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

12th
Jan 2016
Dana Lewis
@danamlewis
Jan 12 2016 00:01
aw :) :) :) :)
Dragan Maksimovic
@dm61
Jan 12 2016 00:12
@TC2013 :smile: :sparkles:
eyim
@eyim
Jan 12 2016 00:13
@TC2013 Wow Toby - amazing!
Jim Matheson
@jmatheson
Jan 12 2016 00:57
Is anyone besides me looking at pulling their BGs from Dexcom G4 directly to the Pi instead of figuring out some other way to get them to into NS and then pull them down to the Pi? I don’t have and am not planning on getting any sort of android device..
Rachel Sandlain
@audiefile
Jan 12 2016 01:01
You can plug the receiver into the pi directly
Scott Leibrand
@scottleibrand
Jan 12 2016 01:02
that is the normal default method
the question is what to do when you unplug the receiver from the Pi
Rachel Sandlain
@audiefile
Jan 12 2016 01:02
:+1:
Means everything works without internet
Jim Matheson
@jmatheson
Jan 12 2016 01:05
yeah, when i first reviewed docs it seemed like that was the default, but mostly seems like people are actually doing it via one of the other methods.
i was assuming i would not ever need to unplug it from the pi, but i guess you need to pull it while charing the receiver?
sadly i am still spinning wheels waiting for my dexcom to show up or else i’d have been looping a month ago. :(
Dana Lewis
@danamlewis
Jan 12 2016 01:11
Nope, the Pi will charge the receiver @jmatheson No need to unplug for that
Jim Matheson
@jmatheson
Jan 12 2016 01:11
oh nice. my old one died at some point so i was thinking it was not charging it.
Dana Lewis
@danamlewis
Jan 12 2016 01:12
It'll keep it topped off.
Matthias Granberry
@mgranberry
Jan 12 2016 01:12
@jmatheson sometimes it can stop charging if there is a ripple in the power supply
If you have issues you might try a different power supply for the pi.
Jim Matheson
@jmatheson
Jan 12 2016 01:13
ok, tx
Jason Calabrese
@jasoncalabrese
Jan 12 2016 01:15
just got a1c results 6.0, last time was 6.2
Dana Lewis
@danamlewis
Jan 12 2016 01:16
:) @jasoncalabrese
Jason Calabrese
@jasoncalabrese
Jan 12 2016 01:16
but even better is we're getting OpenAPS added to the Dr's orders for school
Dana Lewis
@danamlewis
Jan 12 2016 01:17
:) big first :) :)
Jason Calabrese
@jasoncalabrese
Jan 12 2016 01:19
sending it to school next Tuesday
Ben West
@bewest
Jan 12 2016 01:19
I'm currently pulling bg values from g4+share over ble
Matthias Granberry
@mgranberry
Jan 12 2016 01:19
Wow
Ben West
@bewest
Jan 12 2016 01:19
so now it's wireless :-D
Jason Calabrese
@jasoncalabrese
Jan 12 2016 01:19
hoping to have an edison based rig by then, but if not it's going anyway
Dana Lewis
@danamlewis
Jan 12 2016 01:20
Woot @bewest !
@jasoncalabrese Hope you get a rig's first day of school pic :)
Matthias Granberry
@mgranberry
Jan 12 2016 01:23
Something like that would have been great while I was in school back before pumps and bolus calculators were really a thing
Ali Mazaheri
@amazaheri
Jan 12 2016 01:32
that's awesome @jasoncalabrese
Jason Calabrese
@jasoncalabrese
Jan 12 2016 01:58
With the contour next link, they have a model for x23 and x22, anyone know what the difference is?
Pete Schwamb
@ps2
Jan 12 2016 02:05
Oh, that’s interesting.
I have no idea, but I can’t imagine what other kind of difference other than RF issues they would need diff hw for. Most non-rf things could be accounted for in SW, I’d think.
Scott Leibrand
@scottleibrand
Jan 12 2016 02:31
maybe the remote for the x22 used some of the commands that got disabled in the x23?
Dana Lewis
@danamlewis
Jan 12 2016 02:32
I don't think it temp basaled
Scott Leibrand
@scottleibrand
Jan 12 2016 02:35
the x22 remotes may have used the now-disabled bolus commands, and they may have had to replace those with something else in the x23? all speculation on my side...
Ben West
@bewest
Jan 12 2016 02:35
that's all software
on the app side ongsf thi
app side of things
Scott Leibrand
@scottleibrand
Jan 12 2016 02:36
yeah. do we know for a fact that the x23 nextlink uses different hardware than the x22 one, vs. just different firmware?
Pete Schwamb
@ps2
Jan 12 2016 02:37
No, I was just thinking it’s odd to have different model names for the same piece of hw, but you’re right, it could represent different fw.
Jason Calabrese
@jasoncalabrese
Jan 12 2016 02:37
Maybe the x23 one can do something with mysentry?
Ben West
@bewest
Jan 12 2016 02:38
that might be it
dunno, might call them up and ask them
where is link? I can follow up as well
Ben West
@bewest
Jan 12 2016 02:43
weird
yeah, we should generate bunch of tickets for that
Pete Schwamb
@ps2
Jan 12 2016 02:44
What’s the OS that people have been putting on their edison for use with openaps? Any writeup’s on the process?
Ben West
@bewest
Jan 12 2016 02:44
it comes with linux installed
Jason Calabrese
@jasoncalabrese
Jan 12 2016 02:44
Yeah, there's a gist
Pete Schwamb
@ps2
Jan 12 2016 02:44
Yeah, but they’ve been re-installing something else.
Jason Calabrese
@jasoncalabrese
Jan 12 2016 02:44
Looking it up
Ben West
@bewest
Jan 12 2016 02:44
it's a variant of yocto - uses bitbake to build
oh
I think intel has only released limited suite of kernels on bitbake
Pete Schwamb
@ps2
Jan 12 2016 02:45
ubilinux?
Ben West
@bewest
Jan 12 2016 02:45
openembed, yoctolinux
Ubi
Ben West
@bewest
Jan 12 2016 02:46
ah
Pete Schwamb
@ps2
Jan 12 2016 02:46
Thanks!
Ben West
@bewest
Jan 12 2016 02:46
did not realize that, nice
Jason Calabrese
@jasoncalabrese
Jan 12 2016 02:46
Gets updated to Jessie
So just like the pi
Ben West
@bewest
Jan 12 2016 02:46
mmmm
I don't care for systemd though :(
Jason Calabrese
@jasoncalabrese
Jan 12 2016 02:48
Only issues I had with the install was, didn't work from Mac, then did it from pi, but had to add 1G of swap
Ben West
@bewest
Jan 12 2016 02:48
hmm fairly sizeable chunk
Pete Schwamb
@ps2
Jan 12 2016 02:48
The one that was installed (yocto) seemed blazing fast back when I played with it (maybe just in contrast to RPI). But I guess yocto has no package manager and makes you recompile everything?
