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

8th
Feb 2016
Scott Leibrand
@scottleibrand
Feb 08 2016 01:56
@bewest what was that link to the directions you used to set up BT tethering?
Ben West
@bewest
Feb 08 2016 01:57
hmmm
there are a few different ways to do it
I found this most helpful
Scott Leibrand
@scottleibrand
Feb 08 2016 01:59
thx. gonna try it with an Edison
Ben West
@bewest
Feb 08 2016 02:00
did you install my debs?
ah
hmmm
the test scripts he mentions are not distributed properly in the packages
so not sure where they are on your system
on many systems, they end up zipped somewhere in /usr/share/examples/bluez-test-scripts or similar
Scott Leibrand
@scottleibrand
Feb 08 2016 02:00
k
Ben West
@bewest
Feb 08 2016 02:01
also, I did not modify the scirpts because what i was after was slightly different
he's setting up reverse, where rpi is offering net services
and you associate to it
that's what the bridge set up and modifications are for
I had to run dhclient on the new bnep0 interface, just needed to run test-pan after pairing to the phone via bluetoothctl
the daveconroy link is probably good for what you want and automating it
Scott Leibrand
@scottleibrand
Feb 08 2016 02:05
k
Scott Leibrand
@scottleibrand
Feb 08 2016 03:21
seems like http://www.redtreerobotics.com/ip-over-bluetooth-with-linux/ (the "robot" side) is what I want, but:
edison@edison2 ~ $ sudo /root/bluez-5.28/test/test-network DC:2B:2A:45:02:C0 nap
Traceback (most recent call last):
  File "/root/bluez-5.28/test/test-network", line 42, in <module>
    iface = network.Connect(service)
  File "/usr/lib/python2.7/dist-packages/dbus/proxies.py", line 70, in __call__
    return self._proxy_method(*args, **keywords)
  File "/usr/lib/python2.7/dist-packages/dbus/proxies.py", line 145, in __call__
    **keywords)
  File "/usr/lib/python2.7/dist-packages/dbus/connection.py", line 651, in call_blocking
    message, timeout)
dbus.exceptions.DBusException: org.bluez.Error.Failed: Input/output error
any idea what that means?
Got the same thing on both edisons, one with bluez 5.28 from the package, and one with 5.33 (where openxshareble is working)
Jason Calabrese
@jasoncalabrese
Feb 08 2016 03:31
can the edison do both? that would be great
Scott Leibrand
@scottleibrand
Feb 08 2016 03:34
I have never gotten it working anywhere. Would love to have someone who has walk me through it.
Scott Leibrand
@scottleibrand
Feb 08 2016 03:42
I think I just got it working on the Pi
bnep0 Link encap:Ethernet HWaddr 00:19:0e:17:ee:8d
inet addr:172.20.10.3 Bcast:172.20.10.15 Mask:255.255.255.240
strange that the same thing doesn't work on the edison
Ben West
@bewest
Feb 08 2016 04:28
yes
uh
I think the format is aa_bb_cc_dd_ee_ff
no
that likely means
hmm
I think you want test-nap
oh, that's working on pi
did you run dhclient or similar against it?
make sure bluetoothd --experimental is running
on systemd, it's uhhh systemctl status bluetooth.service I think
Scott Leibrand
@scottleibrand
Feb 08 2016 04:34
test-nap is for the AP, not the client.
the 5.33 one has --experimental, or openxshareble wouldn't be working.
The bnep interface never gets created: test-network creates that if it doesn't get an I/O error.
Scott Leibrand
@scottleibrand
Feb 08 2016 04:54
edison@edison2 ~ $ systemctl status bluetooth.service
Failed to get D-Bus connection: Unknown error -1
Ben West
@bewest
Feb 08 2016 05:16
hmm
sounds like bluetoothd is not running
can try running it manually bluetoothd -E &
oh, this is while openxshareble is working already?
