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

4th
Jan 2016
Scott Leibrand
@scottleibrand
Jan 04 2016 00:15
would love to see someone figure out how to use that effectively with the Pi for openaps and document it
Andrew Warrington
@andrew-warrington
Jan 04 2016 00:26
I can take that up. I'm running it in my loop already, but shutting off all ports at once. in a couple of days will work on shutting off individual ports and will document as I go.
Dana Lewis
@danamlewis
Jan 04 2016 00:26
:+1:
Andrew Warrington
@andrew-warrington
Jan 04 2016 00:28
On my RPi2 compact setup, which uses a 2500 mAh batt, I initially got 5h15 minutes on a charge.
I have now implemented two power management schemes:
1- underclock the chip between loops, down to 250 mhz
Ben West
@bewest
Jan 04 2016 00:29
interesting
Andrew Warrington
@andrew-warrington
Jan 04 2016 00:29
2- shut off all ports between loops
my loops are 5 mins apart, due to Medtronic CGM refresh rate
Ben West
@bewest
Jan 04 2016 00:29
fwiw, on g4 + share and up, there is a command that tells the dexcom receiver to stop charging as wel
Andrew Warrington
@andrew-warrington
Jan 04 2016 00:29
I now am getting 6h15 mins, so 20% increase
I am planning to try one or two more things, then will write up the whole outcome together
Tony Zarro
@tzarro
Jan 04 2016 00:37
Quick update and question: had a really great first day on OpenAPS, very fascinating to watch it do its thing and all of its actions have made perfect sense to me thus far...only one issue is that roughly 40 minutes ago (while in the midst of a nap), it started preflight failing and hasn't stopped...I'm guessing a communication issue, and troubleshooting it now.
Main question: is there any advantage to implementing a periodic reboot, or is a periodic USB reset the better way to go (assuming that ends up being the issue)?
Also, how long does the oref0-reset-usb command usually take? It seems to take a very long time in the terminal...
Scott Leibrand
@scottleibrand
Jan 04 2016 00:39
that's because it resets your wifi, so you have to wait for it to re-establish before your terminal comes back
Tony Zarro
@tzarro
Jan 04 2016 00:40
ah, gotcha!
and FYI, resetting usb did indeed fix the issue, back to looping :)
Scott Leibrand
@scottleibrand
Jan 04 2016 00:40
what I do is: openaps alias add preflight '! bash -c "rm -f monitor/clock.json && echo -n \"PREFLIGHT \" && openaps report invoke monitor/clock.json 2>/dev/null >/dev/null && grep -q T monitor/clock.json && echo OK || ( mm-stick warmup || sudo oref0-reset-usb; echo FAIL; openaps get-bg; sleep 120; exit 1 )"'
the logic is, try to refresh clock.json, and if it has a T in it, you're all good. otherwise, (check the carelink stick, and if that fails run oref0-reset-usb) and then go to sleep for 2 minutes in case there's another Pi talking to the pump
Tony Zarro
@tzarro
Jan 04 2016 00:44
ah, very cool...thanks!
Dragan Maksimovic
@dm61
Jan 04 2016 00:46
I have also implemented that check, reset and wait in my preflight and have since had no loop interruptions due to the stick
Dragan Maksimovic
@dm61
Jan 04 2016 01:00
Any reasons you are using clock as a check, as opposed to, say, model?
Scott Leibrand
@scottleibrand
Jan 04 2016 01:09
Yes. Model is too quick a command, and succeeds when longer commands don't.
I believe clock actually does a model first.
Dragan Maksimovic
@dm61
Jan 04 2016 01:13
ok, thanks - I'll update my preflight
eyim
@eyim
Jan 04 2016 01:20
Anyone else trying meal assist? I think I have it right but am getting this error when I ran the monitor/meal.json report: Warning: could not parse meal_input. Meal Assist disabled.
Dana Lewis
@danamlewis
Jan 04 2016 01:23
Did you enter carbs using bolus wizard?
Scott Leibrand
@scottleibrand
Jan 04 2016 01:24
You have to set up some reports and stuff for meal-assist to work: https://gist.github.com/scottleibrand/63a8541397378bf9f4b8
eyim
@eyim
Jan 04 2016 01:24
Yes
I thought I got everything but will try again
Scott Leibrand
@scottleibrand
Jan 04 2016 01:26
also try running oref0-determine-basal manually with all the arguments
Jason Calabrese
@jasoncalabrese
Jan 04 2016 01:46
This message was deleted
diabeticgonewild
@diabeticgonewild
Jan 04 2016 03:14
OK I am at the crontab step
I put in sudo chmod +x /home/pi/test.sh
and my test.sh file looks like:
#!/bin/bash

cd my_openaps
openaps loop
My crontab looks like this:
# m h  dom mon dow   command
@reboot /home/pi/test.sh
*/5 * * * * /home/pi/test.sh
What am I doing wrong?
Dragan Maksimovic
@dm61
Jan 04 2016 03:17
I suggest you look at the example given by @ktomy earlier today; cron is not you, so all paths must be absolute.
diabeticgonewild
@diabeticgonewild
Jan 04 2016 03:17
It's not executing
Scott Leibrand
@scottleibrand
Jan 04 2016 03:34
Redirect output and errors to a file with > and 2>
diabeticgonewild
@diabeticgonewild
Jan 04 2016 03:35
I edited @ktomy 's example at 6:28 today...
Scott Leibrand
@scottleibrand
Jan 04 2016 03:35
and like dm61 said, all the paths in the shell script need to be absolute too
diabeticgonewild
@diabeticgonewild
Jan 04 2016 03:37
PATH=/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin
*/5 * * * * killall -g --older-than 10m openaps
*/5 * * * * cd /home/pi/my_openaps && openaps loop
*/5 * * * * cd /home/pi/my_openaps && ( ps aux | grep -v grep | grep -q 'openaps ' && echo OpenAPS already running || openaps loop ) 2>&1 | tee -a /var/log/openaps/uploader.log

I got the last line correct now....

I just don't have an openaps upload alias yet

Dragan Maksimovic
@dm61
Jan 04 2016 03:40
isn't your loop called 'loop', not 'upload'?
diabeticgonewild
@diabeticgonewild
Jan 04 2016 03:40
yeah
I thought upload was for NS
but I guess not
right?
Dragan Maksimovic
@dm61
Jan 04 2016 03:41
that's just how ktomy called his thing
diabeticgonewild
@diabeticgonewild
Jan 04 2016 03:41
got it
Dragan Maksimovic
@dm61
Jan 04 2016 03:41
nothing to do with NS upload
diabeticgonewild
@diabeticgonewild
Jan 04 2016 03:42
K
thanks
You're an engineer, right?
I am an electrical engineering student at arizona state university....but I have been in school forever.
I even got an internship offer with Intel in Folsom, CA, but I had to turn it down due to medical problems that are not really related to T1D.
