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

21st
Jan 2016
Ben West
@bewest
Jan 21 2016 00:01
that should have suspended pump
it didn't?
personally, like mm-press-key.py --serial 123456 --init down
which should turn on the light
Daniel Bjørnbakk
@danibjor
Jan 21 2016 00:08
@bewest it didn't. 754 with v2.8B 1.1 0B 0B firmware
Alin Artiom Kenibasov
@ktomy
Jan 21 2016 00:08
when trying to debug mocha tests with visual studio I begin to love .net and mstest more and more.....
Daniel Bjørnbakk
@danibjor
Jan 21 2016 00:09
@ktomy .NET and VS.NET is sweet
Alin Artiom Kenibasov
@ktomy
Jan 21 2016 00:09
...but maybe it is only me...
Daniel Bjørnbakk
@danibjor
Jan 21 2016 00:09
we should move it all over to C# on Win10 IOT on the RPI ;)
Alin Artiom Kenibasov
@ktomy
Jan 21 2016 00:10
aha, and to spend 2 years on that :)
Daniel Bjørnbakk
@danibjor
Jan 21 2016 00:11
hehe - we have all the know-how and algorithms
Lori
@LoriLori
Jan 21 2016 00:11
may be win10 phone would be piece of cake after win10 IoT :)
Ben West
@bewest
Jan 21 2016 00:12
@amazaheri ^^^ ;-)
Alin Artiom Kenibasov
@ktomy
Jan 21 2016 00:12
@LorelaiL ...this is how nokia will return to maket :)
Ben West
@bewest
Jan 21 2016 00:12
I think there are some prototypes already
Lori
@LoriLori
Jan 21 2016 00:13
@ktomy would love to get 6230 back :) 20 days of battery life with bluetooth
Ben West
@bewest
Jan 21 2016 00:13
FWIW, with the cc1111 ti usb stick, the whole thing should work as-is on windows and osx as well
Alin Artiom Kenibasov
@ktomy
Jan 21 2016 00:13
tried to port mm-tools once
managed to talk to stick
Ben West
@bewest
Jan 21 2016 00:13
so you can test/develop on your desktop and just sync your git and extra inis to the pi/windows device
the cc1111 ti usb dongle will work everywhere flawless
Daniel Bjørnbakk
@danibjor
Jan 21 2016 00:13
My Lumia 950XL is ready - and has USB host for the medtronic usb stick
matchbox with the cc1111 + ble4 + small battery
Alin Artiom Kenibasov
@ktomy
Jan 21 2016 00:14
@danibjor aha, try to find generic usb2serial driver there :)
Ben West
@bewest
Jan 21 2016 00:15
for the usb, it's generic CDC/ACM
just works :-)
for the cc1111 usb
Alin Artiom Kenibasov
@ktomy
Jan 21 2016 00:16
uart are treue serial, but ms will never accept changing vid/pid as a config on existing drivers :)
so the solution is usblib or something
so you can access up and down streams of usb
Ben West
@bewest
Jan 21 2016 00:17
pretty sure it's just an INF file
Alin Artiom Kenibasov
@ktomy
Jan 21 2016 00:18
yes, and just a small signature makes it unusable
Ali Mazaheri
@amazaheri
Jan 21 2016 00:18
indeed! it works, I ran a full loop on Windows version of OpenAPS on Windows 10
but I am laser focused on Edison for now
Alin Artiom Kenibasov
@ktomy
Jan 21 2016 00:19
and how did you mapped the sticl?
Ben West
@bewest
Jan 21 2016 00:19
https://gist.github.com/bewest/32ccf16c53dca4183c5e this is distributed with the rfcat source
Ali Mazaheri
@amazaheri
Jan 21 2016 00:19
had to change Ben's code
Ben West
@bewest
Jan 21 2016 00:19
ah, for the classic carelink usb, that's more challenging
but for the yardstick one, I think it just works
Ali Mazaheri
@amazaheri
Jan 21 2016 00:20
agree, TI was a breeze
I have both working but as a POC for now
Alin Artiom Kenibasov
@ktomy
Jan 21 2016 00:21
I am not very aware of alternative connection devices, is there anything that already working?
Ben West
@bewest
Jan 21 2016 00:22
several
there is rileylink over ble, uart, spi, yardstick one, cc1111 usb emk TI dongle
Ali Mazaheri
@amazaheri
Jan 21 2016 00:23
dexcom.PNG
carelink.PNG
Jim Matheson
@jmatheson
Jan 21 2016 00:23
So frustrated with insurance in this country. Been waiting two months for a replacement for my out of warranty dexcom. It is amazing how messed up the process is. #venting
Ben West
@bewest
Jan 21 2016 00:23
all those can use the new subg_rfspy firmware
Dana Lewis
@danamlewis
Jan 21 2016 00:23
::wishes for a gist/ list/writeup of all the devices in progress::
@jmatheson I hear you. working on an appeal for my lovely new insurance that denied me a CGM.
Ben West
@bewest
Jan 21 2016 00:23
only one not done for openeaps in particular is the RL+BLE
Ali Mazaheri
@amazaheri
Jan 21 2016 00:24
@bewest :)
come on Ben!
Ben West
@bewest
Jan 21 2016 00:24
hehehe
Ali Mazaheri
@amazaheri
Jan 21 2016 00:24
are we there yet?
flashing the RL
somehow screwed up the license for BLE113 so just requested a new one
Jim Matheson
@jmatheson
Jan 21 2016 00:25
yeah, I sometimes joke that the hardest part about being diabetic is dealing with the healthcare system. Seems so much more painful then treating the disease sometimes :)
Alin Artiom Kenibasov
@ktomy
Jan 21 2016 00:25
so let me understand, there is a "production-grade" solution that can replace the carelink?
Ali Mazaheri
@amazaheri
Jan 21 2016 00:25
yes
we have TI CC1111 in production now
Alin Artiom Kenibasov
@ktomy
Jan 21 2016 00:26
and does it support eu frequency?
