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

23rd
Mar 2016
Tim Howard
@tghoward
Mar 23 2016 00:03
hey @scottleibrand , just confirming, are you using openxshareble and talking to dexcom G4 with share receiver with an edison?
both @bewest and I were having troubles a few days ago :smile:
Dana Lewis
@danamlewis
Mar 23 2016 00:04
@tghoward yes, I am using openxshareble to the Edison.
Also using Share data from Nightscout, too
I had issues last week, but think it was my phone that stopped working sporadically for bluetooth rather than the receiver, since it’s working again with new phone
@garykidd are you working from this for guiding NS setup? https://github.com/openaps/docs/blob/dev/docs/Automate-system/vizualization.md
Scott Leibrand
@scottleibrand
Mar 23 2016 00:07
we have openxshareble working for the most part, but when her Share app was dead and we started trying to rely on it, it wasn't reliable enough. had to go back to Pi with plugged-in Dexcom.
Tim Howard
@tghoward
Mar 23 2016 00:07
@danamlewis good. thanks. That's good to know. Spent a lot of debugging time Sunday ... I think it may be time to back up and start fresh somewhere
Scott Leibrand
@scottleibrand
Mar 23 2016 00:07
so our current config is to try NS first, and only try openxshareble if there's no recent data there.
Tim Howard
@tghoward
Mar 23 2016 00:08
@scottleibrand, ok, that make sense.
We tend to have more wifi connectivity issues, so I was hoping to focus more on the offline mode
garykidd
@garykidd
Mar 23 2016 00:09
using azure
@tghoward ...using azure
Tim Howard
@tghoward
Mar 23 2016 00:11
Is that what you meant by installing the dev version? By re-deploying on Azure?
garykidd
@garykidd
Mar 23 2016 00:11
@danamlewis was using that guide yes,,,,,and following links...am I to use the update my fork tool?
@tghoward I believe so, yes.
Tim Howard
@tghoward
Mar 23 2016 00:13
So you can check on your Azure site what your current deployment is (e.g. master or branch and what commit)
Dana Lewis
@danamlewis
Mar 23 2016 00:13
Can someone remind me who is working on, or who has implemented meal-assist lately? With the latest merge to dev, I think we need to get info into the docs about meal assist, was wondering if someone wanted to volunteer to take the setup gist (https://gist.github.com/scottleibrand/63a8541397378bf9f4b8) and put a PR in to docs, similar to the nice work @amazaheri did on how to set up auto sensitivity mode (https://github.com/openaps/docs/blob/dev/docs/Build-manual-system/Configure%20Automatic%20Sensitivity%20Mode.md) :)
garykidd
@garykidd
Mar 23 2016 00:13
was justtrying to set up NS such that it could interface with openaps ...that is once I get it set up.
and how to you switch to dev branch---sorry if I'm misuing lingo
Tim Howard
@tghoward
Mar 23 2016 00:15
@garykidd , yes, so you need to deploy the dev version, that's right
Scott Leibrand
@scottleibrand
Mar 23 2016 00:17
You only need dev branch of NS for that. No need for dev of openaps or oref0.
Tim Howard
@tghoward
Mar 23 2016 00:17
Step 3 on this page:
http://www.nightscout.info/wiki/welcome/monitor-cgm-web
has one step where you choose master. At that spot, choose dev instead
Tim Howard
@tghoward
Mar 23 2016 00:22
@danamlewis we have meal-assist running but it rarely gets implemented from what I can tell. We aren't the best guinea pigs, I guess! (And rarely implemented by no fault of the code, but the user! )
garykidd
@garykidd
Mar 23 2016 00:35
think i did it...succesfully redeployed but NS looks exactly the same
Tim Howard
@tghoward
Mar 23 2016 00:35
nice!
Dana Lewis
@danamlewis
Mar 23 2016 00:35
@garykidd in settings you’ll need to turn on everything you want to display
@tghoward no worries. you could still help write the setup if you felt like it ;) though
garykidd
@garykidd
Mar 23 2016 00:36
@danamlewis thx...forogot that
Dana Lewis
@danamlewis
Mar 23 2016 00:36
no worries! step by step :)
Tim Howard
@tghoward
Mar 23 2016 00:36
:smile:
Dana Lewis
@danamlewis
Mar 23 2016 00:38
Just wanting someone to write it up as English :)
garykidd
@garykidd
Mar 23 2016 00:40
its working now. thank you.
Dana Lewis
@danamlewis
Mar 23 2016 00:40
woot!
garykidd
@garykidd
Mar 23 2016 00:41
and looks like my profile data from before---is there--it just doesnt show up on th profile page itself---does that sound right?
Tim Howard
@tghoward
Mar 23 2016 00:44
Aaak! I got Dexcom data via openxshareble!! woohoo!
just needed to re-install Adafruit and openxshareble.
Dana Lewis
@danamlewis
Mar 23 2016 00:45
Woot @tghoward!