Dragan Maksimovic
@dm61
Jan 04 2016 03:43
I am an EE, but not software developer, it's been very long time since I was in school
diabeticgonewild
@diabeticgonewild
Jan 04 2016 03:43
That's cool!
I wrote a controller in MATLAB for an artificial pancreas in 2011, when I was a junior in college, just to see if I could do it...It was a sliding mode control.
Do I need to mkdir /var/log/openaps ?
Dragan Maksimovic
@dm61
Jan 04 2016 03:46
probably, I do not know. I just put the log file in my folder
diabeticgonewild
@diabeticgonewild
Jan 04 2016 03:47
that's what I am going to do...can't hurt. I haven't gotten any logs yet. But thanks!
Dragan Maksimovic
@dm61
Jan 04 2016 03:47
It's cool you've worked on your own algorithm
diabeticgonewild
@diabeticgonewild
Jan 04 2016 03:48
It was based off of a paper....I could tell that it was (somehow) solvable, and I solved it just for fun, as a spring break project...at age 21.
Dragan Maksimovic
@dm61
Jan 04 2016 03:48
BTW, private chat is probably better for this type of conversation, so we do not clog this here with irrelevant stuff
diabeticgonewild
@diabeticgonewild
Jan 04 2016 03:49
sorry
diabeticgonewild
@diabeticgonewild
Jan 04 2016 04:07
OK so my crontab looks like this and I initiated it using crontab -l
PATH=/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin
*/5 * * * * killall -g --older-than 10m openaps
*/5 * * * * cd /home/pi/my_openaps && openaps loop
*/5 * * * * cd /home/pi/my_openaps && ( ps aux | grep -v grep | grep -q 'openaps ' && echo OpenAPS already running || openaps loop ) 2>&1 | tee -a /var/log/openaps/uploader.log
I don't know what's wrong with it
Scott Leibrand
@scottleibrand
Jan 04 2016 04:09
Does openaps loop work? Does /var/log/openaps/ exist?
You have two entries running at the same time both trying to run openaps loop.
Presumably the latter will usually say the former is already running, but the former isn't logging anywhere useful.
diabeticgonewild
@diabeticgonewild
Jan 04 2016 04:10
yeah openaps loop works, on it's own and I did mkdir /var/log/openaps
Scott Leibrand
@scottleibrand
Jan 04 2016 04:11
Ok, try commenting out the first one and just leave the one that's logging.
Then look at that log.
For debugging, I would change it from /5 to so you don't have to wait as long.
diabeticgonewild
@diabeticgonewild
Jan 04 2016 04:11
OK, I don't even have a log, I think
yeah, no log...
Scott Leibrand
@scottleibrand
Jan 04 2016 04:12
It should make one if that dir exists.
Did you do crontab -e?
diabeticgonewild
@diabeticgonewild
Jan 04 2016 04:12
this is my pwd right now -- pi@raspberrypi:/var/log/openaps $
Scott Leibrand
@scottleibrand
Jan 04 2016 04:12
-l just shows it.
diabeticgonewild
@diabeticgonewild
Jan 04 2016 04:12
yeah
yeah, I did crontab -e too
Scott Leibrand
@scottleibrand
Jan 04 2016 04:13
K. Make sure you have permissions by doing touch /var/log/openaps/loop.log
diabeticgonewild
@diabeticgonewild
Jan 04 2016 04:13
I commented out the second line
Scott Leibrand
@scottleibrand
Jan 04 2016 04:13
after that tail -F is your friend.
diabeticgonewild
@diabeticgonewild
Jan 04 2016 04:13
it should be uploader.log
Tony Zarro
@tzarro
Jan 04 2016 04:14
@diabeticgonewild , do you have the "SHELL=/bin/bash" line at the top of your crontab?
Scott Leibrand
@scottleibrand
Jan 04 2016 04:14
Well, ya, except it's not an uploader. ;-)
diabeticgonewild
@diabeticgonewild
Jan 04 2016 04:14
no @tzarro
Scott Leibrand
@scottleibrand
Jan 04 2016 04:14
but based on current crontab, ya. That was taken from an upload-only loop.
Hence the weird naming.
diabeticgonewild
@diabeticgonewild
Jan 04 2016 04:15
yeah
so I need SHELL=/bin/bash

pi@raspberrypi:/var/log/openaps $ crontab -e
crontab: installing new crontab
My crontab
SHELL=/bin/bash
PATH=/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin
*/5 * * * * killall -g --older-than 10m openaps
# */5 * * * * cd /home/pi/my_openaps && openaps loop
*/5 * * * * cd /home/pi/my_openaps && ( ps aux | grep -v grep | grep -q 'openaps ' && echo OpenAPS already running || openaps loop ) 2>&1 | tee -a /var/log/openaps/uploader.log
Tony Zarro
@tzarro
Jan 04 2016 04:18
@scottleibrand mentioned you might want to take out the "/5" from your timing configuration, just so that it runs every minute (better than waiting every five minutes when testing)...
diabeticgonewild
@diabeticgonewild
Jan 04 2016 04:19
I could just do /1
yeah changed it to /1 and I got an uploader.log file now. I needed the SHELL=/bin/bash in the first line
uploader.log is empty :(
diabeticgonewild
@diabeticgonewild
Jan 04 2016 04:24
Why would uploader.log be empty?
Scott Leibrand
@scottleibrand
Jan 04 2016 04:24
Did you touch it?
touch creates an empty file if none exists.
diabeticgonewild
@diabeticgonewild
Jan 04 2016 04:25
Oh I think so. Let me redo it, if not....
Yeah, I touched it, but I don't have anything in the file!
I just got lost cause I thought you may have mentioned a secondary command....sorry my BG is like 251...
Scott Leibrand
@scottleibrand
Jan 04 2016 04:27
So touch didn't complain? That was to check to make sure you could write to it.
diabeticgonewild
@diabeticgonewild
Jan 04 2016 04:27
Yeah I did sudo touch /var/log/openaps/uploader.log
Scott Leibrand
@scottleibrand
Jan 04 2016 04:28
There's your problem
the directory needs to be writable without sudo
so you'll need to chown it to be owned by the pi user instead of root
diabeticgonewild
@diabeticgonewild
Jan 04 2016 04:28
touch: cannot touch ‘/var/log/openaps/uploader.log’: Permission denied
Scott Leibrand
@scottleibrand
Jan 04 2016 04:29
thats the problem cron is having too
diabeticgonewild
@diabeticgonewild
Jan 04 2016 04:29
So what do I need to do?
Scott Leibrand
@scottleibrand
Jan 04 2016 04:29
chown the directory
this is something you need to actually understand, so it's worth a few minutes to Google how.
diabeticgonewild
@diabeticgonewild
Jan 04 2016 04:30
pi@raspberrypi:/var/log/openaps $ chown /var/log/openaps
chown: missing operand after ‘/var/log/openaps’
Try 'chown --help' for more information.