Scott Leibrand
@scottleibrand
Feb 08 2016 05:17
that was edison2, which isn't configured right. so let's focus on edison-rl, where openxshareble is working
Ben West
@bewest
Feb 08 2016 05:18
so that one has bluetoothd working currently
Scott Leibrand
@scottleibrand
Feb 08 2016 05:18
yes
Ben West
@bewest
Feb 08 2016 05:18
and test-network raises that error
Scott Leibrand
@scottleibrand
Feb 08 2016 05:18
correct
Ben West
@bewest
Feb 08 2016 05:18
failed input output error
oh
did you pair to the device via bluetooothctl?
think this is only needed first time
I did it via bluetoothctl
or did I
maybe I used dbus-send, not sure
Scott Leibrand
@scottleibrand
Feb 08 2016 05:21
I was able to pair the phone (getting a pin code on both and accepting it both places) and then connect them in the bluetooth settings
did all that through bluetoothctl, both on the edison (not working) and pi (working)
Ben West
@bewest
Feb 08 2016 05:21
hmmm
so edison was able to pair to it
Joakim Ornstedt
@joakimornstedt
Feb 08 2016 05:21
@timomer I have seen HAPP but I have created my own app just so I now exactly what every line of code dose.
Ben West
@bewest
Feb 08 2016 05:22
I have no idea, you did same steps on rpi and edison it sounds like
it worked on rpi and not on edison
Scott Leibrand
@scottleibrand
Feb 08 2016 05:23
yup
Ben West
@bewest
Feb 08 2016 05:23
I'd have to really play with it, I was surprised it more or less worked first time
well, I tried it 9 months ago as I was boarding a plane and failed
was hoping to get both laptop and openaps running off iphone hotspot on airline wifi
maybe less than 9 months ago
Scott Leibrand
@scottleibrand
Feb 08 2016 05:24
I could see that working on a pi with two NICs
or a laptop
Ben West
@bewest
Feb 08 2016 05:24
but this time it worked pretty quickly
yeah, thinking about adding second nic
err, second wifi dongle maybe
Scott Leibrand
@scottleibrand
Feb 08 2016 05:24
but iphones can't be in both AP and client mode simultaneously afaik
Ben West
@bewest
Feb 08 2016 05:24
always do hostapd?
Joakim Ornstedt
@joakimornstedt
Feb 08 2016 05:25
@scottleibrand @danamlewis @bewest @jasoncalabrese What is your experience on sens under the night. Is it constantly the same or different like just befor time to wake up?
Scott Leibrand
@scottleibrand
Feb 08 2016 05:25
I did that on another Pi
Ben West
@bewest
Feb 08 2016 05:25
iphone can do wifi on it's own while offering ble network though
I think
I use one sensitivity all day
Scott Leibrand
@scottleibrand
Feb 08 2016 05:25
ah, that'd be interesting
@joakimornstedt we don't vary sensitivity throughout the day/night, just basals
Ben West
@bewest
Feb 08 2016 05:26
we need tools to analyze basal vs sensitivity vs carb, etc
we don't even have good tools to analyze basals yet
would be great to offer a basal trainer/tuner
same as mmtune
but for basal rates
then add on with sensitivity-tuner
and carb-effect-tuner
Scott Leibrand
@scottleibrand
Feb 08 2016 05:31
gonna try setting bluetooth up from scratch on another pi and see if I can get it working there using the standard packages, or if it's somehow dependent on something I did the last time I fiddled with all this.
Joakim Ornstedt
@joakimornstedt
Feb 08 2016 05:34
@bewest I want to team up on creating a tool for analyzing basal and sens. We run with dubbel as sensitive under the night, but I plan to maybe very the sens from hour to hour
Ben West
@bewest
Feb 08 2016 05:34
you can't use the standard packages
the debs I gave you are not distributed
Scott Leibrand
@scottleibrand
Feb 08 2016 05:35
those are for openxshareble
Ben West
@bewest
Feb 08 2016 05:35
well
hmm
Scott Leibrand
@scottleibrand
Feb 08 2016 05:35
no reason I should need nonstandard just for tethering
Ben West
@bewest
Feb 08 2016 05:35
well, I was unable within last year to do it using standard packages alone
looks like those guides also demand more recent versions of bluez than are often available, not sure
Scott Leibrand
@scottleibrand
Feb 08 2016 05:36
this one is Version 5.23
have also seen 5.28 standard
Ben West
@bewest
Feb 08 2016 05:36
yeah, seems old
that's the working pi?