Ali Mazaheri
@amazaheri
Jan 21 2016 00:26
Ben did a TI USB extension which works great
Ben West
@bewest
Jan 21 2016 00:26
ah, snap, EU is still undergoing tuning, which is doable in app/client
Ali Mazaheri
@amazaheri
Jan 21 2016 00:26
u need to order ERF module, ping @oskarpearson
:)
Alin Artiom Kenibasov
@ktomy
Jan 21 2016 00:27
and the last question - does it require soldering?
Ali Mazaheri
@amazaheri
Jan 21 2016 00:27
ERF yes, but you can do it
Ben West
@bewest
Jan 21 2016 00:27
the usb support means you plug into usb
I think yardstick one should support EU with usb
Ali Mazaheri
@amazaheri
Jan 21 2016 00:27
nice
Ben West
@bewest
Jan 21 2016 00:28
I think TI dongle does as well, not sure though
Ali Mazaheri
@amazaheri
Jan 21 2016 00:28
gotta love the smell of soldering
Ben West
@bewest
Jan 21 2016 00:28
seen too many blue ghosts escaping in my youth
Ali Mazaheri
@amazaheri
Jan 21 2016 00:28
lool\
Alin Artiom Kenibasov
@ktomy
Jan 21 2016 00:29
aha, thank you, so in a few weeks I'll give it a try, tired of "preflight fail" :)
Ali Mazaheri
@amazaheri
Jan 21 2016 00:29
:)
with TI we get 100% coverage and no cable
now RL/BLE is something else :)
Alin Artiom Kenibasov
@ktomy
Jan 21 2016 00:36
so PI+TI stick (USB or BT) it a working setup, right?
ah, usb, as "TI" is an usb stick
Scott Leibrand
@scottleibrand
Jan 21 2016 00:36
right
Dana Lewis
@danamlewis
Jan 21 2016 00:41
heh that url
Scott Leibrand
@scottleibrand
Jan 21 2016 00:42

mine ;-)

haha even better
Dana Lewis
@danamlewis
Jan 21 2016 00:43
LOL @scottleibrand
Alin Artiom Kenibasov
@ktomy
Jan 21 2016 00:44
@bewest thank you, I think I'll convince my wife that $150 for a "toy" is not too much :))
Scott Leibrand
@scottleibrand
Jan 21 2016 00:44
TI stick is about half the price, and smaller
so if cost is a concern...
Alin Artiom Kenibasov
@ktomy
Jan 21 2016 00:45
anyway, i'll talk to @oskarpearson about EU frequency first
Scott Leibrand
@scottleibrand
Jan 21 2016 00:45
I ordered a yard stick, but we haven't tested that one yet. @amazaheri did test the TI and it works quite well, even without the antenna and amplifiers the yard stick has
but yeah, frequencies. :)
Alin Artiom Kenibasov
@ktomy
Jan 21 2016 00:54
Thank you all again and have a good night/day :) as for me it's enough js debugging for a day.
Ali Mazaheri
@amazaheri
Jan 21 2016 00:55
:smile:
Jim Matheson
@jmatheson
Jan 21 2016 02:50
Hey @danamlewis, I just saw my former diabetes camp counselor from when I was a kid post a link to your article on facebook…said he was going to give it a try. Pretty awesome to see the reach the group is starting to get :) https://www.raspberrypi.org/blog/artificial-raspberry-pi-pancreas/
Dana Lewis
@danamlewis
Jan 21 2016 02:50
Hehe cool :) it's a small world!
Scott Leibrand
@scottleibrand
Jan 21 2016 02:53
:-) :+1:
Dana Lewis
@danamlewis
Jan 21 2016 02:56
been getting lots of new folks to openaps-dev, and hopefully starting to lurk here as well as they get ready to jump in!
LadyViktoria
@LadyViktoria
Jan 21 2016 03:10
:)
Aaron Michelson
@Aaybob
Jan 21 2016 04:26
@danibjor I typed that command in response to your question of how to find out how to get glucose readings from the medtronic cgm. That command has a lot of options to read glucose.
did my commit to the docs repo happen? I used the gui to create a copy local. Then I clicked on sync and merge. I don't know if anything happened.
I wrote a section and I explain it in the sync comment.
Aaron Michelson
@Aaybob
Jan 21 2016 04:59
how does one use historical data? does that require nightscout setup?
Scott Leibrand
@scottleibrand
Jan 21 2016 05:01
@Aaybob I never saw a PR. Do you have a green "compare and pull request" button or similar on your repo on GitHub?
generally yes, historical data is kept in Nightscout. The openaps tools can pull up to about 30d of data if necessary, but generally we only do a few hours.
Aaron Michelson
@Aaybob
Jan 21 2016 05:05
hi thanks for helping with this repository stuff... i have github docs ; a check mark in a box that says 1 change; under that a check box checked that says docs\getting-started\rpi.md and five green little boxes in a line next to rpi.md. I see that when I click on the box at the top Changes next to History
that's all in GitHub
Aaron Michelson
@Aaybob
Jan 21 2016 05:08
my eyes glaze over
I've tried to understand that document but can't
Scott Leibrand
@scottleibrand
Jan 21 2016 05:10
Have you tried just following the steps?
Just the Initiating part
Aaron Michelson
@Aaybob
Jan 21 2016 05:23
I 'll try that just following the steps.
out
Kevin Wells
@wkcwells
Jan 21 2016 05:31
Running openaps with Dexcom, Share, Minimed pump and Carelink USB. I only have a few days experience, but it’s beginning to seem like USB needs to be reset whenever the MM stick goes out of range of the pump. Does that sound right?
Scott Leibrand
@scottleibrand
Jan 21 2016 05:34
It shouldn't do that every time.
What is your reset-usb conditional on?
Kevin Wells
@wkcwells
Jan 21 2016 05:36
I’m just doing it manually for now as I am trying to get a handle on robustness issues.
The error I get (permanently) after being out of range is decocare.errors.AckError: ACK is 0 bytes:
Which is fixed by the reset.
I’ll do some more experimenting.
Scott Leibrand
@scottleibrand
Jan 21 2016 05:44
Hmm. Maybe that is more common than I realized.
I would definitely put mm-stick warmup || oref0-reset-usb in your preflight if it fails.
Kevin Wells
@wkcwells
Jan 21 2016 05:47
I will get there. So far I’ve had OK reliability without any auto-resets. And great results!! (When the loop is up.) So I am taking advantage of the ability to go slow to get a feel for where things break down.