Scott Leibrand
@scottleibrand
Jan 04 2016 04:30
Linux permissions will give you no end of grief until you understand them.
-h or man is your friend, or google
diabeticgonewild
@diabeticgonewild
Jan 04 2016 04:30
chown root I guess
Scott Leibrand
@scottleibrand
Jan 04 2016 04:30
it needs to know what user to change ownership to
diabeticgonewild
@diabeticgonewild
Jan 04 2016 04:31
That didn't work...
Scott Leibrand
@scottleibrand
Jan 04 2016 04:31
ls -l will show you what it is now
also read up, what I said earlier.
diabeticgonewild
@diabeticgonewild
Jan 04 2016 04:33
/usr/bin/chown user:group $file
Scott Leibrand
@scottleibrand
Jan 04 2016 04:34
Group is optional
diabeticgonewild
@diabeticgonewild
Jan 04 2016 04:34
what would the group be in this case?
local?
Scott Leibrand
@scottleibrand
Jan 04 2016 04:35
Use ls -l on a file you own, like the stuff in your homedir
youll see the user and group those files have. You want the log to be the same user at least.
diabeticgonewild
@diabeticgonewild
Jan 04 2016 04:36
OK I am slightly confused. You said the directory had to be chowned....are you sure it is not the file?
Scott Leibrand
@scottleibrand
Jan 04 2016 04:37
If you chown the directory and delete the file, it can be recreated correctly. Or you could chown each file you want to be able to write to, individually.
diabeticgonewild
@diabeticgonewild
Jan 04 2016 04:38
OK, thanks.
So I have a problem, and I don't know what to do with it
pi@raspberrypi:/var/log/openaps $ /usr/bin/chown pi:pi
-bash: /usr/bin/chown: No such file or directory
I guess I need to create that directory?
Scott Leibrand
@scottleibrand
Jan 04 2016 04:41
You shouldn't be specifying a full path to a command interactively. Just chown.
Every *nix system puts it somewhere different.
but it'll always be in your PATH. if you need to know where it is for some reason, you can do which
diabeticgonewild
@diabeticgonewild
Jan 04 2016 04:45
Now I am getting this:
pi@raspberrypi:/var/log $ chown pi:pi openaps
chown: changing ownership of ‘openaps’: Operation not permitted
pi@raspberrypi:/var/log $
Jason Calabrese
@jasoncalabrese
Jan 04 2016 04:47
you need to be root, try sudo chown ...
diabeticgonewild
@diabeticgonewild
Jan 04 2016 04:47
that's what I just found out! Thanks!
Scott Leibrand
@scottleibrand
Jan 04 2016 04:48
There you go. Yep. The Pi user doesn't have permissions to write to that directory, so of course it doesn't have permissions to change the permissions either, unless sudo says so.
diabeticgonewild
@diabeticgonewild
Jan 04 2016 04:48
pi@raspberrypi:/var/log/openaps $ ls -l
total 4
-rw-r--r-- 1 pi pi 215 Jan  3 22:48 uploader.log
I just got thrown off because I used sudo for the touch command
Scott Leibrand
@scottleibrand
Jan 04 2016 04:49
yep. Sudo is not magic, it just lets you run some things as root. In this case, that's the opposite of what you wanted, because cron doesn't sudo
diabeticgonewild
@diabeticgonewild
Jan 04 2016 04:50
Now I have to do touch /var/log/openaps/uploader.log
Scott Leibrand
@scottleibrand
Jan 04 2016 04:50
now you can tail -F that log to watch things happen
no, it already wrote to the file
touch was just for testing.
diabeticgonewild
@diabeticgonewild
Jan 04 2016 04:50
tail -F uploader.log?
Scott Leibrand
@scottleibrand
Jan 04 2016 04:51
If you're in that directory. Otherwise full path.
diabeticgonewild
@diabeticgonewild
Jan 04 2016 04:51
yeah I'm in that directory. It looks like it's working on its own
Have to enter in another calibration cause CGM was sooooooo offf, but otherwise it is good...
I'm 306 right now, sigh....
Yeah, so it's definitely looping...
Scott Leibrand
@scottleibrand
Jan 04 2016 04:55
Nice. Doing a sensible thing?
diabeticgonewild
@diabeticgonewild
Jan 04 2016 04:59
Yeah except I am high...but yeah it's working...
Except it says no CareLink stick available
ValueError: Attempting to use a port that is not open
/usr/local/bin/mm-stick FAIL
fail NO MEDTRONIC CARELINK STICK AVAILABLE
No CGM data.
Dana Lewis
@danamlewis
Jan 04 2016 05:01
Is cgm and Carelink plugged in?
diabeticgonewild
@diabeticgonewild
Jan 04 2016 05:01
Yeah but I was moving around, so IDK
Dana Lewis
@danamlewis
Jan 04 2016 05:02
If in doubt, unplug and re plug Carelink stick
diabeticgonewild
@diabeticgonewild
Jan 04 2016 05:04
I just restarted the Pi, and the same error came up. It's either cause it's not plugged in right or my dog chewed up the stick too badly back from when she was a puppy!
Yeah, it's working fine now.
I just ordered a bunch of cords and a couple more batteries.
I'm going to affix some of the more vulnerable connections with electrical tape, such as the CareLink stick
This is what I am going to try tomorrow (along with getting a second Pi up and working...)...having an offline hotspot and a wifi connection for connecting to my iPhone's hotspot http://raspberrypi.stackexchange.com/questions/12276/how-to-make-pi-an-ap-with-2-wifi-adapters
You use 2 Edimax adapters ($20 total)
There are adapters that can run in hotspot and access point mode at the same time but they are bulkier...
diabeticgonewild
@diabeticgonewild
Jan 04 2016 05:10
Obviously that will be added to docs after testing...
Also if anyone needs a keyboard that fits in their pocket, this is the one to go with (I have one...)
stardrift Foldable Bluetooth Keyboard Ultra Slim Pocket Size Standard Keypad HQ Durable Aluminum Structure with remote Camera for iPhone iPad Tablet Window PC Android Smartphones Black https://www.amazon.com/dp/B010VEQIEQ/ref=cm_sw_r_cp_awd_7VFIwbJ5QX8ZP
Scott Leibrand
@scottleibrand
Jan 04 2016 05:21
Is it executing temps?
diabeticgonewild
@diabeticgonewild
Jan 04 2016 05:22
It says like "no action required"
I turned it off for the night...
Scott Leibrand
@scottleibrand
Jan 04 2016 05:23
K. Worth making note of the reason field to understand why it doesn't think action is needed.
Have a good night!
diabeticgonewild
@diabeticgonewild
Jan 04 2016 05:25
You too!
diabeticgonewild
@diabeticgonewild
Jan 04 2016 05:30
Now I am getting "no CGM data" and I unplugged and replugged in the cable for the CGM. Anyway, I am going to bed. So good night, all!