Tim Howard
@tghoward
Mar 23 2016 00:45
Must have done something out of order first time.
garykidd
@garykidd
Mar 23 2016 00:45
I just got the NS set up initially about a week ago...and running on a pebble...and encouraged/determined to loop...but even being able to glance at my wrist and instantly know where I am out is such a n=major step. Thank you--you are not only pioneers but also bodhisattva
Tim Howard
@tghoward
Mar 23 2016 00:45
very exciting.
@garykidd yes, we don't have watches yet, I think it would be a good advancement too. Glad you are doing it.
garykidd
@garykidd
Mar 23 2016 00:49
i really love it. and definitely works better than my glucowatch did...anyone ever try one?
also recently opened and set up g5 ...was using enlite...g5 rocks IMO.
@danamlewis @scottleibrand @tghoward THANK YOU!
Eric Luhrs
@eluhrs
Mar 23 2016 01:04
@tghoward can
@tghoward can you explain openxsharable? Ive seen others mention it but im not sure what it dies.
Tim Howard
@tghoward
Mar 23 2016 01:06
If you have dexcom G4 plus share, it allows Bluetooth (ble) communication between the g4 receiver and your gadget.
Eric Luhrs
@eluhrs
Mar 23 2016 01:07
@danamlewis I will be looking at meal-assist next, probably starting tomorrow. Will add what i can via PR.
@tghoward wow. Thats great. Are there any docs? Does it work with the g5 too? I have both, so guess im just curious.
Tim Howard
@tghoward
Mar 23 2016 01:08
Signing off for the night. Take care everyone.
@eluhrs on my phone now, but yes, some docs
Not for g5, I think
Eric Luhrs
@eluhrs
Mar 23 2016 01:09
@tghoward very cool. I'll leep an eye out for docs.
Dana Lewis
@danamlewis
Mar 23 2016 01:32
@eluhrs excellent, thank you!
And re openxshareble, it's one of @bewest 's awesome tools so look in his repo's.
Scott Leibrand
@scottleibrand
Mar 23 2016 01:34
No G5 BLE from Linux yet. @loudnate got it working from iOS though.
Eric Luhrs
@eluhrs
Mar 23 2016 01:37
Very cool. I've got my g5 working with a cable, but I'm eager to get BTLE working for my g4.
Milos Kozak
@MilosKozak
Mar 23 2016 06:01
blob
hard night for openaps but morning at 6.6 mmol ..... coooool
Matthias
@ceben80
Mar 23 2016 06:03
:clap:
Bill Stackpole
@billstackpole
Mar 23 2016 06:15
does the pump needs to be in units mode rather than percent of basal mode in order for the temp basal to enact?
Scott Leibrand
@scottleibrand
Mar 23 2016 06:19
Yes
Scott Leibrand
@scottleibrand
Mar 23 2016 06:43
@ceben80 with regards to using the TI stick with openaps, pretty much everything is documented in the mmeowlink wiki.
Matthias
@ceben80
Mar 23 2016 06:46
Oki - Then I probably had tomatoes on the eyes..... :smile:
Scott Leibrand
@scottleibrand
Mar 23 2016 06:50
There's also a mmeowlink gitter channel.
Matthias
@ceben80
Mar 23 2016 06:53
Thanks, I know....Thought I would have found nothing how to declare the TI as an device for the pi (for Carelink replacment). I'll looking at the mmeowlink wiki.....
Ben West
@bewest
Mar 23 2016 06:59
absolute vs percent
Bill Stackpole
@billstackpole
Mar 23 2016 07:55
@bewest - thank you. With that, it looks like my loop is working. All I need to do now is create the cron job (but it's late in EDT, I'm tired, so will plan to do that tomorrow). Very exciting stuff! I may want you to review my logic (tomorrow). Seems like I have some redundant instructions and would appreciate another pair of eyes on this. Will also want to better understand my git options for review / replay / etc. Thanks again for the input and help. B.
Scott Leibrand
@scottleibrand
Mar 23 2016 07:57
:-)
Bill Stackpole
@billstackpole
Mar 23 2016 07:58
looking at someone elses commentary they have something about "status upload" and "upload recent treatments" in their cron - will want to ask about those too. Gnite for now!
Scott Leibrand
@scottleibrand
Mar 23 2016 07:59
That would be what powers the NS OpenAPS pill and the NS temp basal viz, respectively.
Night
Bill Stackpole
@billstackpole
Mar 23 2016 11:52
trying to get my loop to run in cron and getting an error "openaps: not found".
crontab -e used to edit crontab as pi user
crontab entry:
/5 * (cd /home/pi/billopenaps/ && openaps retry-loop 2>&1 | logger -t openaps-retry-loop)
whoops - once more with QUOTES...