Ben West
@bewest
Jan 12 2016 02:48
it has opkg
you can publish your own lists/builds fairly easily
just need the build directory hosted by http
Jason Calabrese
@jasoncalabrese
Jan 12 2016 02:49
Nice to have Ubuntu, then the normal OpenAPS docs work
Pete Schwamb
@ps2
Jan 12 2016 02:49
How hard would it be to build & host an openaps build for edison on yocto?
Ben West
@bewest
Jan 12 2016 02:50
we can include things like https://github.com/codazoda/hub-ctrl.c slightly more easily
also might pave the way to doing full embedded device
at some point we'll want to start powering down linux most of time?
rely purely on embedded coms?
becomes ble appliance at some point?
all that is slightly easier, probably with yocto/OpenEmbedded builds
some maintenance hassle in setting up/hosting the builds
Pete Schwamb
@ps2
Jan 12 2016 02:51
yocto isn’t a RTOS, right?
Ben West
@bewest
Jan 12 2016 02:51
no, it's not, it's basic linux
can't use it to truly control tasking/scheduling probably
if that's needed...
Pete Schwamb
@ps2
Jan 12 2016 02:53
I was only interested in that from a bit-banging perspective. Not sure, but I think the cc-debug protocol will be doable w/o rtos, since there is a clock line that the host controls.
How would using hub-ctrl.c be easier on yocto than ubilinux?
Ben West
@bewest
Jan 12 2016 02:54
meh, in terms of distributing a package
posting to ppa on launchpad is fairly high-hassle experience
this is simpler, it's more like python's pypi, it just requires a link on a page somewhere
technically, maybe publishing deb list is just as simple
I've only done it in proprietary environments and on launchpad
to add hub-ctrl, it's just an o-e/bitbake build, and then upload/rsync/ftp
Pete Schwamb
@ps2
Jan 12 2016 02:57
I’m wondering what limits @oskarpearson hit with yocto that made him want to switch to ubilinux? yocto has python, I believe, and you could do easy_install stuff pretty easily, I’d think.
Ben West
@bewest
Jan 12 2016 02:57
ah, it was the carelink serial
has to do with kernel, intel made something like 3.12 available
really need 3.18 or above for reliable legacy usb handling
Pete Schwamb
@ps2
Jan 12 2016 02:57
Hmm, so if going UART that might not be an issue.
Jason Calabrese
@jasoncalabrese
Jan 12 2016 02:58
Yocto sounded like a pain from everything I've seen
Ben West
@bewest
Jan 12 2016 02:58
it's just another build system
openwrt is also a pain
it's a funded marketplace for linux, so that seems cool
Jason Calabrese
@jasoncalabrese
Jan 12 2016 02:59
Ubi feels comfortable, maybe yocto will be an optimization
Ben West
@bewest
Jan 12 2016 02:59
it's all "free", but it's also paid for, properly
Pete Schwamb
@ps2
Jan 12 2016 02:59
You guys were running out of space on ubi?
Ben West
@bewest
Jan 12 2016 03:02
the "base" for yocto is suitable for embedded devices...
basically just openbox, very simple linux
not even bash, it's the simplest possible linux, basically
then everyone adds patches for their stuff
so the result is a custom distro, every time you build it
opkg is similar to apt-get
there are often some basic package repos setup/maintained
also the vendors often create their own...
so intel made one for edison
but they didn't bother to maintain it since release
Jason Calabrese
@jasoncalabrese
Jan 12 2016 03:05
The default partitions aren't great
Ben West
@bewest
Jan 12 2016 03:06
the resin.io install script reformats partitions, I think
Jason Calabrese
@jasoncalabrese
Jan 12 2016 03:06
Split between / /home
Ben West
@bewest
Jan 12 2016 03:06
ah, I always split /home off from everything now
ensures you can bring up completely new system and re-attach your home
Jason Calabrese
@jasoncalabrese
Jan 12 2016 03:07
Its just bigger than needed, and root is a little small
Ben West
@bewest
Jan 12 2016 03:07
ah, I see
just a tad aggressive, sounds like
I agree with most of that post
Jason Calabrese
@jasoncalabrese
Jan 12 2016 03:07
Yeah, don't know if it's really a problem
Ben West
@bewest
Jan 12 2016 03:07
ubilinux is a distro, so you get all benefits of a familiar distro
packages are names you can install, you can copy/paste basically any debian base instructions
on other hand, we're starting to customize more and more
already have several custom hardwares...
bitbake/openembed/yocto is great for maintaining custom builds like that
especially if we get a pizero
a custom pizero or similar, we'll need bitbake to organize/build
Ben West
@bewest
Jan 12 2016 03:13
@ps2 any chance you have a rough idea of the changes needed to subg_rfspy to make it work on the t1 cc1111 usb dongle aka "don's dongle?"
I see rfcat's usb stuff, thinking would need layer that does that similar to the "submodules" already there?
Ben West
@bewest
Jan 12 2016 03:17
yuppers
I have 2 of these, sdcc, and a goodfet
Jason Calabrese
@jasoncalabrese
Jan 12 2016 03:18
Just got that today
Ben West
@bewest
Jan 12 2016 03:18
and I like your C better than atlas' ;-)
Jason Calabrese
@jasoncalabrese
Jan 12 2016 03:18
Going to try flashing it tonight
Pete Schwamb
@ps2
Jan 12 2016 03:18
Let me take a look at the datasheet. I don’t really know how the USB stuff works.
or, conversely, what should I treat my RL?
well, some of the comments in there may explain why I could never transfer bytes
Pete Schwamb
@ps2
Jan 12 2016 03:23
Yeah, that c file you linked to looks like it has most of the stuff you’d need to implement a serial.c file for subg_rfspy; there’d be a bit of work to integrate that code with what subg_rfspy expects.
Ben West
@bewest
Jan 12 2016 03:23
fair, that's to be expected
Pete Schwamb
@ps2
Jan 12 2016 03:23
What should you treat your RL? I don’t understand.
Ben West
@bewest
Jan 12 2016 03:24
I have two RL and two of those
none of them work great, so far
I noticed you had bunch of updates to RL firmware
also subg_rfspy
but I only have sdcc, also
I just got a windows laptop, I could retry bluegiga et al
Pete Schwamb
@ps2
Jan 12 2016 03:25
Yes, I’m running my RL with subg_rfspy and a new ble113 firmware that talks to it. It dumps history nicely now, though I’m still working through some app issues.