I have a neuro-ophthalmologist appointment in the morning...
Tony Zarro
@tzarro
Jan 04 2016 16:18
Quick update: my first overnight on OpenAPS went very well...my BG remained in a pretty tight range all night (between 130 - 150...note that my target BGs are set higher for safety while I'm trialing the system)...
Dana Lewis
@danamlewis
Jan 04 2016 16:19
Nice!
Did the loop get a lot of action?
Tony Zarro
@tzarro
Jan 04 2016 16:21
Yep, it did...my overnight basals have never been extremely accurate, so the BGs remaining in-range had much to do with OpenAPS
Scott Leibrand
@scottleibrand
Jan 04 2016 16:21
:+1:
Tony Zarro
@tzarro
Jan 04 2016 16:22
one non-system issue, though: I'm not getting a full 3 feet of range with my CareLink USB...had quite a few preflight fails because of this
Scott Leibrand
@scottleibrand
Jan 04 2016 16:22
story of our lives.
Tony Zarro
@tzarro
Jan 04 2016 16:22
started wearing the pump on my arm and it was a little better...
are there any projects ongoing to use different hardware with better range, or is that completely out of the question due to the proprietary nature of the stick?
Dana Lewis
@danamlewis
Jan 04 2016 16:23
Yes
See the openAPS winter update post for hardware info
Tony Zarro
@tzarro
Jan 04 2016 16:23
ok
Andrew Warrington
@andrew-warrington
Jan 04 2016 16:47
congrats tzarro!
Tony Zarro
@tzarro
Jan 04 2016 17:19
One semi-related question for those on Dexcom: do you find it sufficient to only enter a calibration glucose reading when prompted, or do you get much better results doing it more often?
Dana Lewis
@danamlewis
Jan 04 2016 17:20
Definitely recommend calibrating 2x/day. I do mine when waking up and before going to bed.
Scott Leibrand
@scottleibrand
Jan 04 2016 17:20
The first or second day, it's good to also calibrate when high and flat or low and flat, so it can figure out how high the highs really are.
other than that, I'd do 2x a day or whenever you test for another reason (like when you don't trust its readings), whichever is more frequent.
Dana Lewis
@danamlewis
Jan 04 2016 17:21
This message was deleted
Dragan Maksimovic
@dm61
Jan 04 2016 17:27
@tzarro you are likely to get worse results if you calibrate more often. I calibrate only when prompted and when the bg curve is really flat, 2x a day, unless, as Scott suggested, there are reasons to question the CGM, e.g. readings look really unexpected or noisy.
Chris Oattes
@cjo20
Jan 04 2016 17:28
I find that calibrating a noisy CGM is just going to throw the calibration off further
generally, a noisy cgm is a dying cgm, calibrating won't fixt hat
LadyViktoria
@LadyViktoria
Jan 04 2016 17:31
@cjo20 you can remove the transmitter of a noisy dexcom and move the little rubber thing with the wire a bit in and out and cleaning it with a desinfection spray. the sensor will stop being noisy ;)
Scott Leibrand
@scottleibrand
Jan 04 2016 17:33
Interesting.
LadyViktoria
@LadyViktoria
Jan 04 2016 17:33
i also saw some picture of a "refubished" sensor in a resettet setter which is able to set again
the people who tried this say it works without noise and last one or 2 weeks
Scott Leibrand
@scottleibrand
Jan 04 2016 17:34
For noisy/dying sensor, I agree calibration doesn't help much. But for "jumpy" overreactive new sensors it helps a lot. Keep in mind that the Dexcom (adult 505 firmware) keeps last 11 calibrations. So more than 6/day would be excessive.
LadyViktoria
@LadyViktoria
Jan 04 2016 17:35
you can also try to make a new hole with a cannula and stick the sensor needle carefully in this hole
Tony Zarro
@tzarro
Jan 04 2016 17:36
thanks...never got a good answer on other boards, I figured you folks would know better :)
LadyViktoria
@LadyViktoria
Jan 04 2016 17:37
but keep in mind to use desinfection spray ;) when do such surgeries with your sensor :angel:
Scott Leibrand
@scottleibrand
Jan 04 2016 17:37
The first day you also don't want to do too many, as the body is still reacting to the new foreign object, so the slope is changing. Adrian, Costik and others suggested calibrating more on day 2. We also just started letting a new sensor embed, protected by an old dead transmitter for the first day, and then switching over to it after 24h or so, after most of the wonky reactions are done.
obviously if you have 2 working transmitters you can do even better.
Chris Oattes
@cjo20
Jan 04 2016 17:39
I use enlite, but I tend to insert the night before and start the sensor in the morning
that gets rid of most of the wonkyness
Kevin Wells
@wkcwells
Jan 04 2016 19:44
Hi all. I am new to OpenAPS but have an open loop system running with remarkable little trouble. Thanks for all the great documentation and info!
Dana Lewis
@danamlewis
Jan 04 2016 19:45
Welcome @wkcwells ! Glad it's going well so far.
We love feedback, so please share any thoughts you've had along the way or submit PRs to the docs.
Kevin Wells
@wkcwells
Jan 04 2016 19:54
Thanks Dana. Before I close the loop, I’d like to feed my OpenAPS data automatically up to my Nightscout console. (Am currently using the Dexcom bridge for CGM data.) I saw a screen shot showing a basal rates graph overlay (can’t find it now), and would like that plus IOB. Does anyone have a pointer to steps to accomplish this? Thanks in advance! Kevin Wells
Dana Lewis
@danamlewis
Jan 04 2016 19:56
Good Q @wkcwells . We just added some things to the docs to help with that. It’s not 100% done but should get you started, and maybe you could add some more notes to it. https://github.com/openaps/docs/blob/master/docs/Automate-system/vizualization.md @eyim and @audiefile did a great job with first-round writeup on it.
Dragan Maksimovic
@dm61
Jan 04 2016 20:02
A side question: is it ok to just edit openaps.ini directly, instead of using add/remove commands. I am planning to clean up my reports and aliases, and I would find direct editing easier to do. But, I am not sure if add/remove do anything that's hidden in addition to modifying the openaps.ini file?
Scott Leibrand
@scottleibrand
Jan 04 2016 20:03
editing directly is fine for an end user
Dragan Maksimovic
@dm61
Jan 04 2016 20:04
Great, thx
Scott Leibrand
@scottleibrand
Jan 04 2016 20:04
it's just hard to document what you've done for other people (or for you to do again later)
but if you're just doing something that's already documented, edit away. :)
Dragan Maksimovic
@dm61
Jan 04 2016 20:04
Sure, makes sense
Tony Zarro
@tzarro
Jan 04 2016 20:08
Question about integrating to nightscout: I did everything in the "Nightscout Integration" section of "Visualization and Monitoring", but not the "Upload Latest Treatments" section...is the second section required to get basic glucose data to nightscout? I'm not seeing any data at all even though I'm pretty sure I've done everything to the letter...any idea where I would start my troubleshooting?