''' /5 * (cd /home/pi/billopenaps/ && openaps retry-loop 2>&1 | logger -t openaps-retry-loop)
hmm
"*/5"
"/5 * (cd /home/pi/billopenaps/ && openaps retry-loop 2>&1 | logger -t openaps-retry-loop) "
nope - doesnt like my stars (star/5 star star star star (cd /home/pi/billopenaps/ && openaps retry-loop 2>&1 | logger -t openaps-retry-loop)
there.
if I run as pi user at cmd line, runs fine (tail -f /var/log/messages shows entries being added, decisions being made, and temp-basals set or unset accordingly)
Colin Lennon
@colinlennon
Mar 23 2016 11:58
Hi @billstackpole , I was facing the same issue last night so I've been looking into this. Try putting /usr/bin/local/ in front of openaps - so -
(cd /home/pi/billopenaps/ && openaps retry-loop 2>&1 | logger -t openaps-retry-loop)
I think you can remove those opening and closing brackets too
So try -
cd /home/pi/billopenaps/ && /usr/bin/local/openaps retry-loop 2>&1 | logger -t openaps-retry-loop
Bill Stackpole
@billstackpole
Mar 23 2016 12:01
doing that as we speak (will go back and "which openaps" as well to be sure my path is same as yours - tnx!)
yup - it's /usr/local/bin (switcheroo the local and bin)
Colin Lennon
@colinlennon
Mar 23 2016 12:04
Ah okay! :)
I'll try this when I get home from work tonight, I think that's all I'm missing too. I'm not looping yet, but the plan is to run a cron job overnight tonight to generated suggestions and log them to a file which I can review in the morning.
Bill Stackpole
@billstackpole
Mar 23 2016 12:08
not yet working for me - rebooting pi
Colin Lennon
@colinlennon
Mar 23 2016 12:09
Did you get any output in your openaps-retry-loop log file?
Bill Stackpole
@billstackpole
Mar 23 2016 13:37
nope. Some kind of permission error. Now my sensor has expired so I have to wait at least two hrs. Whee.
gonna try to run mine today on battery once my sensor is back up and running. Backed all up via tar to another machine, so should be able to recover should battery die and corrupt flash card. Will advise later.
Colin Lennon
@colinlennon
Mar 23 2016 13:52
Good luck!
Eric Luhrs
@eluhrs
Mar 23 2016 15:06
Anyone have suggestions for dealing with this mongodb error I'm getting on heroku?
2016-03-23T14:58:50.615590+00:00 app[web.1]: Error connecting to MongoDB: {"name":"MongoError","message":"connect ECONNREFUSED"} - retrying in 18 sec
was working fine and I haven't changed anything...
kameamea
@kameamea
Mar 23 2016 15:06
the mongo service may just be down temporarily
Eric Luhrs
@eluhrs
Mar 23 2016 15:08
Hmmm. I'm able to view my tables on mlab.com...
Okay, it's back. Was down for at least 30m, which seems odd.
Scott Leibrand
@scottleibrand
Mar 23 2016 16:20
You can also specify a PATH at the top of your crontab file rather than fully qualifying every command.
Colin Lennon
@colinlennon
Mar 23 2016 17:00
Ah, I'll do that then. Seems like a cleaner solution. Thanks
Jim Matheson
@jmatheson
Mar 23 2016 18:25
@scottleibrand how do i switch from dev branch to advanced meal assist branch? searching through old thread here but can’t seem to find it
Scott Leibrand
@scottleibrand
Mar 23 2016 18:28
Do a git checkout advanced-meal-assist in ~/src/oref0/
you'll need to modify a bunch of reports too.
Jim Matheson
@jmatheson
Mar 23 2016 18:39
is this the report changes?
Scott Leibrand
@scottleibrand
Mar 23 2016 18:49
Those are for getting meal assist working on dev. I'll gist up the AMA stuff this afternoon.
benhamou
@benhamou
Mar 23 2016 18:51
may u state a code I enter to check, whether openaps is properly installed it didnt show the version at the end, while entering: openaps --version
it shows: openaps 0.0.10
pi@raspberrypi:~ $
Scott Leibrand
@scottleibrand
Mar 23 2016 18:54
That means the openaps toolset is installed. What exactly are you trying to check?
benhamou
@benhamou
Mar 23 2016 18:58
To check whether the whole package is properly installed
Scott Leibrand
@scottleibrand
Mar 23 2016 18:58
There are multiple packages.
benhamou
@benhamou
Mar 23 2016 18:58
as it didnt come up at the end of installation process, but only if entering version..
Scott Leibrand
@scottleibrand
Mar 23 2016 18:59
Wouldnt hurt to just rerun the quick- script.
this one ?
Scott Leibrand
@scottleibrand
Mar 23 2016 18:59
Ya, just do that again.
benhamou
@benhamou
Mar 23 2016 19:00
I saw danamlewis video on youtube :-)
interview
hmm nothing happens, when redoing it
just having the next command line no installation runs
Scott Leibrand
@scottleibrand
Mar 23 2016 19:11
you forgot the | bash part
otherwise i just waste your time, i dont know what bash is ... my last issues with such thinks where in dos
curl -s https://raw.githubusercontent.com/openaps/docs/master/scripts/quick-packages.sh | bash -
that tells the system "download the contents of this URL, and run it using the bash shell"
benhamou
@benhamou
Mar 23 2016 19:15
i hacked that until the bash, but i dont have entered bash and the sign bevore (I use a putty terminal from my windows pc)
-the sign before bash
Scott Leibrand
@scottleibrand
Mar 23 2016 19:15
you should be copying and pasting stuff like that
the | symbol is a "pipe". it's Shift-\
usually next to backspace on your keyboard
benhamou
@benhamou
Mar 23 2016 19:16
thank you copy and paste doesnt work unfortunately
thank you!!