Ben West
@bewest
Jan 12 2016 03:25
maybe I'll give a shot creating a serial.c for the usb
interesting
so bluegiga might be worth a shot as well
extending range and making everything wireless dampens some pressure on mqtt/deamonizing everything
Pete Schwamb
@ps2
Jan 12 2016 03:27
I was going to look at updating mmblelink with subg_rfspy style commands and the new ble attributes, since it simplifies some things. Though mmeowlink has the commands for subg_rfspy already in it, so it might be easier to add ble support to mmeowlink.
scottleibrand @scottleibrand nods
Scott Leibrand
@scottleibrand
Jan 12 2016 03:27
I'd rather focus on fixing range than adding more complexity to compensate for it
Pete Schwamb
@ps2
Jan 12 2016 03:27
Too many things to do!!!!
Ali Mazaheri
@amazaheri
Jan 12 2016 03:28
range plsssss
Scott Leibrand
@scottleibrand
Jan 12 2016 03:28
really looking forward to start testing out the longer-range radios
Pete Schwamb
@ps2
Jan 12 2016 03:28
And I’m supposed to be catching up on my homework for the stanford machine learning course I’m taking. :) Weird having homework again.
Ben West
@bewest
Jan 12 2016 03:28
mmeowlink has made some good progress, I especially appreciate the ability to dynamically switch the "transport" backend
if that continues, it might find itself merged into mmblelink
or it might supersede mmblelink
Pete Schwamb
@ps2
Jan 12 2016 03:29
So I’m conflicted about claiming longer range. I still can’t talk to the pump sometimes when I’m up < 2 meters away. Most of the time it works, and it works at much greater distances as well, but It’s hard to say yes, this is better.
Ben West
@bewest
Jan 12 2016 03:29
been thinking about splitting up decocare to be easier to swap out the backend
Jason Calabrese
@jasoncalabrese
Jan 12 2016 03:30
Smarter retries might help a lot
Scott Leibrand
@scottleibrand
Jan 12 2016 03:30
@ps2 what do you think the biggest remaining impediment is? detuning?
Ali Mazaheri
@amazaheri
Jan 12 2016 03:30
with TI its intermittent
Ben West
@bewest
Jan 12 2016 03:30
too much to do, indeed
Scott Leibrand
@scottleibrand
Jan 12 2016 03:30
and :+1: to smarter retries
Ben West
@bewest
Jan 12 2016 03:31
I'd be interested in creating smart modem
try frequency hopping a bit, to tune to the pump
Pete Schwamb
@ps2
Jan 12 2016 03:32
The subg_rfspy is very well suited to experimenting with that kind of stuff. That’s actually the main reason I built it. I only decided to switch RL stuff over to it after I wrote it and decided I really preferred the command structure and code clarity.
Ben West
@bewest
Jan 12 2016 03:32
the history decoding and model logic code is all independent from the transport used to get the payloads
Pete Schwamb
@ps2
Jan 12 2016 03:32
(i.e. I didn’t want to go look at my old code again. ;)
Ben West
@bewest
Jan 12 2016 03:32
hehe, it takes a few tries to get it
look at insulaudit and precursors ;-)
yeah, subg_rfspy is looking really promising
if someone were to try to extend to cover usb on "don's dongle"
.usb_alt0? or what name should be used?
Pete Schwamb
@ps2
Jan 12 2016 03:34
So you should be able to use the cc1111 dongle with subg_rfspy if you hook up to it via UART pretty easily.
Ben West
@bewest
Jan 12 2016 03:34
maybe
oh using jumpers perhaps
Pete Schwamb
@ps2
Jan 12 2016 03:35
yeah
Ben West
@bewest
Jan 12 2016 03:35
from test to pi?
hmmmm
well now e-ink and this both want the pins
this is why I have 4 pi though
I'll try that in a bit probably, that sounds fastest
Pete Schwamb
@ps2
Jan 12 2016 03:36
Well, I think the debug group of pins has the uart1_alt2 pins
Ben West
@bewest
Jan 12 2016 03:36
ah, I see
Pete Schwamb
@ps2
Jan 12 2016 03:36
That’s how I do UART to RL.
Ben West
@bewest
Jan 12 2016 03:36
you get those names from the spec sheet?
ah, nice
so RL -> pi same way
Pete Schwamb
@ps2
Jan 12 2016 03:37
Yes
Ben West
@bewest
Jan 12 2016 03:38
on one of my pis, I've been using some of these pins to drive an e-ink display
one could use a breadboard with wiring to use both, or I'm just out of pins?
probably don't need to use both at same time
Pete Schwamb
@ps2
Jan 12 2016 03:41
Just verified in the cc1111 dongle user’s guide that it has the same pins exposed as RL. So you would use the uart1_alt2 setup
Ben West
@bewest
Jan 12 2016 03:41
ah great
ok, you just compare the pins on the pdf spec for both sheets?
now I feel bad that I could have done that
Pete Schwamb
@ps2
Jan 12 2016 03:42
Well, I don’t have a sheet for the RL. :)
And then look at the “UART on RileyLink” here https://github.com/ps2/subg_rfspy
P1_(4-7) are what you would connect to the pi.
Although I just realized that that’s with HW flow control, which I guess is hard to expose on the rpi?
Ben West
@bewest
Jan 12 2016 03:45
the usb code in rfcat bothers me
Pete Schwamb
@ps2
Jan 12 2016 03:45
I’ve been using FTDI boards to hook up to the RL when doing UART over those pins.
Ben West
@bewest
Jan 12 2016 03:46
looks like you rely on basically 4 functions
Pete Schwamb
@ps2
Jan 12 2016 03:46
Yes, the serial.h interface is what you need to implement.
Ben West
@bewest
Jan 12 2016 03:46
why diff in between rx and tx in word vs byte?
sorry
word vs str
they both have _byte
Pete Schwamb
@ps2
Jan 12 2016 03:47
Nothing in the protocol sends a word.
Ben West
@bewest
Jan 12 2016 03:48
ok, nice
Matthias Granberry
@mgranberry
Jan 12 2016 03:59
@scottleibrand I'm really impressed at how openaps catches lows well before they happen. My past 3 days of "open looping" with data fed in from my t:slim have been pretty incredible.
If I'd known it would matter this much I'd have bought a Medtronic pump months ago
Dana Lewis
@danamlewis
Jan 12 2016 04:05
Glad to hear @mgranberry!
Open loop or closed loop, glad this stuff is working for so many people :)
This open loop style is basically what I had with DIYPS, and wanted to make more available for other people , just didn't call it an open loop because we never imagined we'd close the loop :)
Scott Leibrand
@scottleibrand
Jan 12 2016 04:07
interestingly, I remember when I first started suggesting we use temps to zero to preempt the need for carb corrections, @danamlewis was also skeptical.