Scott Leibrand
@scottleibrand
Feb 08 2016 05:37
this is the one I'm setting up new
Version 5.33 on the working one
Ben West
@bewest
Feb 08 2016 05:37
yeah, I like > 5.30
Joakim Ornstedt
@joakimornstedt
Feb 08 2016 05:38
@scottleibrand Do ju think if we have a tool that we would find that the sens very under the night/day?
Dana Lewis
@danamlewis
Feb 08 2016 05:40
We just started testing a sensitivity detector.
Scott Leibrand
@scottleibrand
Feb 08 2016 05:40
possibly. we have a PhD who's willing to help us analyze the data. see #machine-learning on the OmniAPS Slack (where I just invited you)
what we have so far detects 24h changes in sensitivity
Dana Lewis
@danamlewis
Feb 08 2016 05:40
It doesn't seem to vary that much hour to hour, but may be different in kids, and we need to do more analysis anyway
Scott Leibrand
@scottleibrand
Feb 08 2016 05:40
we would like to analyze historical data and see if we can detect anything that varies by time of day
but probably more important first step is to figure out basal tuning
since we know that varies a lot by time of day for just about everyone
@bewest bound to 172.20.10.4 -- renewal in 35836 seconds.
pi@rpi2 ~ $
that's with Version 5.23
so that process works flawlessly on Pi, regardless of bluetooth version
gonna see if I can start over on Edison
Ben West
@bewest
Feb 08 2016 05:43
yeah, guess it should/would
expected it to work on edison as well though
i remember something funny about creation of bnep0
Joakim Ornstedt
@joakimornstedt
Feb 08 2016 05:44
@danamlewis Interesting, How dose the sensitivity detector work?
Scott Leibrand
@scottleibrand
Feb 08 2016 05:45
@joakimornstedt openaps/oref0#58
Dana Lewis
@danamlewis
Feb 08 2016 05:45
^
Joakim Ornstedt
@joakimornstedt
Feb 08 2016 05:49
#58 Tise approach is interesting.
This....
Scott Leibrand
@scottleibrand
Feb 08 2016 07:14
more hours of trying, and I can't seem to figure out this Input/Output error. I did notice that the blue tethering bar shows up momentarily on the iphone before it fails and goes away again...
Jim Matheson
@jmatheson
Feb 08 2016 07:19
damn, the old firmware only lets me change my active insulin time by whole hours
Scott Leibrand
@scottleibrand
Feb 08 2016 07:30
@TC2013 you said you were able to get bluetooth tethering to work on the Edison?
did you use the test-network DC:2B:2A:45:02:C0 nap method to turn up bnep0, or something else?
Aaron Michelson
@Aaybob
Feb 08 2016 07:39
Hi, I'm trying to follow the visualization documentation.....I don't see OpenAPS on my NightScout website....It is the first step I didn't understand and believe may have erred on....I did the release install and initiated a profile. I then repeated the documentation steps: 1.Set-up Github at http://www.nightscout.info/wiki/welcome/monitor-cgm-web and substituted
https://github.com/nightscout/cgm-remote-monitor/tree/dev for https://github.com/nightscout/cgm-remote-monitor/fork I don't know that the new git is active anywhere because I didn't do a clone in a directory on RPi2 like I'm used to doing.
blob
Jason Calabrese
@jasoncalabrese
Feb 08 2016 07:45
looks like you're running the master branch, 0.8.4, instead of the dev branch
use can pick a different branch from azure
Jim Matheson
@jmatheson
Feb 08 2016 07:46
yep, switch to dev and you should be good
Aaron Michelson
@Aaybob
Feb 08 2016 07:47
Oh great, thanks for the help. Looking around Azure now
blob
Jim Matheson
@jmatheson
Feb 08 2016 07:48
it’s under publishing -> continued deployment I think
you have to disconnect then reconnect and choose dev
Aaron Michelson
@Aaybob
Feb 08 2016 07:51
I "stopped" does that disconnect?
Jim Matheson
@jmatheson
Feb 08 2016 07:52
no. see in your screenshot….at the bottom in the column to the right of that graph. choose continued deployment. then from there you will see disconnect
you need to disconnect from github and reconnect to the dev branch. i think stop just stops your web service.