Scott Leibrand
@scottleibrand
Mar 23 2016 19:16
why not?
you'll want to fix copy-paste before you continue
way too many things will get typo'd otherwise
you might need to right-click, middle-click, or Shift-Insert to paste, depending on your putty settings
that's what I used (with openaps --import) to get my rigs all set up for advanced-meal-assist.
benhamou
@benhamou
Mar 23 2016 19:20
haa
thank you :-)
right click is just perfect
and it runs again :-)
Scott Leibrand
@scottleibrand
Mar 23 2016 19:22
:)
LMK if you think that's good enough for what we wanted. Tested it overnight and just updated all my rigs with that.
Jim Matheson
@jmatheson
Mar 23 2016 19:25
nice. i like the #78
are these the only reports and aliases I need or do i also need to to do the pieces i pasted above if i did not turn on meal assist for this rig previously
Scott Leibrand
@scottleibrand
Mar 23 2016 19:29
you should probably get meal-assist set up first. not sure how much of it overlaps.
Jim Matheson
@jmatheson
Mar 23 2016 19:34
ok
Jim Matheson
@jmatheson
Mar 23 2016 19:49
been paying close attention to optimizing pre-bolusing the last week or so. I never realized how long it actually took for Novalog to start working. It’s actually an amazement that openaps can do so well with the 30 minute lag :)
Scott Leibrand
@scottleibrand
Mar 23 2016 19:49
yeah. :)
Dana Lewis
@danamlewis
Mar 23 2016 19:51
:)
Jim Matheson
@jmatheson
Mar 23 2016 19:59
so i pre-bloused 30m ago (still waiting,zzzz). is meal assist or anything taking into account COB?
Scott Leibrand
@scottleibrand
Mar 23 2016 20:00
meal assist on dev does not. advanced-meal-assist does.
meal assist does know about total carbs over the last DIA hours though
and compares that to total boluses and high-temps over the last DIA hours to decide whether to high-temp
Jim Matheson
@jmatheson
Mar 23 2016 20:01
so when i get that going what is the best way to enter things in. I normally just use the bolus wizard with no BG value and enter in my carb estimate. if i prebolus the carbs aren’t on board yet actually.
benhamou
@benhamou
Mar 23 2016 20:01
@scottleibrand may you help me getting the pi configured to nightscout sticked to this information:
[device "curl"]
fields =
cmd = bash
vendor = openaps.vendors.process
args = -c "curl -s https://yourwebsite.azurewebsites.net/api/v1/entries.json | json -e 'this.glucose = this.sgv'"
Jim Matheson
@jmatheson
Mar 23 2016 20:03
so for example in NS i see my COB dropping every update, but in reality i have not eaten yet. so the system would be misinformed and think those carbs would stop being on board sooner than they really will be.
Scott Leibrand
@scottleibrand
Mar 23 2016 20:03
are you entering carbs before you eat them?
Jim Matheson
@jmatheson
Mar 23 2016 20:04
yeah that is what i am saying. i use bolus wizard for my pre-bolus currently. so i enter carbs and let medtrionic do the math for me.
i dont use NS, etc
Scott Leibrand
@scottleibrand
Mar 23 2016 20:04
so the way AMA calculates COB is way more advanced than NS or anything else I've seen. it actually looks for positive BG deviations, and assumes those represent carb absorption.
so if you enter carbs and your BG behaves as if you haven't eaten them (for example, because you haven't) then it won't start decaying COB
Jim Matheson
@jmatheson
Mar 23 2016 20:05
oh, awesome.
benhamou
@benhamou
Mar 23 2016 20:06
how do I get AMA?
i am currently using happo
Scott Leibrand
@scottleibrand
Mar 23 2016 20:06
advanced-meal-asisst branch of oref0
Jim Matheson
@jmatheson
Mar 23 2016 20:06
i think it’s probably best to log the carbs when i start consuming them. but i tend to use the wizard since i dont have a simple 10:1 type of ratio
and i hate the manual interface for NS.
Scott Leibrand
@scottleibrand
Mar 23 2016 20:07
yeah, that'd be slightly better, but it won't matter much with AMA
benhamou
@benhamou
Mar 23 2016 20:07
it makes always sense to use a wizzard
Jim Matheson
@jmatheson
Mar 23 2016 20:07
yeah most value is down the line for analytics
Scott Leibrand
@scottleibrand
Mar 23 2016 20:08
another thing you could do is switch from calculated preboluses to smaller early "eating soon" prebolues
basically, correct to 80 instead of your normal target 30-60 minutes prior to the meal
you can do those as easy boluses, as the exact size doesn't matter as much as getting it in early enough
so if @danamlewis has an eventualBG of ~100, she'll do something like a 0.5U easy bolus 60 minutes prior, or 1U 30 minutes prior.