Dana Lewis
@danamlewis
Jan 12 2016 04:07
Wondering how many other people would open loop if they can't/don't want to get a Medtronic pump, and if it would be worthwhile to call out in the docs as making a difference
Still a lot of work to get the open loop, but for some it might be worth it?
Scott Leibrand
@scottleibrand
Jan 12 2016 04:08
wonder if we should make HAPP an official part of the openaps repo
Dana Lewis
@danamlewis
Jan 12 2016 04:08
Skeptical as in you need the predictive features. CGM alone alarms are too late.
@mgranberry Thoughts on making open loop more well articulated in the docs?
Matthias Granberry
@mgranberry
Jan 12 2016 04:14
I think an app is the way to go for open loop. HAPP needs some love to read in treatments etc from an external source and some way of keeping the interfaces current will be necessary. I'd like to clean up what I've done and make a PR, but it's the kind of hack I'd push back on if someone tried committing it into my code base
I really don't want to carry an extra hunk of computer with only 12h of battery life.
But all the predictive alarm stuff could be a part of Nightscout and use all its machinery too. With pushover notifications it could work fairly well. The nighttime exposure is the big problem with that approach. 30-minute basals sound like they would be a bigger liability than a boon there.
Dana Lewis
@danamlewis
Jan 12 2016 04:20
Not necessarily. They're calculated to be the absolute safest thing to do. We started with temp to zero for 30, no high temps (just bolus alerts for >.5 needed).
Only recommending a conservative temp that is safe to run for 30 (high or low) is better than no alert to take action, iMO
I agree it could be part of NS
Scott Leibrand
@scottleibrand
Jan 12 2016 04:21
overnight drifting low, a temp to zero for 30m (or very rarely 60m) is often all you need to go back to sleep and end up in range
going into bedtime headed low with IOB left over from dinner is a different story of course
Matthias Granberry
@mgranberry
Jan 12 2016 04:22
I'm not worried about safety. They do pretty well for me. I'm very impressed so far. It's loss of sleep I'm worried about. Homeostasis does a lot for me at about 85 mg/dL, but I usually aim for a little above that and have my lower limit set at 90 right now to lend some safety.
I can tell that openaps does a really good job even when correcting for some sugar surfing corrective boluses. It takes action right about where I would to prevent lows, only it it always watching. Highs it's a little less aggressive about, but that's reasonable and it has helped keep my variability down in a "good day" range even when having tacos and chips from an unknown restaurant and pizza.
Toby Canning
@TC2013
Jan 12 2016 04:28
You just named 75% of my diet
;)
JaysonEwer
@JaysonEwer
Jan 12 2016 04:31
Good evening, all.
Matthias Granberry
@mgranberry
Jan 12 2016 04:31
It probably helps that they're both slow carbs, but I can never predict when it will actually end. It's nice just typing in numbers without having to do mental math
JaysonEwer
@JaysonEwer
Jan 12 2016 04:32
I'm working on getting oref0 installed tonight and ran into a few interesting messages on install. It appears the install successfully completed, but was wondering if I need to clear out the npm warn messages?
Toby Canning
@TC2013
Jan 12 2016 04:33
I looks like we will need a 2" x 2.5" x 1" box for the Edison setup.
I've never done anything about those...
The npm warnings that is
And good evening @JaysonEwer
JaysonEwer
@JaysonEwer
Jan 12 2016 04:35
Thanks for the feedback @TC2013. I'll proceed. :-)
Toby Canning
@TC2013
Jan 12 2016 04:36
You bet
JaysonEwer
@JaysonEwer
Jan 12 2016 04:50
Silly Question, I'm sure, but where do find my hashed version of my API_Secret?
Scott Leibrand
@scottleibrand
Jan 12 2016 04:51
not silly at all. we should probably document that. :-)
$ echo asdf | sha1sum
7d97e98f8af710c7e7fe703abc8f639e0ee507c4 -
make note of where in the docs that needs to be added, and when you're ready make the edit to your fork of the docs so you can submit a pull request to openaps/docs and we can get them updated
JaysonEwer
@JaysonEwer
Jan 12 2016 05:01
Because when I click on my API_SECRET field in my azure website for NS, It goes back to clear text, doesn't it? I know I JUST tonight got my IOB showing up in my nightscout, and selected save hash here because its a personal device when I Authenticated my device. :-). I will be honest, I haven't try to read every document to see if there are other hints elsewhere, but I'm working through the oref0 stuff at http://github.com/openaps/oref0 under the ns-upload-entries section.
@scottleibrand , As far as I know I haven't associate my pi with my Nightscout environment yet. Will running that command on my pi provide me with the API_SECRET hash?
Does that make sense?
@scottleibrand, Thank's for the info, btw.
Thanks for all of this! You guys are amazing! I picked up my copy of Popular Science today. :-). Just so I can hang it up for my grandchildren to see. ;-)
JaysonEwer
@JaysonEwer
Jan 12 2016 05:06
I've got 3 kids, 7,5, 18...Assuming I can start looping, perhaps i'll be around long enough to see the grandchildren. :-)
18months, that is
Scott Leibrand
@scottleibrand
Jan 12 2016 05:08
@jaysonewer so if I wasn't clear, doing echo mypasswordcleartext | sha1sum will return the hashed version
JaysonEwer
@JaysonEwer
Jan 12 2016 05:09
Do I include the hyphen at the end of the hash in the string to enable the ns-upload-entries for the oref0?
oh, my bad
perfect
thank you
Dana Lewis
@danamlewis
Jan 12 2016 05:10
:) re: hanging on wall @JaysonEwer
Scott Leibrand
@scottleibrand
Jan 12 2016 05:10
I don't think the hyphen is useful. man sha1sum will probably explain
JaysonEwer
@JaysonEwer
Jan 12 2016 05:12
k, that's what I thought, i'll do my best to determine and then update the docs with as much detail as I can.
Scott Leibrand
@scottleibrand
Jan 12 2016 05:12
thx
Dana Lewis
@danamlewis
Jan 12 2016 05:18
Ps - roll tide! :)
Toby Canning
@TC2013
Jan 12 2016 05:29
What a game!
Dana Lewis
@danamlewis
Jan 12 2016 05:30
:+1:
Toby Canning
@TC2013
Jan 12 2016 05:32
I often root for teams because of their coach and Nick Saben is a top notch fella!
Unlike the Clemson coach who has a chip on his shoulder that's massive
Kevin Wells
@wkcwells
Jan 12 2016 07:05
I’m running open loop and about half the time when I manually enact the suggested temp basal I get a python exception:
  File "/usr/local/lib/python2.7/dist-packages/decocare-0.0.21-py2.7.egg/decocare/commands.py", line 277, in Program
    assert duration % 30 is 0, "duration {0} is not a whole multiple of 30".format(duration)
TypeError: unsupported operand type(s) for %: 'NoneType' and 'int'
Has anyone else run into this?