Kevin Wells
@wkcwells
Jan 04 2016 20:11
Thanks @danamlewis & @scottleibrand. Looking forward to digging in on integration with Nightscout.
Dana Lewis
@danamlewis
Jan 04 2016 20:11
did you have nightscout setup before, @tzarro ?
Jason Calabrese
@jasoncalabrese
Jan 04 2016 20:11
the current NS integration is only uploading treatments and status, not cgm data
Dana Lewis
@danamlewis
Jan 04 2016 20:11
Not sure if @eyim documented from perspective of someone already setup on nightscout or not
Tony Zarro
@tzarro
Jan 04 2016 20:12
BTW, one thing that is really a pleasant surprise for me about OpenAPS is how much it prevents spikes, and how smooth the curves are...in the past I've always been afraid to be as aggressive as I should be with boluses in fear of going low, but OpenAPS has helped out a lot because if I bolus as I should, for the 3-hr duration of my bolus OpenAPS has an opportunity to adjust my basal every minute in case the bolus was too much...I hadn't completely appreciated those effects before I went live :)
no, this is a brand new implementation...installed master branch, setup profile, then installed dev branch, enabled OpenAPS plugin, etc...
Dana Lewis
@danamlewis
Jan 04 2016 20:13
we like pleasant surprises :)
Tony Zarro
@tzarro
Jan 04 2016 20:13
@jasoncalabrese ah, ok...so then the blood sugar data will come from the treatment uploads, right? Sounds like I just need to finish implementing that section...
Dragan Maksimovic
@dm61
Jan 04 2016 20:14
One more detail: my glucose.json has 5 entries, which I think was suggested somewhere in the docs. As far as I could tell, the code is using just last 3 entries (is that so?); any reasons to have 5 (or any more?) entries in glucose.json?
Dana Lewis
@danamlewis
Jan 04 2016 20:14
Hm, @tzarro we may need to make this more clear for someone who didn’t already have NS setup
I think you also need to follow normal NS setup instructions
Scott Leibrand
@scottleibrand
Jan 04 2016 20:14
@dm61: I think it uses current bg and 15m ago, so last 4
Dragan Maksimovic
@dm61
Jan 04 2016 20:15
yes, current and past 3 values, so 4
Scott Leibrand
@scottleibrand
Jan 04 2016 20:15
doesn't really matter: the Dex still needs to read and return one page (several hours or maybe up to a day) of records to get the last few
Tony Zarro
@tzarro
Jan 04 2016 20:15
I thought I did follow normal NS setup, but I'll go back to those instruction pages to see if I missed anything...but I haven't setup the treatment uploads, sounds like that's a definite missing piece in my implementation...
Dana Lewis
@danamlewis
Jan 04 2016 20:15
@tzarro did you start here? http://nightscout.info/
seeing the openaps docs, may not be clear if you didn’t have nightscout already, you need to get the CGM data in through normal NS process (use the http://nightscout.info/ site), then come back to our instructions.
Dragan Maksimovic
@dm61
Jan 04 2016 20:16
But it might be useful to look into better ways to compute the "average" slope?
Scott Leibrand
@scottleibrand
Jan 04 2016 20:16
we need to make that part of the code aware of the timestamps on the records
Jason Calabrese
@jasoncalabrese
Jan 04 2016 20:16
right now we're assuming that the dex data is uploaded to NS some other way (not from the pi), but it's possible to upload it from the pi, just no tool for it yet
Scott Leibrand
@scottleibrand
Jan 04 2016 20:16
right now it overreacts to big gaps in data when rising/falling
Dragan Maksimovic
@dm61
Jan 04 2016 20:17
yes it does, I've noticed that
Tony Zarro
@tzarro
Jan 04 2016 20:17
@danamlewis yep, I started there...I haven't setup any devices (android uploader phone, or anything else) to upload the cgm data though, since I assumed that would flow through OpenAPS...so will I need to do something like an android uploader phone?
Scott Leibrand
@scottleibrand
Jan 04 2016 20:17
I also want to do a 30m lookback in addition to the 15m, and take the lesser of the two deltas so as to avoid overreacting to jumpy sensor data or compression events
@tzarro do you have Share?
if so, nightscout can get the BG directly from Dexcom's Share servers
Tony Zarro
@tzarro
Jan 04 2016 20:18
Nope, I don't have Share
Dragan Maksimovic
@dm61
Jan 04 2016 20:18
@scottleibrand I think that would be very good
Jason Calabrese
@jasoncalabrese
Jan 04 2016 20:19
if the dex is plugged into the pi we have the data, just need to transform it some and post it to NS
Dragan Maksimovic
@dm61
Jan 04 2016 20:19
But, you do not want to add more delay either...
Paul Andrel
@stavlor
Jan 04 2016 20:19
should be relatively easy to send data from dex plugged into pi to ns
Scott Leibrand
@scottleibrand
Jan 04 2016 20:19
@dm61 only thing that would delay would be responding to a sudden reversal over < 30m
Tony Zarro
@tzarro
Jan 04 2016 20:20
So I'm concluding that its been rare for someone to implement OpenAPS before they have implemented Nightscout? :)
Scott Leibrand
@scottleibrand
Jan 04 2016 20:20
yes, quite. :)
Dana Lewis
@danamlewis
Jan 04 2016 20:20
correct :)
Tony Zarro
@tzarro
Jan 04 2016 20:20
lol
Dana Lewis
@danamlewis
Jan 04 2016 20:20
but this means an awesome opp to help fix the docs for the next person ;)
Jason Calabrese
@jasoncalabrese
Jan 04 2016 20:20
if we were reading raw from the pi it could replace an android uploader
Dragan Maksimovic
@dm61
Jan 04 2016 20:20
@scottleibrand in my 'manual' practice, the inflection points are actually very important
Tony Zarro
@tzarro
Jan 04 2016 20:21
I'll volunteer to add some notes to the docs about this...is editing the docs similar to editing code in the branch, i.e. Edit and Commit?
Dana Lewis
@danamlewis
Jan 04 2016 20:22
You need to fork and create a new branch, edit there, then PR back to master
(but yes)
Tony Zarro
@tzarro
Jan 04 2016 20:22
ah, ok...never done that before, but I'll give it a shot tonight
Dana Lewis
@danamlewis
Jan 04 2016 20:23
follow the green buttons :) for the pull request, and we can help with any questions.
Scott Leibrand
@scottleibrand
Jan 04 2016 20:23
@dm61 so if you're flat and then have a sudden up/down tick, it will respond the same. if you're rising and then start to fall, it would reduce the response until the 15m ago <= 30m ago, so on average an extra 5-10m
Paul Andrel
@stavlor
Jan 04 2016 20:23
its relatively simple and github is quite useful for guiding people through a pr
Tony Zarro
@tzarro
Jan 04 2016 20:23
if I've been forked for a while, I assume there is a way to get the latest version of a file from master, then edit/commit, then PR, right?