Jim Matheson
@jmatheson
Mar 23 2016 20:10
yeah, been thinking about that. 30 minutes is clearly not early enough for me. 41 minutes since i took ~6 units and my BG has dropped maybe 10 points
Scott Leibrand
@scottleibrand
Mar 23 2016 20:10
obviously if your ratio is more like 20 than 40, you can double those
and then she does her full bolus, minus however much she did for eating soon, when she actually eats
Jim Matheson
@jmatheson
Mar 23 2016 20:11
well my target BG in is 90, so in a perfect world i’d be sitting closer to 80 anyway. usually thats not the case of course.
but i am essentially in a constant correct to 90 anyway
Dana Lewis
@danamlewis
Mar 23 2016 20:12
Yep. Do a bonus without carbs, because you're not eating. Then when you eat, use wizard to bonus and enter carbs normally.
Scott Leibrand
@scottleibrand
Mar 23 2016 20:12
bonus bolus! ;-)
autocorrect is so much fun. :)
Jim Matheson
@jmatheson
Mar 23 2016 20:13
i’ve just been pre-bolusing for my whole meal. then wait until i am either in the double digits or headed down fast enough i think i am ready.
Dana Lewis
@danamlewis
Mar 23 2016 20:13
:p
This isn't pre bolus though
This is amount based on BG, not to do with amount of carbs you'll eat
Jim Matheson
@jmatheson
Mar 23 2016 20:14
yeah, i have seen other people use the term differently. but i do understand what you do. i just don’t think it will have enough of an effect for me.
i mean i am at 45 minutes past BG of 123 saying i ate 65 carbs and my BG is now just down to 109
and openaps had high temped me 20 minutes before that bolus
Dana Lewis
@danamlewis
Mar 23 2016 20:15
Insulin peaks at 60 so that makes sense
You won't always see a BG drop.
The big impact will be not having a humongous carb spike
Scott Leibrand
@scottleibrand
Mar 23 2016 20:22
we found that trying to do a big prebolus early enough to make BG start to drop before eating has too much risk of going low if you mistime it. but doing a smaller earlier pre-bolus whose peak activity coincides with the beginning of the meal, before your regular meal bolus kicks in, will blunt any meal rise without much risk of low
benhamou
@benhamou
Mar 23 2016 20:24
I give half an hour a square wave prebolus and when eating I bolus normal
but i have to find out the proper time before ... cant await having that loop closed :-)
Jim Matheson
@jmatheson
Mar 23 2016 20:25
yeah. so far i’ve only been doing it when i know for sure. like sitting at my house with cold pizza waiting in the fridge :)
dropping quickly now and ate about 5-10mins ago. will see how we go :)
Colin Lennon
@colinlennon
Mar 23 2016 21:31
Is it normal to see negative values for snoozeBG and eventualBG after meal boluses? I've just been reading up on how OpenAPS doesn't account for carbs on board so I'm guessing that's why I'm seeing that?
(I'm on master)
Scott Leibrand
@scottleibrand
Mar 23 2016 21:31
yeah, that's normal for eventualBG
Colin Lennon
@colinlennon
Mar 23 2016 21:31
Oh, okay, great
Scott Leibrand
@scottleibrand
Mar 23 2016 21:31
it's also normal for snoozeBG if the snooze wears off and you haven't risen much
Colin Lennon
@colinlennon
Mar 23 2016 21:32
Here's my most recent suggestion -
{"temp":"absolute","bg":150,"tick":-6,"eventualBG":-128,"snoozeBG":-79,"reason":"Eventual BG -128<118, no temp, setting -1.36U/hr","duration":30,"rate":0}
Scott Leibrand
@scottleibrand
Mar 23 2016 21:33
it won't low-temp as long as BG is rising, but if it starts to fall and it's predicting a low, it'll low-temp
Colin Lennon
@colinlennon
Mar 23 2016 21:33
I have about 4.1u of IOB
Scott Leibrand
@scottleibrand
Mar 23 2016 21:33
in the dev branch it's more sensitive to the rate of rise/fall, and how that compares to BGI, the rate you would be falling without any carb absorption
but I think in master it's just a binary cutoff
Colin Lennon
@colinlennon
Mar 23 2016 21:35
Okay yip, makes sense. I think I was just a little surprised because both times I've generated suggestions have been an hour or two after meals
I'll run my loop overnight tonight, logging out suggestions, and review tomorrow
Thanks :)
Scott Leibrand
@scottleibrand
Mar 23 2016 21:35
:-) :+1:
if you want to see it make more recommendations, you can set your target range narrower (min=max if you want)
Colin Lennon
@colinlennon
Mar 23 2016 21:37
I'll try that, thanks
Scott Leibrand
@scottleibrand
Mar 23 2016 21:37
we have had min=max in production for months, and it works quite well
initially wanted to minimize # of temps, but now we like it constantly adjusting. :)
Colin Lennon
@colinlennon
Mar 23 2016 21:38
Do you know if it's possible to have visualisation in NS before actually closing the loop? Would be nice to use NS to review rather than just the raw output
Dana Lewis
@danamlewis
Mar 23 2016 21:39
You could if you had a second test pump with no insulin in it..