Aaron Michelson
@Aaybob
Feb 08 2016 07:55
blob
Jim Matheson
@jmatheson
Feb 08 2016 07:55
yep. at the top right in the black bar
Microsoft UI is so intunitive :)
Aaron Michelson
@Aaybob
Feb 08 2016 08:04
thanks ...on to step four
Jim Matheson
@jmatheson
Feb 08 2016 08:12
nice
Aaron Michelson
@Aaybob
Feb 08 2016 08:25
blob
I don't have any of these reports...
Where did one first encounter them?
timomer
@timomer
Feb 08 2016 08:59
@joakimornstedt snap, that was my thought process as well. feel free to dip in and reuse any code. would be great to see yours when you are ready to share
Joakim Ornstedt
@joakimornstedt
Feb 08 2016 10:39
The date in entries table, does any one know the format? Like -> "date": 1443357159000 - I dos not semes to ticks or?
@timomer Do you use HAPP today, and follow suggestions from HAPP?
timomer
@timomer
Feb 08 2016 10:45
@joakimornstedt yep, been using it as a Open Loop solution for almost 6 months, I have it set to notify me every 15mins if an adjustment in Temp Basal is suggested. Also I log my meals there to use the Bolus Wizard. surprisingly useable just being in open loop
Joakim Ornstedt
@joakimornstedt
Feb 08 2016 10:46
@timomer Cool :-)
timomer
@timomer
Feb 08 2016 10:48
welcome to join the chat room https://gitter.im/timomer/HAPP
Sulka Haro
@sulkaharo
Feb 08 2016 16:54
12669640_10156515112165344_402639481810170654_n.jpg
Found the diagram to setup the APS
Eric
@ecc1
Feb 08 2016 16:54
@scottleibrand I missed the start of this saga :-) What's the goal of pairing with your phone? If it's network connectivity, why not run a local WiFi hotspot on the phone? As for the BT difficulties, it wouldn't surprise me if there were kernel limitations on the Edison. I found some already related to BLE ("discovery filters" to specify BLE-only don't work, but they do in more recent kernels.)
Jason Calabrese
@jasoncalabrese
Feb 08 2016 17:09
nice to be able to watch mmtune remotely https://github.com/jasoncalabrese/indy-ti-edison/blob/master/monitor/mmtune.txt, I'd like to get some of that into the device status
Aaron Michelson
@Aaybob
Feb 08 2016 17:36
blob
from: https://github.com/openaps/docs/blob/master/docs/Automate-system/vizualization.md
I haven't seen the reports suggested. What am I missing?
Jason Calabrese
@jasoncalabrese
Feb 08 2016 17:39
the reports have different names depending on how you set them up
predict/oref0.json might be called suggest.json and control/enacted.json might be enact/enacted.json
Aaron Michelson
@Aaybob
Feb 08 2016 17:39
ohhhh
Jason Calabrese
@jasoncalabrese
Feb 08 2016 17:39
etc
Pete Schwamb
@ps2
Feb 08 2016 17:42
I’d like to get mmtune working in python, since edison doesn’t seem to support ruby well. I’ll see if I can get that going today.
Eric
@ecc1
Feb 08 2016 17:45
My life with Edison improved a lot when I upgraded its userland to Debian stretch, from jessie. More things work out of the box, like BLE and nodejs. But of course the kernel is still outdated.
Aaron Michelson
@Aaybob
Feb 08 2016 17:53
So for battery.json I would "add" a report that runs openaps use <> read_battery_status? And then when upload-status is invoked it invokes the report? And would clock_zoned.json come from read_clock?
@ecc1 I just deleted nodejs from a new empty hard drive. was it important to keep that program?
Jason Calabrese
@jasoncalabrese
Feb 08 2016 17:56
@ps2 that would save a bunch of space on the edison too, the root partition is getting full
Eric
@ecc1
Feb 08 2016 18:03
@Aaybob On what machine? The oref0 functions are written in Javascript, so they need nodejs. And if you're running your own cgm-remote-monitor (Nightscout server), it does too.