Scott Leibrand
@scottleibrand
Jan 12 2016 07:07
Can you paste the suggested.json?
Kevin Wells
@wkcwells
Jan 12 2016 07:08
{"temp":"absolute","bg":260,"tick":-3,"eventualBG":144,"snoozeBG":253,"reason":"Eventual BG 144>110, temp 2.25 >~ req 1.96U/hr”}
The "duration": 0,s appear to be in pumphistory.json.
Maybe not. They were in my boluses. I edited them to “1” and still have the same error.
Scott Leibrand
@scottleibrand
Jan 12 2016 07:15
You're trying to enact a no-action suggested.json
notice the complete absence of rate or duration.
You need to check for the presence of those using the json tool or grep and only enact if they're present.
Kevin Wells
@wkcwells
Jan 12 2016 07:17
Ah. Thanks. The upstream problem is that I am trying to upload OpenAPS status to Nightscout, and without an enacted.json, I get a ValueError: No JSON object could be decoded error.
Scott Leibrand
@scottleibrand
Jan 12 2016 07:17
A lot of people get stuck on that. We should make the docs more clear, or make the error message more explicit. If you have any suggestions for where to clarify that in the docs, please feel free to fork and edit them, and submit a pull request to update.
Kevin Wells
@wkcwells
Jan 12 2016 07:18
With an enacted.json it seems to work fine. I’m still very new to all of this, so my configuration may not be right. (The open loop seems to run fine.)
Scott Leibrand
@scottleibrand
Jan 12 2016 07:18
Do you have one from when that 2.25U/hr temp was set?
Ben West
@bewest
Jan 12 2016 07:18
that error is expected
Kevin Wells
@wkcwells
Jan 12 2016 07:19
Not any more. I’ll keep a sample around from now on.
Scott Leibrand
@scottleibrand
Jan 12 2016 07:19
I never delete my old enacted.json
so the upload always gets whatever was last enacted, with its old timestamp.
nightscout is quite happy with that.
Kevin Wells
@wkcwells
Jan 12 2016 07:21
I think I know what I am doing wrong. I am creating enacted.json by >> the output of set_temp_basal, which is zeroing out the file if there is not basal to enact.
I think I need to set up a proper report for enacting. Does that sound right
?
Ben West
@bewest
Jan 12 2016 07:22
this is a common error, I just added a patch to eliminate the error message
Scott Leibrand
@scottleibrand
Jan 12 2016 07:22
Yes, enact should be a report, but not sure it's what's causing this.
You also need to avoid invoking the enact report unless there is a duration in the suggested.json, if you don't want the uploader to complain about an empty enacted.json
maybe we should make the uploader stop complaining about that, though.
Ben West
@bewest
Jan 12 2016 07:23
this patch should help with that
Scott Leibrand
@scottleibrand
Jan 12 2016 07:24
your patch will help with the enact. Doesn't change upload though, right?
Can you link your commit?
Ben West
@bewest
Jan 12 2016 07:29
This message was deleted
oops
there we go
$ openaps use pump set_temp_basal  oref0-predict/oref0.json
{ 
  "bg": 121,
  "temp": "absolute",
  "snoozeBG": 113,
  "timestamp": "2016-01-11T23:27:37.087634",
  "missing": [
    "duration",
    "rate"
  ],
  "reason": "Eventual BG 113>106 but Avg. Delta -2.67 < Exp. Delta 0.1; no temp to cancel",
  "eventualBG": 113,
  "error": "missing required input fields",
  "input": {
    "reason": "Eventual BG 113>106 but Avg. Delta -2.67 < Exp. Delta 0.1; no temp to cancel",
    "bg": 121,
    "eventualBG": 113,
    "temp": "absolute",
    "snoozeBG": 113,
    "tick": -2
  },
  "tick": -2
}bewest@bewest-MacBookPro:~/Documents/openaps$
Scott Leibrand
@scottleibrand
Jan 12 2016 07:33
Ah, cool. So that writes an enacted.json regardless. Just looks way different if it's missing inputs.
Will need to test that with the Nightscout uploader and display stuff, but I like that approach.
@jasoncalabrese ^^
Kevin Wells
@wkcwells
Jan 12 2016 07:34
OK. I created a proper report (I think). But since I have no temp basal to enact, I created an empty JSON dictionary for enacted.json. That seems to workaround this to get the uploader going.
Scott Leibrand
@scottleibrand
Jan 12 2016 07:34
cool, good th
Good thinking.
Kevin Wells
@wkcwells
Jan 12 2016 07:35
Thanks @scottleibrand & @bewest very much! I’ve got an early morning, so I am going to step out. I’ll install the patch and work on getting the UI showing in NS tomorrow. Thanks again!!
Ben West
@bewest
Jan 12 2016 07:36
fwiw, I only upload events from pump-history to ns
Scott Leibrand
@scottleibrand
Jan 12 2016 07:38
Yeah, you can now also upload enacted.json's and the openaps pill will display them.
Good for going back and seeing why it did something in retro mode.
Sulka Haro
@sulkaharo
Jan 12 2016 08:13
Hmm, I've now had two nights where kid goes low and stays below the target, so OpenAPS zero-temped for 2.5 hours. Results in ketone buildup due to lack of insulin, measured ketones at 2.5 last morning. I think I need a warning for when it looks like the system will zero-temp for too long.
Thomas Emge
@ThomasEmge
Jan 12 2016 08:21
my cron job isn't quite working as expected. I know that when I call it manually it work ok but not in the scheduler
I can see in Jan 12 00:10:01 socalAPS CRON[8233]: pam_unix(cron:session): session opened for user pi by (uid=0) Jan 12 00:10:01 socalAPS CRON[8240]: (pi) CMD (/home/pi/red_1/loop_aps.sh >/home/pi/red_1/logs/loop_$(date +) Jan 12 00:10:01 socalAPS CRON[8233]: (CRON) info (No MTA installed, discarding output) Jan 12 00:10:01 socalAPS CRON[8233]: pam_unix(cron:session): session closed for user pi
let me try that again
Jan 12 00:10:01 socalAPS CRON[8233]: pam_unix(cron:session): session opened for user pi by (uid=0) Jan 12 00:10:01 socalAPS CRON[8240]: (pi) CMD (/home/pi/red_1/loop_aps.sh >/home/pi/red_1/logs/loop_$(date +) Jan 12 00:10:01 socalAPS CRON[8233]: (CRON) info (No MTA installed, discarding output) Jan 12 00:10:01 socalAPS CRON[8233]: pam_unix(cron:session): session closed for user pi
Thomas Emge
@ThomasEmge
Jan 12 2016 08:27
Jan 12 00:10:01 socalAPS CRON[8233]: pam_unix(cron:session): session opened for user pi by (uid=0)
Jan 12 00:10:01 socalAPS CRON[8240]: (pi) CMD (/home/pi/red_1/loop_aps.sh >/home/pi/red_1/logs/loop_$(date +)
Jan 12 00:10:01 socalAPS CRON[8233]: (CRON) info (No MTA installed, discarding output)
Jan 12 00:10:01 socalAPS CRON[8233]: pam_unix(cron:session): session closed for user pi
Oskar Pearson
@oskarpearson
Jan 12 2016 09:17
@ps2 I changed away from yocto because it doesn't have a large number of useful packages, and doesn't have things like a continuous security upgrade process. You need to rebuild the whole image. Useful things like strace, tcpdump, and more were missing. The packages in opkg aren't really much use at all - though the unofficial repos by "some random dude on the internet" seemed to have more things.
