabrese
@jasoncalabrese
Feb 11 2016 16:40
or use logger, and send it to syslog
I just add something like 2>&1 | logger -t openaps-loop to the end of the line in the crontab
Scott Leibrand
@scottleibrand
Feb 11 2016 16:46
:+1:
Jason Calabrese
@jasoncalabrese
Feb 11 2016 16:51
also for logging I edit the log rotate confs
Scott Leibrand
@scottleibrand
Feb 11 2016 16:56
Yeah, my repo has some logrotate conf files. Not sure anything installs them automatically though.
Jason Calabrese
@jasoncalabrese
Feb 11 2016 17:03
are we avoiding adding that to the docs? could have enough info to get started at https://github.com/openaps/docs/blob/dev/docs/Automate-system/create-schedule.md
an example cron would save people some time
Dana Lewis
@danamlewis
Feb 11 2016 17:09
we were specifically avoiding that, because the cron process was where people were learning the most about how the system works
we're leaning away from that in the future, but it needs to be done in concert with provdiing more stepping stones in a different way of the process.
new way to walk people through the overall setup process, more clear timeline and goals at each step. but then could include more examples of cron (or daemon if we end up going that route)
Jason Calabrese
@jasoncalabrese
Feb 11 2016 17:11
if there's some line of what we don't want in the docs we should put that in the docs
Dana Lewis
@danamlewis
Feb 11 2016 17:11
so it was not about saving time, it was about making sure people understand how the process works and what the key components are. at the time we did the docs, we decided it was safer for people to start from scratch than having an example cron.
Jason Calabrese
@jasoncalabrese
Feb 11 2016 17:12
a little section called something like "Things you won't find in these docs"
Dana Lewis
@danamlewis
Feb 11 2016 17:12
you should comment on the commit thing above with that idea ;)
(please. I'm going to use all the comments there to stub out the new docs outline, if it's not included there there's a chance I might accidentally forget it)
Jason Calabrese
@jasoncalabrese
Feb 11 2016 17:13
added 1
Dana Lewis
@danamlewis
Feb 11 2016 17:14
:+1:
Thx
Matthias
@ceben80
Feb 11 2016 19:00
based to this passage here I would like to add the export lines, but I am unable to find the .profile in the pi home directory, please advice what to do, thanks
a.JPG
Scott Leibrand
@scottleibrand
Feb 11 2016 19:01
You can create one
Matthias
@ceben80
Feb 11 2016 19:02
empty file and simply paste the export lines?
nothing more?
Scott Leibrand
@scottleibrand
Feb 11 2016 19:03
Correct. Google for "bash .profile" if you want to learn more.
Matthias
@ceben80
Feb 11 2016 19:04
great Scott, thx :+1:
Matthias
@ceben80
Feb 11 2016 19:12
the exact name is ".bash_profile"(with an underscore) - correct?
JaysonEwer
@JaysonEwer
Feb 11 2016 19:14
ls -alh ~ look for a .profile file
It should be there, you may not be looking in your home directory
If it's not as Great Scott has already said you may create one.
Jim Matheson
@jmatheson
Feb 11 2016 19:17
i am making some edits to the instructions to make it more clear.
Matthias
@ceben80
Feb 11 2016 19:17
a.JPG
JaysonEwer
@JaysonEwer
Feb 11 2016 19:18
Yep, NOT THERE. 😀
vi .profile if your adventurous or nano .profile if not.
Matthias
@ceben80
Feb 11 2016 19:20
jupp, thanks Jayson!
@scottleibrand @danamlewis - should be modified in openAPS doc that this file does not exist per default
Dana Lewis
@danamlewis
Feb 11 2016 19:23
@ceben80 would love to review a PR if you want to put one in to update the docs :)
Jim Matheson
@jmatheson
Feb 11 2016 19:24
@ceben80 @JaysonEwer is this more clear? if so I can add a PR to get it updated. I also included the suggestion @ps2 made yesterday to create your hash locally. https://github.com/jmatheson/docs/blob/dev/docs/Automate-system/vizualization.md
Matthias
@ceben80
Feb 11 2016 19:26
nano .profile shows....
a.JPG
Interesting. I would have thought, it is an empty file
Jim Matheson
@jmatheson
Feb 11 2016 19:27
just add the lines below all of that
Matthias
@ceben80
Feb 11 2016 19:27
jupp
JaysonEwer
@JaysonEwer
Feb 11 2016 19:28
I ended up adding the lines to my crontab which did the trick, but I also didn't have the quotes included.
Jim Matheson
@jmatheson
Feb 11 2016 19:35
the instructions actually say to add to your crontab at the bottom, but that is not required if we just have people add them to the .profile, right? I think env var is best, no need to confuse people with different options IMHO
Scott Leibrand
@scottleibrand
Feb 11 2016 19:38
I don't think cron uses .profile
Jim Matheson
@jmatheson
Feb 11 2016 19:41
i dont have anything set in my cron
oh, i forgot. yea…in my cron i make sure to reference the .profile first
Scott Leibrand
@scottleibrand
Feb 11 2016 19:42
ah, yeah, that works
Jim Matheson
@jmatheson
Feb 11 2016 19:42
but we dont tell anyone to do that anywhere
so will just leave that in there.
Matthias
@ceben80
Feb 11 2016 20:07
@jmatheson great - will read "vizualization.md" tomorrow
Jim Matheson
@jmatheson
Feb 11 2016 20:09
if anyone is in need of a carelink let me know. i have an extra i won’t be using.
Jim Matheson
@jmatheson
Feb 11 2016 23:34
how are folks using oref0-reset-git in their cron? checking for a certain fail message and resetting ?
Ben West
@bewest
Feb 11 2016 23:36
muahahahaha