Pete Schwamb
@ps2
Feb 08 2016 18:03
@jasoncalabrese are you actually running ruby mmtune on edison?
Jason Calabrese
@jasoncalabrese
Feb 08 2016 18:04
yeah
on 2 edisons
Pete Schwamb
@ps2
Feb 08 2016 18:04
Which ruby did you use?
Jason Calabrese
@jasoncalabrese
Feb 08 2016 18:05
it does work, I think the issue @TC2013 had was trying to use wheezy instead of jessie
$ ruby -v
ruby 2.1.5p273 (2014-11-13) [i386-linux-gnu]
installed with sudo apt-get install ruby-full
Pete Schwamb
@ps2
Feb 08 2016 18:06
Oh yeah, you’re on debian. I am trying to get this working on yocto.
Jason Calabrese
@jasoncalabrese
Feb 08 2016 18:06
I guess yocto can read the lipo voltage
that would be nice
Pete Schwamb
@ps2
Feb 08 2016 18:07
There are sleep modes in yocto, too.
Jason Calabrese
@jasoncalabrese
Feb 08 2016 18:07
I've said it before, but it's just so awesome using the RL for real now
Dana Lewis
@danamlewis
Feb 08 2016 18:08
:+1:
Pete Schwamb
@ps2
Feb 08 2016 18:08
:)
Jason Calabrese
@jasoncalabrese
Feb 08 2016 18:09
hoping to battery makes it through the day, but so far first day with RL at school is going good
Eric
@ecc1
Feb 08 2016 18:10
The sleep modes are all kernel features, right? So should be doable on Debian using the right sysfs magic
Jason Calabrese
@jasoncalabrese
Feb 08 2016 18:10
I think the battery-voltage too, saw some threads about it
Eric
@ecc1
Feb 08 2016 18:14
Just looked at the source code for battery-voltage, it reads /sys/devices/platform/pmic_ccsm/battery_level Don't have my Edison handy atm, does this still appear in ubilinux?
Jason Calabrese
@jasoncalabrese
Feb 08 2016 18:16
it's not there for me, but there is /sys/devices/platform/pmic_ccsm/
Eric
@ecc1
Feb 08 2016 18:18
Huh. I just assumed ubilinux used the latest yocto kernel. Will have to investigate!
Jason Calabrese
@jasoncalabrese
Feb 08 2016 18:20
once we figure this stuff out getting a custom image will be great
Aaron Michelson
@Aaybob
Feb 08 2016 18:32
Visualizing documentation implies I might have: clock-zoned.json iob.json oref0.json enacted.json battery.json reservoir.json and status.json by different names. I haven't by following directions added a report similar to clock-zoned, battery, or reservoir. If so would I add reports using read_clock, read_battery_status, and reservoir. Those being the names documented in -h.
Jason Calabrese
@jasoncalabrese
Feb 08 2016 18:37
the -zoned files are created using a tz tool, see https://github.com/openaps/oref0/blob/master/bin/ns-uploader-setup.sh#L82
Aaron Michelson
@Aaybob
Feb 08 2016 18:48
openaps-report add: error: argument device: invalid choice: 'tz' (choose from u'calculate-iob' is the result
you may have to adjust things to match your existing reports
Scott Leibrand
@scottleibrand
Feb 08 2016 19:04
@ecc1 the reason I want BT tethering is so I don't have to manually activate wifi tethering on my iPhone every time I leave the house. With BT the Pi can activate it on demand.
Jason Calabrese
@jasoncalabrese
Feb 08 2016 19:05
maybe lower battery use too
Aaron Michelson
@Aaybob
Feb 08 2016 19:12
got clock-zoned.json and my openaps.ini file was updated with the new commands. Thanks, I think I'm close now to having visualization.
Jason Calabrese
@jasoncalabrese
Feb 08 2016 19:15
great
PieterGit
@PieterGit
Feb 08 2016 20:10
i think i found a small bug in ns-upload-setup.sh.... gather alias should be openaps get-bg ; openaps get-settings instead of openaps get-bg && openaps get-settings
if there is no bg value, why should the get-settings be uploaded. or am i missing something here
is there an more efficient way to debug openaps commands than: openaps command ; echo $?