Additionally, I wanted to be able to point people at the same scripts that we use for the Raspberry Pi
The current install gist is a it of a hack though
Something like resin.io is closer to the ideal - but it does mean moving away from the current install process - and to something a lot more pre-packaged.
I'm certainly happy for someone to start maintaining a better set of opkg packages, or building something using yocto that we can all use... but I don't really want to do that personally
Scott Leibrand
@scottleibrand
Jan 12 2016 10:40
@ThomasEmge so does that log file get created? What if anything gets written to it?
And have you tried running the exact same command (including the log file output redirection) from the CLI?
you might also try 2> to redirect stderr to that log file too if it's not clear from the log or manual run what's failing.
Dana Lewis
@danamlewis
Jan 12 2016 10:41
@sulkaharo Hm, good idea. Might also help to alarm for negative IoB before normal am wake up time to help with breakfast? Some possibilities along the same lines there.
Matthias Granberry
@mgranberry
Jan 12 2016 13:19
It's not safe to zero temp > 2h. Ketones are only part of it. Insulin will also crystlaize in tubing and begin to clog and cause occlusions
Jason Calabrese
@jasoncalabrese
Jan 12 2016 13:24
I was thinking of adding a -IOB alarm, but might be better to do it based on time
Milos Kozak
@MilosKozak
Jan 12 2016 13:40
let's move the calculation to NS too and do it using NS alarming
Matthias Granberry
@mgranberry
Jan 12 2016 14:56
@MilosKozak I think that's a good open-loop solution
Thomas Emge
@ThomasEmge
Jan 12 2016 15:01
@scottleibrand correct the same line executed in the CLI works as expcted but not as part of crontabs ```
/10 * /home/pi/red1/loopaps.sh >/home/pi/red_1/logs/loop$(date +%Y%m%d%H%M%S).log 2>&1
@scottleibrand my markdown skills, oh boy,....
*/10 * * * * /home/pi/red_1/loop_aps.sh >/home/pi/red_1/logs/loop_$(date +%Y%m%d_%H%M%S).log 2>&1
Jason Calabrese
@jasoncalabrese
Jan 12 2016 16:05
was wondering if moving a bunch of stuff from oref0 to dcalc makes sense or if we should add oref0 as a dependency for NS
if we move determine basal, iob, etc to dcalc, what would stay in oref0? just the /bin scripts?
Sulka Haro
@sulkaharo
Jan 12 2016 16:21
I think dcalc would would have the math bits? oref0 would contain the logic to load and save json in case of openaps
so we import dcalc to both openaps and nightscout as a dependency that contains all at the algorithms, but the where the data comes from and is put into is totally different
@mgranberry right, yes.
Jason Calabrese
@jasoncalabrese
Jan 12 2016 16:23
so determine-basal.js would be in dcalc? to me that seems like the heart of oref0
but would be nice to use that in BWP, for the basal recomendation
Sulka Haro
@sulkaharo
Jan 12 2016 16:28
Ah yes that's the grey area piece of code
Jason Calabrese
@jasoncalabrese
Jan 12 2016 16:29
got really lucky with a high temp before breakfast today, need to find a way to replicate it, 0 breakfast spike, strange but good
Screen Shot 2016-01-12 at 8.27.04 AM.png
Sulka Haro
@sulkaharo
Jan 12 2016 16:29
But also yes, NS would benefit from using it for the open loop
Do you pre-bolus usually?
Jason Calabrese
@jasoncalabrese
Jan 12 2016 16:30
yeah, the carbs were 20mins later
and only half of them, he hasn't had the milk yet
Scott Leibrand
@scottleibrand
Jan 12 2016 16:38
I was thinking that if we know what time breakfast will be, we could have oref0 adjust the target range over the 2h or so prior: initially raise it to get BG up, then lower it to get +IOB and insulin activity, and then normalize it by 30m after breakfast.
Matthias Granberry
@mgranberry
Jan 12 2016 16:39
I think you will run into some insulin resistance issues if you do that
Scott Leibrand
@scottleibrand
Jan 12 2016 16:43
What do you mean?
Jason Calabrese
@jasoncalabrese
Jan 12 2016 16:46
exactly @scottleibrand, I think we need a way to schedule it, maybe manually the night before
Oskar Pearson
@oskarpearson
Jan 12 2016 16:47
Hey all. I've a question for you. I'd like to create some sort of visibility for Siobhan of what the closed loop system is doing, but the device is going to be offline and won't have nightscout access. Are all the visibility options people are using on their phones/watches effectively displaying nightscout information? Is there a solution where things are being pushed to a iphone or even apple watch directly?
Scott Leibrand
@scottleibrand
Jan 12 2016 16:48
Is hotspotting her phone not an option?
we briefly had ours set up a local hotspot so we could access a web server on the Pi itself, serving pebble.json. Could also access the openaps-predict visualizations, or even run Nightscout on the Pi itself...
i have never figured out how to get the Pi to switch between being a hotspot and being a client. Maybe with two wifi adapters?
Matthias Granberry
@mgranberry
Jan 12 2016 16:57
@scottleibrand decreasing basal rates will influence insulin resistance for a 2-6h period
Scott Leibrand
@scottleibrand
Jan 12 2016 17:06
We haven't really seen problems with rebounds or resistance after long low temps. Any idea whether that effect has ever been quantified?
For the long-zero-temp ketone thing, I wonder if we'd want to put a minimum on the total insulin delivery over the length of the pumphistory: maybe set the minimum low-temp to something like 25% of normal if total insulin delivery over that timeframe drops to 25% of normal?