Dana Lewis
@danamlewis
Jan 21 2016 05:48
:+1:
Kevin Wells
@wkcwells
Jan 21 2016 05:51
@bewest fixed a different exception the other night that I haven’t had time to patch in yet. (zero duration in the temp basal file.) Is the general idea that python exceptions are OK and will be handled by the invoking script etc., or that they should be caught and reported up in the module where they occur (decocare in this case)?
Kevin Wells
@wkcwells
Jan 21 2016 06:00
Based on a trivial amount of experimenting, I get a diffenent exception when the pump is out of range, and that self-recovers when back in-range. And I can recover from the original problem by pulling the stick out and plugging back in. I’ll keep going and writeup what I find.
Scott Leibrand
@scottleibrand
Jan 21 2016 06:07
Exceptions and tracebacks are obnoxiously verbose, but otherwise don't cause issues.
My loop has lots of 2>/dev/null
would be nice to catch exceptions and report more user-friendly error messages, but it hasn't need a priority yet.
I've started doing it in oref0, but I don't touch the Python code.
Kevin Wells
@wkcwells
Jan 21 2016 06:10
Sounds good. I’ll DIY my own path on this for a while. As I said, when it’s all working, it really works! And I love the NS console visualization.
Scott Leibrand
@scottleibrand
Jan 21 2016 06:11
:+1:
Kevin Wells
@wkcwells
Jan 21 2016 06:15
Plus I’m clueless on shell scripting. So I really need to spool up on that. :)
Thomas Emge
@ThomasEmge
Jan 21 2016 07:06
we have TI CC1111 in production now
@amazaheri What do you mean? How would one get this TI CC1111 to replace the gnarly carelink stick?
Thomas Emge
@ThomasEmge
Jan 21 2016 07:43
@bewest tried a different bluetooth dongle today but the results are very much the same. Sometimes Failed to find UART device., sometimes
  File "/usr/local/lib/python2.7/dist-packages/openaps-0.0.9-py2.7.egg/openaps/vendors/dexcom.py", line 91, in main
    for item in self.dexcom.iter_records('EGV_DATA'):
AttributeError: 'NoneType' object has no attribute 'iter_records'
are you using wheezy or jessie on your PI?
Oskar Pearson
@oskarpearson
Jan 21 2016 07:56
Hey @ktomy - the eu frequency for erf isn't a problem. That's what we use. At the moment we've got some issues with certain pump models - but it would be incredibly unlikely that we can't fix it in software
Scott Leibrand
@scottleibrand
Jan 21 2016 07:57
What about the TI stick or yard stick?
any idea if those can do EU frequencies well?
Oskar Pearson
@oskarpearson
Jan 21 2016 08:06
Ti stick definitely does - it's what we used originally here in the uk with success. Yardstick hasn't been tested yet, but I'd be super surprised if it doesn't work
at the moment I don't have proper verification that the erf does USA frequencies with a reasonable range. The devices are designed for the eu range primarily. However from my experience using a USA-designed device in the uk it doesn't seem to be a problem
Scott Leibrand
@scottleibrand
Jan 21 2016 08:08
:+1:
Just opened some new oref0 issues for discussion: https://github.com/openaps/oref0/issues
Oskar Pearson
@oskarpearson
Jan 21 2016 08:10
range seems to be much more about the tuning of parameters on the cc1110/cc1111 device. Thanks to ps2's subg rfspy code that's doable in software
Scott Leibrand
@scottleibrand
Jan 21 2016 08:11
Those new issues relate to some new features I think we might want to consider, so would love input from as many oref0 users as possible.
Jason Calabrese
@jasoncalabrese
Jan 21 2016 08:12
is the carelink doing auto tuning of some type?
I was able to get usb subg_rfspy to kind of work on 1 pump, but not another
Oskar Pearson
@oskarpearson
Jan 21 2016 08:15
@jasoncalabrese i think the reason we get such terrible reception on the carelink is that it's the most general parameters possible, so it'll work on all pumps
i think the mmcommander settings are pretty close to optimal, in that they seem to work with
Ben West
@bewest
Jan 21 2016 08:16
likely
Oskar Pearson
@oskarpearson
Jan 21 2016 08:16
most pumps and give better range
the issue I think we have now is that the radio frequency parameters are "over tuned" - they work very well with some pumps, and not at all with others
I'm going to put some profile parameters into the Mmeowlink code that look at the pump model and then try choose the optimal radio parameters. The next step is to figure out what those parameters are for the different models
it seems there might also be some variance within model numbers too, unfortunately
so we might default to something like mmcommander settings, and then give people the option to try other values
Jason Calabrese
@jasoncalabrese
Jan 21 2016 08:21
maybe a script to test/auto tune?
looking for error rates
Oskar Pearson
@oskarpearson
Jan 21 2016 08:28
Have a look at the mmeowlink and ps2/rileylink channels - there's some info in there about auto-frequency-adjustment and scripts to tune things there
Ben West
@bewest
Jan 21 2016 08:36
would be interesting to work on more tuning scripts
would be absolutely lovely to have some kind of profile + override settings that can be used in openaps
so an autotune thingy could print/save exactly the right config
Jason Calabrese
@jasoncalabrese
Jan 21 2016 08:44
just merged a little pr to fix/improve some things with the pump/openaps plungin based on some things I ran into over the last couple of days
nightscout/cgm-remote-monitor#1481
Dana Lewis
@danamlewis
Jan 21 2016 16:28
@Aaybob Per your Facebook post, do you have Nightscout set up to visualize your loop? And have you checked your logs? This will be where you find it what the loop is doing...
Dana Lewis
@danamlewis
Jan 21 2016 16:35
@Aaybob Here's an example of what NS visualization can look like, so you can see high and low temps from the past: https://twitter.com/danamlewis/status/688050867983888384 and
and the logs are even better because they'll show the reason logic for what the thought process was
Scott Leibrand
@scottleibrand
Jan 21 2016 16:45
You can also see that reason logic if you set up ns-status upload and enable the OpenAPS pill in NS.