Ben West
@bewest
Feb 08 2016 20:43
I'm prepping a new high level configurable object for openaps
right now there are four
vendors, devices, reports, aliases
these all have matching add, show, remove commands, plus "the main thing" that they do
eg, devices have use, reports have invoke
the proposed fifth thing is called schedules
so openaps schedule {add,show,remove,???}
could go with invoke again, or introduce a new word, maybe trigger
Dana Lewis
@danamlewis
Feb 08 2016 20:47
I like a new word, @bewest
Ben West
@bewest
Feb 08 2016 20:50
oh really, why?
leaning towards trigger then
or maybe arm, not sure
probably trigger
Jason Calabrese
@jasoncalabrese
Feb 08 2016 20:51
I'd like a way to do a single commit per loop, if we use invoke for a schedule and that was 1 commit it might make sense
Ben West
@bewest
Feb 08 2016 20:52
currently report invoke only does one commit for the entire batch
was thinking of having a new thing called openaps do, which would run any high level openaps command similarly as a single commit, hopefully
so you could say openaps do preflight gather-data predict-outcomes enact-suggestions
an in theory that would run all the aliases same way as report invoke ...
Jason Calabrese
@jasoncalabrese
Feb 08 2016 20:54
I like that
Ben West
@bewest
Feb 08 2016 20:54
there are few gotchas in the code that complicate doing that, so not sure
will have to parse the alias and emit events to dispatch internally
rather than execing to shell
Jason Calabrese
@jasoncalabrese
Feb 08 2016 20:54
would be nice to get rid of all the bash -c's in the ini too
Ben West
@bewest
Feb 08 2016 20:54
either that or pass some environment flag that says "do not commit"
Jason Calabrese
@jasoncalabrese
Feb 08 2016 20:54
it's mostly for && and ||
Ben West
@bewest
Feb 08 2016 20:55
yeah, maybe --git-skip-all vs --git-part-of-workflow or something
Jason Calabrese
@jasoncalabrese
Feb 08 2016 20:56
ideally enact or a new loop would trigger a commit
Ben West
@bewest
Feb 08 2016 20:57
lot of work to get it all in
think there's a path to get there, but this initial batch will be quite simple/limited
Jason Calabrese
@jasoncalabrese
Feb 08 2016 20:57
small steps
Ben West
@bewest
Feb 08 2016 20:57
yeah, small steps
ok, trigger it is, I guess
Jason Calabrese
@jasoncalabrese
Feb 08 2016 21:03
so for the first step we'd have cron trigger a schedule?
Dana Lewis
@danamlewis
Feb 08 2016 21:05
Yes trigger is good
Ben West
@bewest
Feb 08 2016 21:10
```
No handlers could be found for logger "recurrent"
bewest@bewest-MacBookPro:~/Documents/openaps$ openaps schedule add main "every 5 minutes"
removed schedule://main/RRULE:FREQ=MINUTELY;INTERVAL=5
bewest@bewest-MacBookPro:~/Documents/openaps$ openaps schedule remove main
bewest@bewest-MacBookPro:~/Documents/openaps$ openaps schedule show main --ini
[schedule "main"]
rrule = RRULE:FREQ=MINUTELY;INTERVAL=5
bewest@bewest-MacBookPro:~/Documents/openaps$
added schedule://main/RRULE:FREQ=MINUTELY;INTERVAL=5
{"main": {"phases": "", "rrule": "RRULE:FREQ=MINUTELY;INTERVAL=5"}, "type": "schedule", "name": "main"}
phases =
bewest@bewest-MacBookPro:~/Documents/openaps$ openaps schedule show main --json
```
well, I'm taking a fairly large set of first steps
taking cron out of the loop/setup
I've got a daemon that sits on dbus
Jason Calabrese
@jasoncalabrese
Feb 08 2016 21:11
oh, not small steps then
Ben West
@bewest
Feb 08 2016 21:11
this is ideal in linux, it gives us full access to networkmanager, bluetooth, etc
it's already exporting all the openaps config and stuff
have a webapp with ajax gui that can fetch the properties
but it's not clear to a gui what the status of loop is with cron
so enter the daemon :-)
so this is the reason for the new fifth high level object, the schedules
the daemon will arm upcoming events, then "trigger" them
and there will be a tool to exercise/test from cli
this daemon auto-starts itself if the system needs one of it's methods/properties and it's not already running
Jason Calabrese
@jasoncalabrese
Feb 08 2016 21:14
and the daemon will know not to try talking to the pump at the same time? so you could have get settings ever 15m and loop every 1m, but they wouldn't step on each other?