Or 50%, or who knows. so with 50%, if your loop does iter_pump_hour 4, then after 2 hours of temp to zero (or 4 hours averaging 50%) it wouldn't low-temp below 50%
Scott Leibrand
@scottleibrand
Jan 12 2016 17:14
or could do something like minimum temp basal = (current basal - pumphistory average basal) once 4h average drops below 50%
Jason Calabrese
@jasoncalabrese
Jan 12 2016 17:15
doesn't seem like we should stop low temping below some level
think alarms around long 0 temps would help more there
Matthias Granberry
@mgranberry
Jan 12 2016 17:15
I think extended non-delivery of insulin is asking for ER visits.
Scott Leibrand
@scottleibrand
Jan 12 2016 17:15
This wouldn't stop it, just ensure a trickle. Agreed alarms are also needed.
Matthias Granberry
@mgranberry
Jan 12 2016 17:15
for DKA.
Scott Leibrand
@scottleibrand
Jan 12 2016 17:16
I'm trying to imagine the situation where insulin activity ever gets to zero. I've never seen that.
I've only ever seen long low temps when bolus insulin is still active at the start.
Matthias Granberry
@mgranberry
Jan 12 2016 17:17
I worry about sensor issues or calibration problems coupled with an SGLT-2 inhibitor
Scott Leibrand
@scottleibrand
Jan 12 2016 17:17
When that all decays, BG starts to rise, and insulin delivery resumes. If the low temps were really long, it'll rebound faster and trigger a high temp.
Yeah, sensor issues was one thing that occurred to me.
Remind me which mechanism the SGLT-2 inhibitors trigger?
Matthias Granberry
@mgranberry
Jan 12 2016 17:24
the SGLT-2 inhibitors prevent the reuptake of glucose by the kidneys, bringing BG down in a glucose-dependent but insulin-independent manner
they can cause DKA at normoglycemic levels in T1s, but they seem to work very well in us otherwise
the later generations work extremely well
Jason Calabrese
@jasoncalabrese
Jan 12 2016 17:31
along with the cut off based on time we also need something based on BG level, below 55 (and probably a little higher) seems like you'd be better off continuing the temp to 0
Scott Leibrand
@scottleibrand
Jan 12 2016 17:31
Yeah, seems like SGLT-2 inhibitors would be a contraindication for using a closed loop like OpenAPS.
If we're worried about people combining those, we could add a warning to the docs.
Otherwise, I'd say we design for people not using them.
Alex
@SuicidalTissue
Jan 12 2016 18:30
Little bit off topic, does anyone know where i can get a carelink USB shipped to the UK?
Oskar Pearson
@oskarpearson
Jan 12 2016 18:31
@SuicidalTissue hey there.
Alex
@SuicidalTissue
Jan 12 2016 18:32
@oskarpearson hi
Oskar Pearson
@oskarpearson
Jan 12 2016 18:32
We're in London. Many of us are moving away from the Carelink due to poor range. (The alternatives are still not great, but have been better for us). If you're having difficulty getting a carelink it might be worth skipping straight to the alternative.
It's still fairly early days - but https://github.com/oskarpearson/mmeowlink/wiki is where the info is living
Alex
@SuicidalTissue
Jan 12 2016 18:33
How is the compatibility with Openaps?
Oskar Pearson
@oskarpearson
Jan 12 2016 18:33
There's also a gitter oskarpearson/mmeowlink channel that you can search for
It's been designed for OpenAPS
Sulka Haro
@sulkaharo
Jan 12 2016 18:33
Yeah @SuicidalTissue I'd recommend using the CC1111 stick or ERF. Also - is local MedTronic not shipping you one? I think I got my first one in London
Alex
@SuicidalTissue
Jan 12 2016 18:33
Ah right, this looks brilliant then, I've been searching for a carelink for days now
They're being a bit iffy since I've imported my pump from canada
Oskar Pearson
@oskarpearson
Jan 12 2016 18:35
BTW: we originally spoke to our Diabetes specialist nurse and she was going to organise a second hand one that they had lying around. You might try that route if you want
The ERF stick setup is at https://github.com/oskarpearson/mmeowlink/wiki/Writing-Firmware-to-the-ERF-stick - I need to expand the documents
Sulka Haro
@sulkaharo
Jan 12 2016 18:36
LOL @jasoncalabrese have you checked what the Nightscout daily report graphs look like with OpenAPS?
Oskar Pearson
@oskarpearson
Jan 12 2016 18:37
Buying the TI debugger can be quite expensive (£50). I'm in the process of building a batch of ERF devices to send to people though - send me a message with the title 'ERF' and your address if you want me to add you to the list. It'll take a little time though as I need to sort out a backlog. I'm on oskar@deckle.co.uk
Jason Calabrese
@jasoncalabrese
Jan 12 2016 18:39
@sulkaharo they look ok for me? what are you seeing
Sulka Haro
@sulkaharo
Jan 12 2016 18:39
I need to hide notes
Jason Calabrese
@jasoncalabrese
Jan 12 2016 18:39
yeah, the notes are a mess
Oskar Pearson
@oskarpearson
Jan 12 2016 18:39
If you're set on USB devices you would need to order the TI stick and follow the mmcommander setup. It's much simpler than the ERF stick. Note that using a USB stick increases the power consumption a lot, so you need a bigger battery.
Jason Calabrese
@jasoncalabrese
Jan 12 2016 18:41
should probably hide most of those notes, I need to look at your js for converting the history
keep forgetting about that
Oskar Pearson
@oskarpearson
Jan 12 2016 18:41
I've got to run for now. Send me an @oskarpearson or catch me on oskarpearson/mmeowlink if you need a hand
Jason Calabrese
@jasoncalabrese
Jan 12 2016 18:41
is there a PR?
Oskar Pearson
@oskarpearson
Jan 12 2016 18:41
Oh - also - local hackspaces may have a GoodFet available which could save you some money
bye
Alex
@SuicidalTissue
Jan 12 2016 18:52
Do you mind shooting us some links to suppliers?
Rachel Sandlain
@audiefile
Jan 12 2016 18:55
Suppliers for what exactly? You can get the ti stick and the cc debugger from ti.com the stick is called cc1111
Rachel Sandlain
@audiefile
Jan 12 2016 19:02
I have no idea, you've escaped my realm of knowledge there.
Jason Calabrese
@jasoncalabrese
Jan 12 2016 19:07
Alex
@SuicidalTissue
Jan 12 2016 19:31
@jasoncalabrese Thanks. I'm looking at the docs and trying to figure out how the hell to connect it to the Pi. Any docs on this?