Oskar Pearson
@oskarpearson
Jan 21 2016 16:47
I really must get around to sorting out the NS stuff :)
Dana Lewis
@danamlewis
Jan 21 2016 16:51
Definitely worth it! I use that rather than sshing in on the go to see if loop is running and has run recently. (That + watch face)
Rachel Sandlain
@audiefile
Jan 21 2016 16:57
Do it! It could also give an easy way for your wife to check on things while at work
Dana Lewis
@danamlewis
Jan 21 2016 16:58
Yup. One button press for Ns app. Vs logging in requiring keystrokes
Jason Calabrese
@jasoncalabrese
Jan 21 2016 17:04
now that more people are using it, any requests?
I've been thinking about adding more previous suggestions/enacts
Dana Lewis
@danamlewis
Jan 21 2016 17:05
@jasoncalabrese :+1: Also I think getting net iob calculated there, and then the discussion we started last night about alerts
Dragan Maksimovic
@dm61
Jan 21 2016 17:05
@danamlewis I've made an attempt at updating the visualization section - have you seen my PR ?
Scott Leibrand
@scottleibrand
Jan 21 2016 17:05
Would love if the iOS app killed itself when you closed it instead of trying to background.
Would load faster on relaunch starting from scratch.
Dana Lewis
@danamlewis
Jan 21 2016 17:06
@dm61 Seen it come through, yes :) haven't had a chance to review thoroughly yet, will try to do so today!
(And thank you :))
Dragan Maksimovic
@dm61
Jan 21 2016 17:07
no, thanks go to you - I am not super happy about how that section looks like
Scott Leibrand
@scottleibrand
Jan 21 2016 17:07
On the c-r-m side I think IOB is the big one.
Jason Calabrese
@jasoncalabrese
Jan 21 2016 17:09
yes, maybe we should do a quick port first, @sulkaharo started updating dcalc
I like the idea of the loop being able to ask for help
maybe -IOB warnings to start with
Jason Calabrese
@jasoncalabrese
Jan 21 2016 17:16
looking back at results from yesterday, I see some issues around calibrations
maybe we should ignore points before a the last calibration
Scott Leibrand
@scottleibrand
Jan 21 2016 17:25
We need to pull calibrations and raw from Dexcom over USB and upload them to NS.
Jason Calabrese
@jasoncalabrese
Jan 21 2016 17:26
we have them in NS, depends on the uploader
wouldn't work for Share
Scott Leibrand
@scottleibrand
Jan 21 2016 17:26
Yeah, but Pi doesn't know about them
and would be nice not to need android.
Jason Calabrese
@jasoncalabrese
Jan 21 2016 17:27
yeah, think we can pull them with usb, think we just need to add reports
Scott Leibrand
@scottleibrand
Jan 21 2016 17:27
Ya
and document how to use Pi as NS uploader
and of course figure out how to use raw and calibrations in oref0
Jason Calabrese
@jasoncalabrese
Jan 21 2016 17:37
there are 2 things the cal record, and what we call an mbg in NS
you get a mbg every time you enter a bg, but sometime dex won't add a cal
JaysonEwer
@JaysonEwer
Jan 21 2016 17:49
+1 on @scottleibrand comment about being able to get raw through usb.
Ben West
@bewest
Jan 21 2016 19:11
iter_sensor, iter_sensor_hours
Sulka Haro
@sulkaharo
Jan 21 2016 19:17
Did we have support for minimum basal yet? We're seeing unexpected cannula blockages, I suspect the zero temping is causing issues
Dana Lewis
@danamlewis
Jan 21 2016 19:18
Minimum basal?
Scott Leibrand
@scottleibrand
Jan 21 2016 19:18
Just opened an issue you should comment on.
Dana Lewis
@danamlewis
Jan 21 2016 19:18
openaps/oref0#56
Paul Andrel
@stavlor
Jan 21 2016 19:18
guessing he means like a lowest possible basal instead of 0 temping
Dana Lewis
@danamlewis
Jan 21 2016 19:19
^ @sulkaharo That, or something else?
Yes, please comment on the issue. Started a discussion last night but want some broad input and ideas before we figure out the solution(s)
Daniel Bjørnbakk
@danibjor
Jan 21 2016 19:25
Still struggling with the reports.
root@raspberrypi:/home/pi/maylen_openaps_v2# openaps report invoke enact/suggested.json
determine-basal://text/shell/enact/suggested.json
Could not determine last BG time
{"delta":null,"avgdelta":null}
{"duration":0,"rate":50,"temp":"percent"}
{"bolusiob":0,"iob":1.3235715253750002,"activity":0.03365530303030303}
{"max_iob":2,"type":"current","dia":4,"current_basal":2,"max_daily_basal":2,"max_basal":4.4,"min_bg":90,"max_bg":239.4,"sens":36}
Error: could not calculate eventualBG
reporting enact/suggested.json
where do I start looking for strange numbers?
Alin Artiom Kenibasov
@ktomy
Jan 21 2016 19:30
@danibjor , can you please post some last entries from glucose.json?
Daniel Bjørnbakk
@danibjor
Jan 21 2016 19:37
looks like it just got calibration data, and no sensor records
Alin Artiom Kenibasov
@ktomy
Jan 21 2016 19:40
as I can see, in .date you have a ISOxxx format and as I know, determine-basal expects unixtime there.
it can be easily fixed, will include this in my fix
hope today i'll finish debugging an will pr my code :)
Sulka Haro
@sulkaharo
Jan 21 2016 19:41
@danamlewis that. I'll comment on the thread. Our basal during the day is 0.15, which is not enough to open a cannula blockage after a zero temp. We've experimented with blocked cannulas and it looks like 0.3 unit bolus is not enough to unblock whatever is causing the blockage, but 0.5 does cause the insulin to start re-flowing.
Alin Artiom Kenibasov
@ktomy
Jan 21 2016 19:41
but, for you to make it work faster
just a second, i'll tell you the line where to modify
Sulka Haro
@sulkaharo
Jan 21 2016 19:43
Second, it looks like the current logic for dealing with meals is somehow not working for us. I've seen a couple occasions where a flat curve after a meal causes the loop to zero-temp, causing a high. Basically, LCHF style eating breaks the loop.
instead of display_time put date
Dana Lewis
@danamlewis
Jan 21 2016 19:46
@sulkaharo Specifically meal-assist, or not using meal assist?