Colin Lennon
@colinlennon
Mar 23 2016 21:39
Ah, I don't unfortunately. Not a big deal though, was just a thought
Dana Lewis
@danamlewis
Mar 23 2016 21:39
Not sure otherwise how to trick that, @jasoncalabrese may have some ideas
Mark Wilson
@mddub
Mar 23 2016 21:39
you could convert the suggested.json report into nightscout temp basal treatments and upload those
not sure that would be super helpful since it would always be based on pumphistory which doesn't include its suggestions
so it would very quickly come to not make sense as a simulation
Colin Lennon
@colinlennon
Mar 23 2016 21:42
Ah of course, I hadn't thought of that. Yeah, I'll skip that thought for now. I'll go ahead and switch to the dev branch of NS in preparation for closing the loop anyway. That'll keep me busy!
Mark Wilson
@mddub
Mar 23 2016 21:43
what i did was format suggested.json and show it on the urchin watchface so i could at least monitor its suggestions in real time
Dana Lewis
@danamlewis
Mar 23 2016 21:43
:+1: for the urchin watch face to see suggestions!
Mark Wilson
@mddub
Mar 23 2016 21:44
@scottleibrand recently merged a tool into dev to format it as text https://github.com/openaps/oref0/blob/dev/bin/oref0-html.js
Colin Lennon
@colinlennon
Mar 23 2016 21:44
I'm using urchin already! That sounds like a good option for me
Mark Wilson
@mddub
Mar 23 2016 21:44
you'd run that as part of your loop and then upload its output somewhere (your own server, a public git repo, etc.) and point urchin at that url
Colin Lennon
@colinlennon
Mar 23 2016 21:46
Excellent :+1:
Thanks
Mark Wilson
@mddub
Mar 23 2016 21:46
(btw, @scottleibrand, this reminds me: latest version of urchin has support for cards-for-pebble style json, with a "content" key and an optional "timestamp" key, with which minutes-ago can be shown instead of adding time in the text)
Scott Leibrand
@scottleibrand
Mar 23 2016 21:47
nice!
Mark Wilson
@mddub
Mar 23 2016 21:49
it's still slightly janky in that the recency is computed by the watch when the json is fetched, so if you lose bluetooth or network you'll see a stale minutes-ago (so you'd add the staleness reported next to the bluetooth/network icon to the staleness reported in the status to get the true staleness)
*by the phone, not the watch
need to move it to the watch
there's still like 7 hops too many in monitoring my loop
Scott Leibrand
@scottleibrand
Mar 23 2016 21:50
heh
Mark Wilson
@mddub
Mar 23 2016 21:50
bg: dexcom with share -> bt -> phone -> NS
pump history: pump -> pi -> wifi (possibly tethered) -> NS
viewing: watch -> phone -> NS
i just went camping for 3 days and had the pi paired directly to my pebble, it was just sending a short text notification every time the loop ran, but it was incredibly stable
think that's the direction things need to go
Scott Leibrand
@scottleibrand
Mar 23 2016 21:54
need both local and remote monitoring
Mark Wilson
@mddub
Mar 23 2016 21:54
maybe folks like you and dana do :P
remote data storage/analysis, for sure
remote monitoring has been nothing but a headache
Scott Leibrand
@scottleibrand
Mar 23 2016 21:56
single adults need local monitoring. significant others and parents need remote.
lots of feature requirements break down along that same dividing line
Mark Wilson
@mddub
Mar 23 2016 21:58
there's still a somewhat normative statement there, my long-term girlfriend is unlikely to ever have my bgs on her watch or phone
likewise i doubt everyone with share uses follow even though they can
i think everyone will do it differently
Scott Leibrand
@scottleibrand
Mar 23 2016 22:00
yeah, not all significant others will have that kind of relationship
Bill Stackpole
@billstackpole
Mar 23 2016 22:56
getting a preflight fail today - weird. Indicates something like "loose object <hash> (stored in .git/objects/e5/<hash> is empty." Thoughts?
think I need to reload the git repo or something but not sure...
Scott Leibrand
@scottleibrand
Mar 23 2016 23:07
yeah, you need to run oref0-reset-git
if you get the one from the dev branch it'll try oref0-fix-git-corruption first
otherwise it'll blow away git history and start from the current state
Bill Stackpole
@billstackpole
Mar 23 2016 23:08
just found that link from 9-March discussion. Do we have an idea of why this happens / is required?