Jason Calabrese
@jasoncalabrese
Jan 12 2016 19:32
not sure what pins to use to wire it to the pi
Alex
@SuicidalTissue
Jan 12 2016 19:57
@stavlor brill, thanks. Just ordered that and a lovely chinese cc debugger. Any diagrams for wiring that to the pi
Sulka Haro
@sulkaharo
Jan 12 2016 20:00
Arght getting "ImportError: No module named mmhistorytools" again on a new install. I have openaps-contrib installed, but no luck
Jason Calabrese
@jasoncalabrese
Jan 12 2016 20:04
thats an other repo
but you're probably not using it
I started setting it up, but didn't ever get around to setting up https://github.com/loudnate/openaps-monitor
Sulka Haro
@sulkaharo
Jan 12 2016 20:06
Arght yes just figured it out
Ben West
@bewest
Jan 12 2016 20:13
python has this concept of shared namespaces
where modules can plant themselves in some common namespace
I'm increasingly skeptical
it's been super painful and hard to get right
"namespace packages"
Sulka Haro
@sulkaharo
Jan 12 2016 20:42
Right, I now have an Edison that's setup and working to run the loop
Still waiting for the ERF sticks to make it small
Jason Calabrese
@jasoncalabrese
Jan 12 2016 20:47
nice, what's the rig look like?
I was planning on flashing my cc1111 last night, but feel asleep putting the kids to bed
eszcloud
@eszcloud
Jan 12 2016 20:51
@oskarpearson have you tried @loudnate's monitor?
I really like it and that it doesn't rely in internet access
uses hotspot on phone and then can view the data in a webpage (phone or computer)
Sulka Haro
@sulkaharo
Jan 12 2016 20:57
@jasoncalabrese don't have a good case, sadly. Trying to 3D print something right now
Jason Calabrese
@jasoncalabrese
Jan 12 2016 21:00
just a little otg cable to the CC, edison and battery?
and pulling bgs from NS
Sulka Haro
@sulkaharo
Jan 12 2016 21:03
yes
without the USB cable it'd be tiny
sulkaharo @sulkaharo is contemplating if it's worth soldering a custom USB cable at 11PM
Paul Andrel
@stavlor
Jan 12 2016 21:24
what are you trying to cable?
Sulka Haro
@sulkaharo
Jan 12 2016 21:25
smallest possible package for the CC1111 and Edison
Paul Andrel
@stavlor
Jan 12 2016 21:25
hmm
Sulka Haro
@sulkaharo
Jan 12 2016 21:27
Oh even more dangerous idea! The CC1111 board exposes the USB mount points nicely, could solder the micro-USB cable there directly :D
Paul Andrel
@stavlor
Jan 12 2016 21:27
nice
heh
Toby Canning
@TC2013
Jan 12 2016 21:31
No, solder the serial connection, much less power draw than the USB. Haven't tested it yet, but I just flashed the CC1111 with subg_rfspy firmware
Paul Andrel
@stavlor
Jan 12 2016 21:33
true much lower power draw that way
Sulka Haro
@sulkaharo
Jan 12 2016 21:36
oh I could then hook it up the same way as the ERF?
@tc2013 is the wiring diagram from Edison to CC1111 up somewhere with that solution?
Toby Canning
@TC2013
Jan 12 2016 21:45
Yes, let me grab it
I'm soldering it now to test also
Sulka Haro
@sulkaharo
Jan 12 2016 21:50
Cool
Using an Edison or Pi?
Toby Canning
@TC2013
Jan 12 2016 21:53
Edison
Sulka Haro
@sulkaharo
Jan 12 2016 21:53
<3
Toby Canning
@TC2013
Jan 12 2016 21:53
Edison to RileyLink UART Connections.png
Sulka Haro
@sulkaharo
Jan 12 2016 21:54
And that's the DEBUG connector
Toby Canning
@TC2013
Jan 12 2016 21:55
Edison to RileyLink UART Connections.png
Sulka Haro
@sulkaharo
Jan 12 2016 21:58
the CC1111 pinouts don't seem to match the diagram?
Jason Calabrese
@jasoncalabrese
Jan 12 2016 22:06
that's the RL not the CC
Toby Canning
@TC2013
Jan 12 2016 22:06
They should be the same pinouts
Sulka Haro
@sulkaharo
Jan 12 2016 22:07
I guess maybe the labeling is just different?
Toby Canning
@TC2013
Jan 12 2016 22:08
Well, I know they are programmed over the same pins because that worked fine
Sulka Haro
@sulkaharo
Jan 12 2016 22:08
:D
Toby Canning
@TC2013
Jan 12 2016 22:09
They just don't label the UART pins at all
Sulka Haro
@sulkaharo
Jan 12 2016 22:13
And I'd use the cc debugger to flash the .hex on the stick. Latest mmeowlink supports this already?
I recall @oskarpearson mentioned the EU radios will need tuning
Or some, last I used Oskar's tuning settings, our pump didn't like them at all
Sulka Haro
@sulkaharo
Jan 12 2016 22:26
Right, CC1111 flashed. I think I need to hit the bed now, I'll do the soldering to Edison board tomorrow. Poke me if with your results.
Alin Artiom Kenibasov
@ktomy
Jan 12 2016 22:30
second night with loop passed. It see,s that my son have got cold or something, but, our usual TDD is abou 10-12u, but since yesterday evening it was 24u (including u4 injected by pen). But loop still managed to keep gb at about 200, today it wa between 200 and 350.
in this night I have doubled all constants and will monitor what happens
Scott Leibrand
@scottleibrand
Jan 12 2016 22:32
all constants? doubling sensitivity would make the loop less aggressive, not more
Alin Artiom Kenibasov
@ktomy
Jan 12 2016 22:35
:) doubling not in mathematical but in quantity of insulin sense
lowered isf from 200 to 80
Scott Leibrand
@scottleibrand
Jan 12 2016 22:35
ok, cool
Oskar Pearson
@oskarpearson
Jan 12 2016 23:17
@sulkaharo we can tune the radio parameters from the command line to match mmcommander now thanks to the subg rfspy code - it - wont require reflashing.
@SuicidalTissue the wirelessthings modules all use a srf device for their core radio module. So the pins that you connect for writing the USB firmware will be the same as the pins on the erf stick. They are going to be difficult to connect to on the pi board, as you can't use the clips I use for the erf stick as they aren't near the edges of the board
but it's probably doable if you make a jig of some form, solder wires onto the pins, or just have someone to help you hold the connecting wires in place
Dragan Maksimovic
@dm61
Jan 12 2016 23:21
I'd be careful about artificially lowering isf
Oskar Pearson
@oskarpearson
Jan 12 2016 23:23
i am going to do some experimentation with local hotspots to see if I can get the device to correctly multi home - with a wifi at home and a hotspot if it's not available
i was looking for something that would push alerts to an app or similar if things are out of range, for example
over bluetooth
Jason Calabrese
@jasoncalabrese
Jan 12 2016 23:40
This message was deleted
This message was deleted