Daniel Bjørnbakk
@danibjor
Jan 21 2016 19:47
I see the comments mentions Dexcom - we use the medtronic sensor (if that matters)
@ktomy also, I used the quick-installer, so where is the stuff placed on the filesystem?
Sulka Haro
@sulkaharo
Jan 21 2016 19:50
@danamlewis I thought I'd enabled meal-assist BUT for haven't seen it be active in these cases. The zero temping happens around an hour after the bolus, when there's still IOB but BG curve has been near flat.
I only picked it up because the zero-temping happened when the BG was above target.
Daniel Bjørnbakk
@danibjor
Jan 21 2016 19:52
find . did the trick.. i'll try your fix
Dana Lewis
@danamlewis
Jan 21 2016 19:53
@sulkaharo Do you have some logs? Would be helpful to look into what's happening.
Ben West
@bewest
Jan 21 2016 19:54
for medtronic gucose, save your iter_glucose_hours as glucose-raw.json
Sulka Haro
@sulkaharo
Jan 21 2016 19:54
Sadly not from the Edison. I'll re-enable the git committing.
Dana Lewis
@danamlewis
Jan 21 2016 19:55
K.
Ben West
@bewest
Jan 21 2016 19:56
run the timezone tool, tz rezone --date --date save as glucose-raw-zoned.json, mm-format-ns-glucose --oref0 glucose-raw-zoned.json, save as glucose.json
Sulka Haro
@sulkaharo
Jan 21 2016 19:56
So wish we had event logging that didn't break the loop. Should I just start directing the crontab execution logging into a file? I guess that might have enough data to check what's happened.
Alin Artiom Kenibasov
@ktomy
Jan 21 2016 19:57
@bewest thank you, I didn;t knew there is already-made solution
Ben West
@bewest
Jan 21 2016 19:57
running manually doesn't reproduce?
the design of openaps is modular so it can handle different glucose from different places :-)
Dana Lewis
@danamlewis
Jan 21 2016 19:58
@sulkaharo How does logging break your loop?
Sulka Haro
@sulkaharo
Jan 21 2016 20:07
By logging I mean git committing each cycle. Don't know why, but I've experienced git repo corruption on every time a Pi battery run down or was accidentally disconnected with it enabled. Guess I could add the repo clearing script to the loop, but I suspect that'd just result in increased battery consumption and the repo being reset all the time, for no gain.
Jason Calabrese
@jasoncalabrese
Jan 21 2016 20:08
do you have the reset script in your cron?
Sulka Haro
@sulkaharo
Jan 21 2016 20:10
Nope, due to reasons above. I don't see how it'd not run all the time, which would just lead to no history and faster battery consumption.
Jason Calabrese
@jasoncalabrese
Jan 21 2016 20:10
I resisted adding it for awhile, but now that it first tries to fix the repo first before wiping it seems to do really good
Sulka Haro
@sulkaharo
Jan 21 2016 20:11
I've tried using the repo fix every time, has worked maybe 1 time out of 5.
Again, could just be I've been unlucky.
Jason Calabrese
@jasoncalabrese
Jan 21 2016 20:11
I've only run it manually a few times and it's always worked
Ben West
@bewest
Jan 21 2016 20:12
@sulkaharo if you want to tar/gzip your openaps instance, I might take a look at it to see what's going on
Sulka Haro
@sulkaharo
Jan 21 2016 20:12
Strange. I'm near always left with something that's borked. Sometimes pushing works regardless, often not.
Ben West
@bewest
Jan 21 2016 20:13
I usually look at the reflog
Sulka Haro
@sulkaharo
Jan 21 2016 20:13
@bewest which one? :D I have three.
Ben West
@bewest
Jan 21 2016 20:13
the broken one
or the most broken one
or the most curiously broken one
Sulka Haro
@sulkaharo
Jan 21 2016 20:14
Sec, I'll check if I have git breakage on any right now.
Ben West
@bewest
Jan 21 2016 20:14
or next time it happens
Sulka Haro
@sulkaharo
Jan 21 2016 20:14
There's one in the parent bed, on in the dude's bed, one that's mobile.
Ben West
@bewest
Jan 21 2016 20:15
since I wrote the git fix script, none of mine have corrupted
Sulka Haro
@sulkaharo
Jan 21 2016 20:15
Guess I might have been using a in-development version
Ben West
@bewest
Jan 21 2016 20:15
clearly the solution is to write a script to fix it so that it never happens again
;-)
Sulka Haro
@sulkaharo
Jan 21 2016 20:15
At which point during the loop do you run it? Just once in the beginning?
Ben West
@bewest
Jan 21 2016 20:16
no I mean, having written it will ward off the forces that might requires it's use, apparently
Sulka Haro
@sulkaharo
Jan 21 2016 20:16
:D
Dana Lewis
@danamlewis
Jan 21 2016 20:16
Lol
Ben West
@bewest
Jan 21 2016 20:16
it's like raining and having your car washed
Sulka Haro
@sulkaharo
Jan 21 2016 20:17
opposite of car washing and bird relief?
Ben West
@bewest
Jan 21 2016 20:17
inverse maybe, yeah
not sure where scott and dana put theirs
maybe after or before preflight
check if git is broken, if so do this fix thingy, if not continue
Sulka Haro
@sulkaharo
Jan 21 2016 20:18
hitting the car insurance bonus cap seems to correlate to my wife finding a way to drop me down a few notches, too
Ben West
@bewest
Jan 21 2016 20:18
git status && echo "GIT OK" || echo "GIT BROKEN"
Sulka Haro
@sulkaharo
Jan 21 2016 20:19
hey easier way to do if statements in OpenAPS would be AWESOME
bash scripting with cron is a bit strange to me as a state machine
Jason Calabrese
@jasoncalabrese
Jan 21 2016 20:23
don't think it can be in the loop it self
since the openaps tool won't work if git is screwed up
I'm doing something similar to @scottleibrand and putting it in my cron * * * * * cd /path/to/repo && oref0-reset-git")
@sulkaharo ^^
Sulka Haro
@sulkaharo
Jan 21 2016 20:33
Yup
Adding
Dana Lewis
@danamlewis
Jan 21 2016 20:33
:+1:
Tony Zarro
@tzarro
Jan 21 2016 20:36
Just checking in with a question and a comment regarding insulin pump battery life: it seems like my minimed pump is eating through AAA batteries like never before...haven't tracked it but its something like 3-4 days...is that common?