Dragan Maksimovic
@dm61
Jan 04 2016 20:23
@tzarro once you do it, please get back here and tell me how to do it - I am still hesitant to do any PR or anything else on github...
Tony Zarro
@tzarro
Jan 04 2016 20:23
ok
Scott Leibrand
@scottleibrand
Jan 04 2016 20:23
@dm61 same with a sudden uptick after descending
@dm61 you can't do any harm trying to PR: you don't have permissions to do anything other than generate emails. :)
so give it a try, and if you do it wrong you can close the PR and try again, or one of us can point out how to do what you're trying to
Tony Zarro
@tzarro
Jan 04 2016 20:25
ah, that's comforting...so the PR process routes it to folks who actually know what they're doing :)
Dana Lewis
@danamlewis
Jan 04 2016 20:25
@tzarro yes, re-fork from master, and do the process as you described. if needed you can delete your old fork as long as you didn’t have unsaved changes there that hadn’t already been PRed back
Tony Zarro
@tzarro
Jan 04 2016 20:25
makes perfect sense...thanks!
Dana Lewis
@danamlewis
Jan 04 2016 20:25
yea. your PR will go to people to get multiple +1s for any big changes before it gets officially added
Scott Leibrand
@scottleibrand
Jan 04 2016 20:25
@tzarro exactly. :)
Paul Andrel
@stavlor
Jan 04 2016 20:25
:+1:
Dana Lewis
@danamlewis
Jan 04 2016 20:25
so PRs never ever hurt :) and are a great way to stub out notes, suggest adding a section, ask questions for more details, etc. :)
danamlewis @danamlewis channels @bewest
Dragan Maksimovic
@dm61
Jan 04 2016 20:25
@scottleibrand ok, ok, I'll look into this some more; from limited experience so far, I think I can see some instances where openAPS action could have been faster (or slower), which would have been decided on how the average slope was computed...
Scott Leibrand
@scottleibrand
Jan 04 2016 20:25
:+1:
Dana Lewis
@danamlewis
Jan 04 2016 20:26
and @dm61 the docs are an excellent place to learn PR/Github protocol to get you even more comfortable before doing code-based PRs :)
Scott Leibrand
@scottleibrand
Jan 04 2016 20:26
@dm61 looking forward to it. not enough people are actually drilling down into the algorithm and suggesting improvements. would love to have someone to collaborate with on that
Dana Lewis
@danamlewis
Jan 04 2016 20:26
:+1: (he’s tired of my suggestions ;))
Paul Andrel
@stavlor
Jan 04 2016 20:26
*takes cover while @danamlewis channels @bewest
Dana Lewis
@danamlewis
Jan 04 2016 20:26
lol @stavlor
Dragan Maksimovic
@dm61
Jan 04 2016 20:29
@danamlewis sounds good
Tony Zarro
@tzarro
Jan 04 2016 20:29
interesting...I've only worked on dev projects with very rigid "top-down" development processes...the PR process sounds great for distributed, community-based software development...this has been a very interesting process for me :)
Dana Lewis
@danamlewis
Jan 04 2016 20:29
were those open source projects? and yup :)
Tony Zarro
@tzarro
Jan 04 2016 20:30
no, never worked on anything open-source...all corporate proprietary stuff
and I've never actually done any coding...had the good sense to stay away from their object repository :)
Jason Calabrese
@jasoncalabrese
Jan 04 2016 20:32
one thing that works good for me is to keep the master and dev branches in my fork clean, and create new branches for things I want to PR
eyim
@eyim
Jan 04 2016 20:32
@danamlewis @wkcwells Hi Dana - yes the documentation around Nightscout is from the perspective of someone already setup and using Nightscout. I put a section about how to pull data from NS into OpenAPS but not to push the data into NS
Dana Lewis
@danamlewis
Jan 04 2016 20:32
I’m the opposite - no corporate dev projects, but regardless I really like the flexible structure of this approach. Small changes just need a single +1; medium might get +2 or more; only the big reorg-all-things require the entire cavalry to come in and review. Helps expedite getting a MVP or the next iteration up, and then someone can always come behind and do another PR to edit/adjust as desired.
Jason Calabrese
@jasoncalabrese
Jan 04 2016 20:32
that way it's really easy to update your master/dev branch to get in sync
Paul Andrel
@stavlor
Jan 04 2016 20:34
@jasoncalabrese indeed, its easiest to do most work in a per-feature branch or per issue/fix branch rather than doing stuff in dev/master as
Dana Lewis
@danamlewis
Jan 04 2016 20:34
+1
thanks @eyim for confirming
Paul Andrel
@stavlor
Jan 04 2016 20:35
also helps cut down on potential conflicts from your fork to the main repo
Jason Calabrese
@jasoncalabrese
Jan 04 2016 20:35
I also like to have a prod or deploy branch, that includes anything custom or not merged that I want to use now
Paul Andrel
@stavlor
Jan 04 2016 20:36
@jasoncalabrese tags are quite useful for that aswell
@tzarro ```[pandrel@paul-desktop cgm-remote-monitor]$ git pull
remote: Counting objects: 51, done.
remote: Compressing objects: 100% (25/25), done.
remote: Total 51 (delta 29), reused 17 (delta 17), pack-reused 9
Unpacking objects: 100% (51/51), done.
From https://github.com/nightscout/cgm-remote-monitor
11c77c2..728607a dev -> origin/dev
377a8f7..e38be3a hotfix/node-version-azure -> origin/hotfix/node-version-azure
82e3f55..dda9962 master -> origin/master
  • [new branch] wip/faster-load -> origin/wip/faster-load
  • [new tag] 0.8.4 -> 0.8.4
    There is no tracking information for the current branch.
    Please specify which branch you want to merge with.
    See git-pull(1) for details```in a branch that isn't already on the origin works alot easier than having a bunch of un pushed changes in your local dev or master and trying to prevent conflicts the next time you try to update your local development copy
Scott Leibrand
@scottleibrand
Jan 04 2016 20:46
@stavlor you can edit that to put a newline before and after the ``` and gitter will format it nicely
Paul Andrel
@stavlor
Jan 04 2016 20:50
ah
Ben West
@bewest
Jan 04 2016 21:18
ns-upload can do a lot
use mm-format-ns-entries I think
and ns-upload ... for entries
oh
oh, dexcom_reader to NS
need to run it via json tool, I thinink
then use ns-upload
Jason Calabrese
@jasoncalabrese
Jan 04 2016 21:21
we'd need to do some deduping like treatments
Rachel Sandlain
@audiefile
Jan 04 2016 21:22
@wkcwells or @tzarro let me know if you still have Nightscout set up questions and I'll do what I can to help
Ben West
@bewest
Jan 04 2016 21:23
ns-upload has nice help output and a --config tool to help create reports
$ ns-upload
ns-upload: missing API_SECRET
Usage: ns-upload <NIGHTSCOUT_HOST|localhost:1337> <API_SECRET> [API-TYPE|entries.json] <monitor/entries-to-upload.json> [stdout|-]

ns-upload --config <NIGHTSCOUT_HOST> <PLAIN_API_SECRET> <API-TYPE|entries.json> <monitor/entries-to-upload.json> output-report.json

ns-upload help - This message.