Scott Leibrand
@scottleibrand
Mar 23 2016 23:08
yes, lots of fixes in newer dev versions of openaps
Bill Stackpole
@billstackpole
Mar 23 2016 23:09
how do I / should I - update to a newer dev version?
(literally just started with the cron job today and checked it when I got home - pi had borked via git)
Scott Leibrand
@scottleibrand
Mar 23 2016 23:09
I would just put oref0-reset-git in cron to start with
Bill Stackpole
@billstackpole
Mar 23 2016 23:10
how often? hourly? ev 5 min? 30 min? what is appropriate? (sry for so many q's)
Scott Leibrand
@scottleibrand
Mar 23 2016 23:10
* * * * * cd /home/pi/yourdir && oref0-reset-git
Bill Stackpole
@billstackpole
Mar 23 2016 23:10
ev min. K.
Scott Leibrand
@scottleibrand
Mar 23 2016 23:10
it doesn't do anything unless git status fails
Bill Stackpole
@billstackpole
Mar 23 2016 23:10
how did you get gitter to show the stars?
Scott Leibrand
@scottleibrand
Mar 23 2016 23:10
put ` around it
Bill Stackpole
@billstackpole
Mar 23 2016 23:10
ah - just single quotes - should have known.
doh
Scott Leibrand
@scottleibrand
Mar 23 2016 23:10
no
backticks
on the ~ key
Bill Stackpole
@billstackpole
Mar 23 2016 23:11
understood.
Bill Stackpole
@billstackpole
Mar 23 2016 23:20
one more weird thing "openaps raspi bad zero CRC?" in /var/log/messages (a few repeats here and there).
Scott Leibrand
@scottleibrand
Mar 23 2016 23:21
don't think that's a concern
Bill Stackpole
@billstackpole
Mar 23 2016 23:22
periodically when running my retry-loop I get "no CGM data" when I know (or at least think) that CGM data should be available. If I re-run, I usually get a response / suggestion that can be enacted (or not as the case may be) - ideas about this?
Finally, is there an easy way (good way?) to search the gitter repo? (I find lots of hits for things but they are either not the one that I know I just saw, or unrelated to my search / what I want to find). Would like to be more self sufficient (I know you put a lot of your time into this and don't need to address all the noob questions if I can find the answers in the existing traffic)
Scott Leibrand
@scottleibrand
Mar 23 2016 23:26
do you have something in your aliases that generates "no CGM data"?
unfortunately gitter isn't very searchable. we need people documenting things in the FAQ / troubleshooting sections as they figure them out
Bill Stackpole
@billstackpole
Mar 23 2016 23:27
the loop alias generates the "no CGM data" - wait one and I'll paste it in... (backtics?)
pi@openaps-raspi:~/billopenaps $ openaps alias show | grep CGM loop ! bash -c "openaps monitor-cgm 2>/dev/null && ( openaps preflight && openaps gather-profile && openaps monitor-pump && openaps monitor-cgm && openaps glucosecleaner && openaps checkiob && openaps suggest && openaps enact) || echo No CGM data."
Scott Leibrand
@scottleibrand
Mar 23 2016 23:28
I think that loop alias has faulty logic and needs fixing. lots of people complain about that
Bill Stackpole
@billstackpole
Mar 23 2016 23:29
so sounds like one of my aliases is failing, hence the "or" is triggered and I get "no cgm data"?
Scott Leibrand
@scottleibrand
Mar 23 2016 23:29
right. that's the wrong error message to print there
Tim Howard
@tghoward
Mar 23 2016 23:29
Agreed. The 'no CGM data' really just means that something failed along the way, not necessarily glucose
Bill Stackpole
@billstackpole
Mar 23 2016 23:30
K - so I'll change the error message - is there a way for me to indicate which of the aliases has failed?
rather than get the catchall at the end?
or should I just consider running a series of aliases consecutively to debug?
each with a "fail" message prescriptive to the scxript being run?
Scott Leibrand
@scottleibrand
Mar 23 2016 23:31
I do this: gather ! bash -c "rm monitor/*; ( openaps get-bg | egrep \"reporting|Copied\" && echo -n R && openaps report invoke monitor/status.json 2>/dev/null >/dev/null && echo -n e && test $(cat monitor/status.json | json bolusing) == false && echo -n fr && openaps ns-meal-carbs && echo -n esh && ( (openaps monitor-pump || openaps monitor-pump) >/dev/null && echo ed ) || (echo; sleep 60; exit 1)) 2>/dev/null"
so it prints Refreshed one bit at a time
or if each alias has its own output, you can look at how many of them you got
with the serialized && I don't think you can do an || for each one though
you'd have to switch to full nested if's, which would be ugly
probably best for now to just change the error to be more generic, and submit a PR to the docs to fix it there
Bill Stackpole
@billstackpole
Mar 23 2016 23:33
so the letters (R, e, fr, esh, ed) each indicate the part that succeeds or fails - if you get the whole word, all is well, else the missing part shows the issue - tricky! (Funny part was that I was typing the whole thing and until I saw them run on with commas between I didn't see it. <smile>)
Scott Leibrand
@scottleibrand
Mar 23 2016 23:34
;-)
Bill Stackpole
@billstackpole
Mar 23 2016 23:35
belated congrats to you and Dana btw - only recently stumbled on some wedding pics. :clap:
Dana Lewis
@danamlewis
Mar 23 2016 23:35
heh :) thanks!