Sulka Haro
@sulkaharo
Jan 21 2016 20:36
Yes
You need to get the absolutely best batteries you can find, and it's still going to burn through those pretty fast
Tony Zarro
@tzarro
Jan 21 2016 20:37
Also, I noticed that when the battery life is to the point where it will no longer allow turning on the light on the pump, I get constant errors so I imagine its also not allowing proper communication to the CareLink at that point
Sulka Haro
@sulkaharo
Jan 21 2016 20:37
Hence my earlier interest in not fetching everything from the pump every time
We could get some data like the profile only out every 30 minutes
Tony Zarro
@tzarro
Jan 21 2016 20:37
No worries, a very small price to pay for a working pancreas :)
Scott Leibrand
@scottleibrand
Jan 21 2016 20:37
@tzarro Energizer Lithiums will last a week plus.
Tony Zarro
@tzarro
Jan 21 2016 20:38
@scottleibrand just what I was about to ask :)
Scott Leibrand
@scottleibrand
Jan 21 2016 20:38
They die fast though. Three bars to dead in hours.
Jason Calabrese
@jasoncalabrese
Jan 21 2016 20:38
the new NS pump plugin can send alerts based on battery level
Sulka Haro
@sulkaharo
Jan 21 2016 20:38
Yeah, those are awesome. A month or more when not looping
But yes, with lithium you have to replace immediately at first sign of bars dropping
This is due to the voltage staying up until the battery is depleted, unlike NiMh
Tony Zarro
@tzarro
Jan 21 2016 20:39
@sulkaharo great info, thank you!
Sulka Haro
@sulkaharo
Jan 21 2016 20:39
Pump doesn't have the capability to measure a li-ion
Jason Calabrese
@jasoncalabrese
Jan 21 2016 20:40
the voltage is weird, seen it go up after some usage
Sulka Haro
@sulkaharo
Jan 21 2016 20:41
chemistry <3
Dana Lewis
@danamlewis
Jan 21 2016 20:41
haha.
Sulka Haro
@sulkaharo
Jan 21 2016 20:43
Still kind of blown away by our endo being supportive of OpenAPS. Was expecting the Spanish Inquisition regarding safety.
Dana Lewis
@danamlewis
Jan 21 2016 20:44
@sulkaharo that's awesome :)
Sulka Haro
@sulkaharo
Jan 21 2016 20:44
But she's T1 herself, so...
Didn't see her for a year, as she had a baby. She said her sugars behaved like a 3 year old T1 during the first trimester, then like 8 year old during second and then like a teen during third. :D
Eric
@ecc1
Jan 21 2016 20:46
"Our three main weapons are insulin pumps, CGMs, and an almost fanatical devotion to openaps ..." :-)
Sulka Haro
@sulkaharo
Jan 21 2016 20:46
:D
Dana Lewis
@danamlewis
Jan 21 2016 20:46
@tzarro Along with switching to lithium, I send myself email alerts to track the voltage, so you get the first warning of it dropping a bar & know you will need to replace then or within a few hours. (Don't go to bed if you've gotten an alert, because it'll be zero overnight and no openaps and thus sad sad sad panda.)
Probably something to add to the list for alerts we might want to have people be able to configure, but in the meantime, let me know if you want info on how to
Sulka Haro
@sulkaharo
Jan 21 2016 20:47
Her HbA1c is 5.8% - respectable and makes me believe she's on top of things. She's tried all the equipment herself, too. All CGMs on market, including Libre, so she can make recommendations. She's now on 650G and says it's awesome for her.
Apparently she only really needs the low prediction suspend with her diet and metabolism.
Dana Lewis
@danamlewis
Jan 21 2016 20:48
nice!
Sulka Haro
@sulkaharo
Jan 21 2016 20:51
Somehow have the feel that even endos who've helped T1s for a decade have no clue if they don't have first hand daily experience what it's like. We've seen five different people and the T1 one is amazing, one with 30 years of experience was really good, other three were just out of touch. Sound familiar to others?
Something of interest in addition - the endo wants to do deeper investigation into Eero's subtype. We're closing on three years since diagnosis and the antibodies are still negative and insulin need is < 0.5 units / kg / day, which is indicative of our T1 not being a common form. Will learn more in three months time or so.
Alin Artiom Kenibasov
@ktomy
Jan 21 2016 20:56
avgdelta is per minute or per 5 minutes?
Scott Leibrand
@scottleibrand
Jan 21 2016 20:56
per 5m
Alin Artiom Kenibasov
@ktomy
Jan 21 2016 20:56
thank you
Jason Calabrese
@jasoncalabrese
Jan 21 2016 20:57
our Endo is really supportive, but I should still get consulting fees for our appointments
Dana Lewis
@danamlewis
Jan 21 2016 20:57
@sulkaharo yea, I think it widely varies. My first endo (peds) was AMAZING, although she didn't have T1. The rest of my endos have been…ok, ok, and meh. (None have t1, and the rest of these had been adult endos). But I've also heard stories from friends who have adult endos with T1 that are great, and some with T1 that are really about their way or the highway :(
Scott Leibrand
@scottleibrand
Jan 21 2016 20:57
@jasoncalabrese hehe
Dana Lewis
@danamlewis
Jan 21 2016 20:58
@jasoncalabrese :)
Jason Calabrese
@jasoncalabrese
Jan 21 2016 20:59
he was impressed with how well behaved the loop was, spent some time reviewing the previous day via NS
Ben West
@bewest
Jan 21 2016 21:00
yeah NS is very very good about showing what's happening
Jason Calabrese
@jasoncalabrese
Jan 21 2016 21:00
seeing it real-time while in the office was nice too
Sulka Haro
@sulkaharo
Jan 21 2016 21:01
Ha, I've only gotten excited oggling at the NS graphs, followed by the endo switching back to the MedTronic reporting system
Jason Calabrese
@jasoncalabrese
Jan 21 2016 21:01
think Andrew went low for almost every other appt, always a bad time and long wait, etc
this time it was a perfect flat line, the whole time
maybe more a perfect rolling line
so there was activity from the loop, but just enough
Sulka Haro
@sulkaharo
Jan 21 2016 21:02
LOL yes I remember this one appointment pre-CGM when Eero went lying down on the floor looking pale. I recall we measured BG at 1.8, which resulted in some pretty urgent glucose ingestion.