Usage: ns-upload <NIGHTSCOUT_HOST-http://localhost:1337> <API_SECRET> [entries|treatments|profile/] <file-to-upload.json> 
bewest@bewest-MacBookPro:~$
(aside: RIP hither and patient)
Jason Calabrese
@jasoncalabrese
Jan 04 2016 21:26
? that was your old laptop?
Jason Calabrese
@jasoncalabrese
Jan 04 2016 21:58
lots of hard coded limits for wtf-assist openaps/oref0@9769d18
Paul Andrel
@stavlor
Jan 04 2016 21:59
wtf-assist nice name lol
Jason Calabrese
@jasoncalabrese
Jan 04 2016 21:59
it helps, but hasn't been able to keep up with BG in my tests, even with it was more aggressive
at first it was only based on carb input, but now when there is a fast unexpected rise (a wtf) it high temps sooner and at a higher rate
Paul Andrel
@stavlor
Jan 04 2016 22:02
nice
Jason Calabrese
@jasoncalabrese
Jan 04 2016 22:03
maybe the minDelta limit should only apply when there aren't carbs
Tony Zarro
@tzarro
Jan 04 2016 22:28
Question: is there a way for OpenAPS to use CGM readings from NS instead of readings directly from the Dexcom? I am thinking of doing my NS implementation with a wired connection from the CGM to an uploader phone...
and if so, is that what the get-ns-glucose alias is for?
Dana Lewis
@danamlewis
Jan 04 2016 22:29
Yes, it’s possible.
Jason Calabrese
@jasoncalabrese
Jan 04 2016 22:31
that's how I'm using it most of the time, have an alias that first tries for a local usb dex, and if it fails uses NS
Tony Zarro
@tzarro
Jan 04 2016 22:31
@jasoncalabrese ah, that's awesome...
Jason Calabrese
@jasoncalabrese
Jan 04 2016 22:31
glucose-local = report invoke monitor/glucose.json
glucose-ns = report invoke monitor/glucose-ns.json
get-bg = ! bash -c "rm -f monitor/glucose*.json && openaps glucose-local 2>/dev/null || ( openaps glucose-ns && grep -q sgv monitor/glucose-ns.json && mv monitor/glucose-ns.json monitor/glucose.json )"
Dana Lewis
@danamlewis
Jan 04 2016 22:32
^ same. keep in mind that’ll require connectivity for the Pi, though, so you’ll either have to be in a known network or hotspot from another device
Jason Calabrese
@jasoncalabrese
Jan 04 2016 22:32
based on some tips from @scottleibrand
Dana Lewis
@danamlewis
Jan 04 2016 22:32
so keep that in mind if you ever go hiking/camping/on a plane/situations without good connectivity.
Jason Calabrese
@jasoncalabrese
Jan 04 2016 22:32
nice to just plugin the dex and have it work
Dana Lewis
@danamlewis
Jan 04 2016 22:32
yup.
Tony Zarro
@tzarro
Jan 04 2016 22:32
Yep, fair enough...90% of the time I'm either in the office or at work, and for the 10% of time I'm not I can just connect the CGM to the Pi instead of the uploader...
Dana Lewis
@danamlewis
Jan 04 2016 22:33
+1
ya, I would just make sure you’ve got it built to handle it that way vs skipping that step
Tony Zarro
@tzarro
Jan 04 2016 22:33
definitely, yes
Jason Calabrese
@jasoncalabrese
Jan 04 2016 22:33
and with multiple pis for better coverage you pretty much need something like that
Tony Zarro
@tzarro
Jan 04 2016 22:34
Does everyone else get the feeling that once APSs are commercially available, we'll still be getting better results with OpenAPS for at least 10 years afterward? :)
Jason Calabrese
@jasoncalabrese
Jan 04 2016 22:34
YES
the commercial systems will be limited black boxes
Dana Lewis
@danamlewis
Jan 04 2016 22:36
I hate to be a pessimist, but…think it’s highly likely it’ll take a few commercial options and some fine-tuning options in round 2 or 3 of each device before you can get the same kind of customization we’re getting used to with openaps
Andrew Warrington
@andrew-warrington
Jan 04 2016 22:36
Hi all, would be interesting to get some base statistics from any people who are willing to share. I have a couple of interesting ones from my first week of looping. I will share now.
Dana Lewis
@danamlewis
Jan 04 2016 22:36
so it’ll be a tradeoff for whether the low-burden/easy-to-use but less customized black box is better than the burden of the bigger setup, etc. with responsibility of making sure it works.
Andrew Warrington
@andrew-warrington
Jan 04 2016 22:36
Keep in mind before I share this, that mine is not an OpenAPS implementation, but I think you guys are making the same improvements, or better.
Dana Lewis
@danamlewis
Jan 04 2016 22:37
@tzarro hopefully not as bad as 10, but 3-5 wouldn’t surprise me
Andrew Warrington
@andrew-warrington
Jan 04 2016 22:37
And one other thing to know about my setup is that I am not announcing meals, so my spikes are longer-lasting than most... anyhow, stats:
Milos Kozak
@MilosKozak
Jan 04 2016 22:37
ahhh another step done with danar, decoded bolus history and receiving :)
Andrew Warrington
@andrew-warrington
Jan 04 2016 22:37
week before looping, number of (mild) hypo excursions: 7
first week of looping, number of hypo excursions: 0
week before looping: average blood glucose (in mmol/l): 8.9
first week of looping: 8.8
week before looping, average insulin per day: 39 units
first week of looping: 25.8
Scott Leibrand
@scottleibrand
Jan 04 2016 22:40
@jasoncalabrese were you referring to openaps/oref0@9769d18 or something else re minDelta?
Dana Lewis
@danamlewis
Jan 04 2016 22:41
niiiiice. thanks for sharing @andrew-warrington . Related: I will probably put out a call for data in a few weeks for a possible meta-analysis/aggregate look at what people are seeing before/after they start. Would love ongoing thoughts about what data will be good to track - I think change in insulin per day is interesting, esp. if you reduce basals and prevent the rebounds that require more insulin for correction. I also saw some reduction in TDD (total daily dosage) when I started looping as well.