Scott Leibrand
@scottleibrand
Mar 23 2016 23:35
:-D
Bill Stackpole
@billstackpole
Mar 23 2016 23:35
been married > 20 yrs - hope you get the same. :smiley_cat:
Dana Lewis
@danamlewis
Mar 23 2016 23:36
:D
Dana Lewis
@danamlewis
Mar 23 2016 23:45
once we confirm the stubbed out structure, per someone’s suggestion, I will make a list of all the pages we know need cleanup to be as nice as that one ^^ in terms of usefulness, so it’s easy for people to jump in and help improve
Ben West
@bewest
Mar 23 2016 23:47
the structure is pretty easy to change
that output is actually the output from --help
Tim Howard
@tghoward
Mar 23 2016 23:47
oooh. Like the new format for the docs!
Ben West
@bewest
Mar 23 2016 23:47
I wrote a quick script to automatically organize them in to manual pages
Tim Howard
@tghoward
Mar 23 2016 23:48
TOC on the left is nice.
Dana Lewis
@danamlewis
Mar 23 2016 23:51
if someone wanted to go restructure following this outline (openaps/docs@8ac49f7), that would be great :D
there’s stuff like this, writing up english for advanced and regular meal assist and whatnot, that are just waiting for volunteers :)
Ben West
@bewest
Mar 23 2016 23:54
little confused, do most of those items already map to a document?
Dana Lewis
@danamlewis
Mar 23 2016 23:54
some of them do. that’s the outline for the next gen docs, some of that content exists in current docs, some of it needs to be written.
Ben West
@bewest
Mar 23 2016 23:55
hmmm
Dana Lewis
@danamlewis
Mar 23 2016 23:55
so it’s an exercise in building the new pages, mapping the old content to the new, then starting to fill in the new gaps
Ben West
@bewest
Mar 23 2016 23:55
since we don't currently have a good UI for docs, I'd like to push the sphinx stuff through dev -> master, and then stub out this structure in dev
gitbooks is gone already, I'm assuming
Dana Lewis
@danamlewis
Mar 23 2016 23:56
someone else linked to it earlier today, not sure it’s dead. not sure if @channemann as the original publisher needs to delete something, or ?
Ben West
@bewest
Mar 23 2016 23:56
I'm the original publisher :-)
Dana Lewis
@danamlewis
Mar 23 2016 23:56
ahhh
Ben West
@bewest
Mar 23 2016 23:56
created all the resources :-D
Dana Lewis
@danamlewis
Mar 23 2016 23:56
then question for you, then, if you need to go kill/delete it?
but, i want to make sure sphinx won’t end up the same way in a few months...
Ben West
@bewest
Mar 23 2016 23:57
if it does we just switch again, although it's easier to run sphinx elsewhere than gitbook
rtd also years more experience in the wild than gitbook
that's what I meant, gitbook is business run by single person
Dana Lewis
@danamlewis
Mar 23 2016 23:58
I definitely would like one source of truth for the docs, so there’s not so much time spent upkeeping the formatting for two things
Ben West
@bewest
Mar 23 2016 23:58
ok, I'll replace gitbook with rtd
I feel comfy doing that
Dana Lewis
@danamlewis
Mar 23 2016 23:58
I’m just as likely to spend the time adding nav buttons to the bottom of every page in the github repo to help people navigate, rather than do something else
Ben West
@bewest
Mar 23 2016 23:58
I highly recommend avoiding that
Bill Stackpole
@billstackpole
Mar 23 2016 23:58
any idea what this means? Mar 23 19:58:11 openaps-raspi Stick transmit[TransmitPacket:PowerControl:data:unknown] reader[ReadRadio:size:14] download_i[4] status[<LinkStatus:0x03:status:size=??LinkStatus:error:True:reason:[]:size(0)>] poll_size[0] poll_i[False] command[<LinkStatus:0x03:status:size=??LinkStatus:error:True:reason:[]:size(0)>]:download(attempts[4],expect[0],results[0]:data[0]):BAD AILING
Dana Lewis
@danamlewis
Mar 23 2016 23:59
I would prefer an easier solution, but I also feel a bit burned by the plethora of complaints we were getting about some formatting not rendering in gitbooks for weeks on end
Ben West
@bewest
Mar 23 2016 23:59
it's fine, @billstackpole
Bill Stackpole
@billstackpole
Mar 23 2016 23:59
tnx @bewest
Dana Lewis
@danamlewis
Mar 23 2016 23:59
but then again, maybe it’ll be better/easier this time around since we have so many more loopers who can help respond and troubleshoot :)