No idea what he'd need to be at to pass out. Under 3 mmol values don't seem to result in anything noticeable in his behavior. If anything, looks like he's even more focused than usual when going low.
Jason Calabrese
@jasoncalabrese
Jan 21 2016 21:08
we had a 36 waiting for an appointment before, was pre-NS and had the dex 7+, he was running around bouncing of the walls
Alin Artiom Kenibasov
@ktomy
Jan 21 2016 21:34
I am ready for PR, only one thing - I have introduced a npm library for interpolation and I think it should be "told" somewhere (installation scripts)
Scott Leibrand
@scottleibrand
Jan 21 2016 21:34
do we need a library for that?
Alin Artiom Kenibasov
@ktomy
Jan 21 2016 21:34
i have it in package.json dependencies
  • "natural-spline-interpolator": "^1.0.2",
Ben West
@bewest
Jan 21 2016 21:34
yeah, perfect
Alin Artiom Kenibasov
@ktomy
Jan 21 2016 21:35
but is it the only place to put it?
Ben West
@bewest
Jan 21 2016 21:35
did you publish it somewhere?
you can do npm install --save natural-spline-interpolator
and that will save it/register it in package.json for you
Scott Leibrand
@scottleibrand
Jan 21 2016 21:35
it's an existing npm package right?
ah, cool
Alin Artiom Kenibasov
@ktomy
Jan 21 2016 21:36
it is a package from global npm store it;s not ine
Ben West
@bewest
Jan 21 2016 21:36
if it's not on npm yet, you can use your github repo url
ah
yeah, you can just install --save it in oref0 on a dev branch
maybe dcalc
we should cut over to dcalc at some point
Alin Artiom Kenibasov
@ktomy
Jan 21 2016 21:36
so in this case (globally-existant npm), should I put it somewhere?
Ben West
@bewest
Jan 21 2016 21:36
yes, in dev branch of oref0
do npm install --save natural-spline-interpolator
that saves it in package.json automatically
then you can git diff or git commit -av to save it in git
Alin Artiom Kenibasov
@ktomy
Jan 21 2016 21:37
aha, so I should get it locally and push it to git
Ben West
@bewest
Jan 21 2016 21:38
you should register it as dependency in oref0
so the code in oref0 can use it
when oref0 gets installed, this package will be installed as dependency automatically
Alin Artiom Kenibasov
@ktomy
Jan 21 2016 21:38
it is registered in package.json
Ben West
@bewest
Jan 21 2016 21:38
then anywhere in oref0 you can get it with var interp = require('natural-spline-interpolator');
Alin Artiom Kenibasov
@ktomy
Jan 21 2016 21:38
but the code of this library is ...somewhere on npm global store
Nyadach
@Nyadach
Jan 21 2016 21:38
hi all, just working through initial setup, and on phase 2 now, but find when running gather-profile it returns that it can't parse carbratio_data? thought it might be needing the carb ratios off the pump so created a report for them, but doesn't seem to fix it. Mind me asking what I'm missing please?
Ben West
@bewest
Jan 21 2016 21:39
npm packages install all their packages in ./node_modules on a per project basis
the package.json controls which outside projects get placed in there automatically
Jason Calabrese
@jasoncalabrese
Jan 21 2016 21:39
does that understand time? we'd need 5 minute, 15 minutes deltas, etc
Scott Leibrand
@scottleibrand
Jan 21 2016 21:39
@Nyadach you may be using an old version of oref0
Alin Artiom Kenibasov
@ktomy
Jan 21 2016 21:40
aha, ok, so it will search and download it automaticaly when you'll get last version
Ben West
@bewest
Jan 21 2016 21:40
the npm install --save step puts it in package.json
Scott Leibrand
@scottleibrand
Jan 21 2016 21:40
@Nyadach and/or your reports were set up for an older version
Ben West
@bewest
Jan 21 2016 21:40
whenever npm install happens for oref0, npm manages all that automatically
so in the oref0 project youcan use it anywhere with require('natural-splint-interpolator') to get whatever that project exported
Alin Artiom Kenibasov
@ktomy
Jan 21 2016 21:41
ok, thank you, so I'll make PR and automated testing system will tell if I did something wrong :)
Nyadach
@Nyadach
Jan 21 2016 21:41
just working through the setup manual, is the new one mentioned elsewhere?
Scott Leibrand
@scottleibrand
Jan 21 2016 21:41
it's just a question of how long ago you did it
things have been updated in the past few months. if you did everything in the last couple weeks, that's probably not it.
Ben West
@bewest
Jan 21 2016 21:41
@Nyadach you created a report for the carb ratios?
it should show up in output of openaps report show
Nyadach
@Nyadach
Jan 21 2016 21:42
using the read_carb_ratios yes, saved that in the settings dir
Ben West
@bewest
Jan 21 2016 21:42
ah, ok
ah, so openaps report show settings/read_carb_ratios.json should show something
Scott Leibrand
@scottleibrand
Jan 21 2016 21:43
you can check your version of oref0 by doing npm show oref0 | grep version:
Nyadach
@Nyadach
Jan 21 2016 21:43
told it to save as settings/carbratio_data.json
Ben West
@bewest
Jan 21 2016 21:43
or even npm show oref0 version
Nyadach
@Nyadach
Jan 21 2016 21:44
as it was looking for that in the error
Ben West
@bewest
Jan 21 2016 21:44
ah I see
ok, so openaps report show settings/carbratio_data.json
Nyadach
@Nyadach
Jan 21 2016 21:45
has everything off the pump
Ben West
@bewest
Jan 21 2016 21:45
curious what it shows if you cat that file: cat settings/carbratio_data.json can you paste it here?