Andrew Warrington
@andrew-warrington
Jan 04 2016 22:41
I had to fix two numbers up there just now
Scott Leibrand
@scottleibrand
Jan 04 2016 22:41
I added that because we just saw another case where the combination of Dana high-temping and the loop being in meal-assist mode made it want to high-temp a bit too long. I figure if your BG isn't rapidly rising, there's less need for being meal-assist aggressive
Andrew Warrington
@andrew-warrington
Jan 04 2016 22:41
mistake in Excel!
but still good
Tony Zarro
@tzarro
Jan 04 2016 22:42
@danamlewis yep, I'm just guessing that any first-out solution would end up being frustrating and limiting for folks who are on OpenAPS (especially as OpenAPS progresses further before we even have other commercial options)...
Scott Leibrand
@scottleibrand
Jan 04 2016 22:44
we definitely know that will be true for MDT 670G
still holding out hope that Bigfoot's solution will allow enough tuning of aggressiveness to meet our needs
Tony Zarro
@tzarro
Jan 04 2016 22:52
One other question about connecting the CGM directly to the uploader phone for NS...is there any reason why I can't enable the wifi hotspot on my uploader phone, so that the Pi can always have connectivity to NS for pulling glucose readings?
Jason Calabrese
@jasoncalabrese
Jan 04 2016 22:53
@scottleibrand makes sense in the case where we assume there are carbs, but maybe we should be more aggressive when there are carbs entered, ideally it would work as auto extended bolus
a slow and steady increase would be possible there
Rachel Sandlain
@audiefile
Jan 04 2016 22:56
@tzarro not as far as I know but it will have a major impact on battery life
also most uploader phones can upload or charge, they can't do both
Dana Lewis
@danamlewis
Jan 04 2016 23:00
@tzarro per @audiefile , just battery. Many uploader phones can’t charge and be used at the same time, so when I was using an android uploader I had one for daytime (without hotspotting) and one for evening, so I’d swap them and let the other one charge. So not sure how hard of a turnaround that would be if you were uploading and hot spotting, depending on how you’d needed to charge
will you be able to add the work wifi as a known wifi? if that and home are your 90%, you won’t need to hotspot there, just if you’re out and about, so it may not be as big of a deal to double up on upload and hotspot during those time
if you can’t add work wifi, though, might get annoying
Rachel Sandlain
@audiefile
Jan 04 2016 23:03
What model dexcom do you have?
Scott Leibrand
@scottleibrand
Jan 04 2016 23:06
@jasoncalabrese perhaps. but it's much more common, for us, that a low minDelta indicates carb absorption is ending, and we should stop high-temping, rather than predicting a continued slow rise
what threshold do you think would work to differentiate those? 0 is too low, but 5 mg/dL per 5m may be too high. thoughts?
Dana Lewis
@danamlewis
Jan 04 2016 23:10
3?
4 is my favorite, but ya’ll tend to like odd numbers..so..
Scott Leibrand
@scottleibrand
Jan 04 2016 23:12
yeah, I'm gonna drop it to >3
Tony Zarro
@tzarro
Jan 04 2016 23:12
I've got the old G4, without Share
Yeah, I can add my work network...I think I'll do that and just plug my CGM directly into OpenAPS when I don't have wifi
Dana Lewis
@danamlewis
Jan 04 2016 23:14
@tzarro fwiw, I might consider a SHARE just so you don’t have to fiddle with an uploader phone. I dropped it as soon as the SHARE receiver came in. Esp. with openaps already being a handful/pocketful, any simplification/downsizing on the physical component is nice
Tony Zarro
@tzarro
Jan 04 2016 23:15
@danamlewis how would the Share work with an Android phone?
is there an app yet? Doesn't seem so from the NS site
Rachel Sandlain
@audiefile
Jan 04 2016 23:16
@tzarro I would second that. I personally use share with xdrip on my android phone to upload to nightscout, and then plug in the receiver to the pi. That way pi works with no network, nightscout gets data even if I don't have pi, or pi doesn't have a connection. If phone doesnt have data I can use camping mode to see bg on pebble still
YES. It's called xdrip not nightscout
Dana Lewis
@danamlewis
Jan 04 2016 23:16
do you have an iphone or an android as your normal phone, @tzarro ?
Tony Zarro
@tzarro
Jan 04 2016 23:16
yes, android for normal phone
Dana Lewis
@danamlewis
Jan 04 2016 23:17
k. see what @audiefile just suggested then :) ^ re: xdrip
Tony Zarro
@tzarro
Jan 04 2016 23:17
I got the impression that xdrip was experimental hardware, but are you saying that its an uploader software I can use on android?
Dana Lewis
@danamlewis
Jan 04 2016 23:18
related, @audiefile I think this one has your name on it :) openaps/docs#54
Scott Leibrand
@scottleibrand
Jan 04 2016 23:18
yeah, xdrip was originally hardware, but now it's an app too
unfortunately the software-only solution requires a Share receiver
Tony Zarro
@tzarro
Jan 04 2016 23:18
OK, well I'm not too far from my typical Dexcom upgrade window...
Scott Leibrand
@scottleibrand
Jan 04 2016 23:19
for non-Share G4, your best option is probably to configure ns-upload to upload BG data to NS from the Pi
Rachel Sandlain
@audiefile
Jan 04 2016 23:20
xdrip is an overloaded term. it can mean diy receiver but it's also the name of the app to connect the diy receiver or a share receiver to an android via bluetooth
Tony Zarro
@tzarro
Jan 04 2016 23:22
awesome information as usual, guys...thanks so much!
Rachel Sandlain
@audiefile
Jan 04 2016 23:22
@danamlewis re openaps/docs#54 do you mean how to set up share receiver with android app, share receive to android app to nightscout to pi, diy hardware to android to ns to pi, diy hardware to pi directly?
Scott Leibrand
@scottleibrand
Jan 04 2016 23:22
lol, overloaded indeed!
Dana Lewis
@danamlewis
Jan 04 2016 23:23
I mean add something somewhere to the docs that says it’s possible to openaps to get data from xdrip, then point people to the relevant, multiple places to go fish ;) for what they want
Rachel Sandlain
@audiefile
Jan 04 2016 23:23
lol I'll see what I can do
Dana Lewis
@danamlewis
Jan 04 2016 23:25
Thanks :)
I realized when writing the winter update that a lot of other methods like that, or Ns, or Medtronic cgm , etc aren't referenced
Might could be done on the hardware page with subsections as needed
Toby Canning
@TC2013
Jan 04 2016 23:50
image1.JPG
image2.JPG
Rachel Sandlain
@audiefile
Jan 04 2016 23:51
@danamlewis is this what you're looking for? https://gist.github.com/audiefile/cc7d02793c70112cebd4
Toby Canning
@TC2013
Jan 04 2016 23:51
This is working really well for our child. To keep the setup from getting hot, we underclock the Pi.
We never miss a run though!
I have an intel compute stick and intel edison coming Wednesday to give those a try too.