probably fails with error if there is no file yet
the report add from earlier just saves some config
we have toinvoke the report to actually fetch the data
Alin Artiom Kenibasov
@ktomy
Jan 21 2016 21:47
Not sure I did it right, but please sii this: openaps/oref0#54
Ben West
@bewest
Jan 21 2016 21:48
might try openaps report invoke settings/carbratio_data.json to actually get that data
Alin Artiom Kenibasov
@ktomy
Jan 21 2016 21:48
I hope my new commits are there
Nyadach
@Nyadach
Jan 21 2016 21:48
not easily, as dif machine, but:
{
"units": "grams",
"first": 1,
"schedule": [
{
"q" : 0,
"start": "00:00:00",
"r": 180,
"ratio": 18.0,
"offset": 0,
"i": 0,
"x": 0
}
Ben West
@bewest
Jan 21 2016 21:48
yeah looks good
so any time carb ratios are mentioned in the future, you need to use your name for it: settings/carbratio_data.json going forward
Nyadach
@Nyadach
Jan 21 2016 21:50
ok, and the gather-profile, seems to request it, but "Warning: could not parse carbratio_data. Meal Assist disabled."
Scott Leibrand
@scottleibrand
Jan 21 2016 21:51
oh
that is just a warning saying you haven't set up meal assist yet
Nyadach
@Nyadach
Jan 21 2016 21:51
is that, as no settings/ expecting it in the base dir?
ah
Scott Leibrand
@scottleibrand
Jan 21 2016 21:51
if you want to do so, I have a gist for that
Ben West
@bewest
Jan 21 2016 21:51
should the default setup error, then?
or is meal assist now default?
Scott Leibrand
@scottleibrand
Jan 21 2016 21:52
meal assist is default in dev only
we haven't decided how to enable/disable it: we'll have to do that before it goes to master
so it gives you a warning (not an error) if you haven't set it up
Nyadach
@Nyadach
Jan 21 2016 21:53
ah, so I could just roll on and ignore the warning?
Scott Leibrand
@scottleibrand
Jan 21 2016 21:53
yes
Ben West
@bewest
Jan 21 2016 21:53
sounds like it should simply do nothing
eg, in NS we dont' issue warnings if careportal is not setup, etc
Scott Leibrand
@scottleibrand
Jan 21 2016 21:53
we'll probably have an enable option for it before going to master
Ben West
@bewest
Jan 21 2016 21:53
it's an optional feature, or is more intended
Nyadach
@Nyadach
Jan 21 2016 21:53
great...but meal assist is something that can be enabled later on once everything else is up and running?
Scott Leibrand
@scottleibrand
Jan 21 2016 21:54
at that point, if the feature is enabled, but reports are missing, we'd give this warning
but if the feature is not enabled, we'd silently continue
@Nyadach correct
Nyadach
@Nyadach
Jan 21 2016 21:54
great, thanks guys
Scott Leibrand
@scottleibrand
Jan 21 2016 21:54
if we now have newbies setting themselves up on dev, maybe we need to accelerate that a bit. :)
Nyadach
@Nyadach
Jan 21 2016 21:55
is there a different version somewhere?
might be worth me rolling back to
Scott Leibrand
@scottleibrand
Jan 21 2016 21:56
quick-packages installs master, and older more stable version. but dev is pretty well tested at this point
Alin Artiom Kenibasov
@ktomy
Jan 21 2016 21:57
yes-yes, I learned to use should :)
Scott Leibrand
@scottleibrand
Jan 21 2016 21:57
dev is a bit quicker to low-temp if you're projected to go low, and will high-temp slightly more aggressively if you're above about 10 mmol
Nyadach
@Nyadach
Jan 21 2016 21:57
big issue I had in earlier phase was it not accepting pump running in mmol's, seemed as soon as switched it to mg/dl it cleared a few issues too
Ben West
@bewest
Jan 21 2016 22:01
I was curious if you were using mmol
there's a tool to convert
the units tool
it has tool to convert insulin_sensitivities and bg_targets to mg
Nyadach
@Nyadach
Jan 21 2016 22:05
would that auto convert the bg_targets.json etc? or do those need manual editing once units figured?
Scott Leibrand
@scottleibrand
Jan 21 2016 22:12
you'll need to set up a report that uses units to convert bg_targets_raw.json (in mmol) to bg_targets.json (in mg/dl) or similar
Alin Artiom Kenibasov
@ktomy
Jan 21 2016 22:14
bg_targets_raw.json (in mmol) to bg_targets.json (in mg/dl)
Scott Leibrand
@scottleibrand
Jan 21 2016 22:14
sorry, yes. edited
Nyadach
@Nyadach
Jan 21 2016 22:14
ah right
am loving the work that's been done on this though, really is a great project
Alin Artiom Kenibasov
@ktomy
Jan 21 2016 22:16
@scottleibrand please, if you are giong to accept my PR, test it first on real data as I am not home for some days and don't have real test data. Thank you
Scott Leibrand
@scottleibrand
Jan 21 2016 22:17
k
I'll probably look at it this weekend
Alin Artiom Kenibasov
@ktomy
Jan 21 2016 22:17
ok, thank you
Scott Leibrand
@scottleibrand
Jan 21 2016 22:17
I may not be able to fully test it, though, without a MDT CGM
so I'll probably merge it to a new branch for further testing by MDT CGM users
but I need to actually look at it first. :)
Alin Artiom Kenibasov
@ktomy
Jan 21 2016 22:20
don't worry, if you will commit it to dev wihout MDT CGM testing, @bewest will tell us in next 30 seconds :)
Scott Leibrand
@scottleibrand
Jan 21 2016 22:21
;-)
Toby Canning
@TC2013
Jan 21 2016 22:41
@bewest I was just looking at openxshareble documentation and it doesn't list G5 support. Is that right?
Jason Calabrese
@jasoncalabrese
Jan 21 2016 22:43
yeah, only g5 is Nate's xDripG5 ios library
Toby Canning
@TC2013
Jan 21 2016 22:47
Thanks Jason
Ben West
@bewest
Jan 21 2016 23:58
correct, does not do g5
I need a g5 system to implement it for openaps