how would we do things like waiting for new BG
Ben West
@bewest
Feb 08 2016 21:14
hmm, probably only as a happy side-effect initially
well, this shifts the whole architecture to event drive in the daemon
so just need to augment the phase/trigger/schedule so that signals/triggers can kick off other things
eventually dexcom/medtronic/roche will each have their own system service
and openaps will address those directly with things like OnNewBg
but that's down the road a ways
for now, openaps do will simply take list of things and execute them in order
I'll get a global lock so it's impossible for two things to run at same time
however, it's unclear what to do if that does happen
Jason Calabrese
@jasoncalabrese
Feb 08 2016 21:16
maybe wait, but they you get a backlog
wait with timeout maybe
Ben West
@bewest
Feb 08 2016 21:17
yeah, for now, I'm happy to just keep doing things in serial fashion
Eric
@ecc1
Feb 08 2016 21:17
How about "perform" to go with "schedule"
Ben West
@bewest
Feb 08 2016 21:18
the new d-bus based stuff makes it easier to create new plugins that represent high level service
hmm maybe
was thinking that schedules can arm triggers which then fire
or phases mix in there somehow too
well it's true there will be a list of tasks that get queued up ahead of time
Eric
@ecc1
Feb 08 2016 21:23
or "action"?
Ben West
@bewest
Feb 08 2016 21:23
ah, I already use action way too much in the code
Eric
@ecc1
Feb 08 2016 21:23
yeah, it's pretty generic
Ben West
@bewest
Feb 08 2016 21:25
nifty thing with this though is that it can handle asymmetric loops
might be feature without much of a point for our use case
but you can create several loops which are not symmetric... first thursdays of month and things like that
Jason Calabrese
@jasoncalabrese
Feb 08 2016 21:26
I could see that
Ben West
@bewest
Feb 08 2016 21:26
so referesh settings on weekends, etc
Jason Calabrese
@jasoncalabrese
Feb 08 2016 21:26
would be nice to push settings to the pump with it
Ben West
@bewest
Feb 08 2016 21:26
yeah!
need more work on the command, params format
it's mainly the params, not sure how to handle it
it's a large payload, usually
probably easier to get right with the subg_rfspy, ironically
```
added schedule://handle-weekends/RRULE:BYDAY=FR,SA;INTERVAL=1;FREQ=WEEKLY
No handlers could be found for logger "recurrent"
$ openaps schedule add handle-weekends "every Friday, Saturday morning"
```
bewest@bewest-MacBookPro:~/Documents/openaps$
ah, here we go
```
No handlers could be found for logger "recurrent"
```
bewest@bewest-MacBookPro:~/Documents/openaps$ openaps schedule add handle-weekends "every Friday, Saturday at 6am for 6 hours"
added schedule://handle-weekends/RRULE:BYMINUTE=0;BYDAY=FR,SA;BYHOUR=6;FREQ=HOURLY;INTERVAL=6
Scott Leibrand
@scottleibrand
Feb 08 2016 21:32
@bewest your ``` are getting sent as independent messages. need to have them as part of a single mutliline message to get them to render
Ben West
@bewest
Feb 08 2016 21:32
ah, it won't work over irc I guess
Scott Leibrand
@scottleibrand
Feb 08 2016 21:33
can you embed newlines in irc?
Ben West
@bewest
Feb 08 2016 21:33
yeah, I did in the last one
I messed up the prior one
oh
Scott Leibrand
@scottleibrand
Feb 08 2016 21:33
the "here we go" one shows as three messages in gitter
Ben West
@bewest
Feb 08 2016 21:33
no, one of them is messed up
yeah
can't fix it from here, sorry
Scott Leibrand
@scottleibrand
Feb 08 2016 21:33
no worries, just wondering if it's possible to do right in future
scottleibrand @scottleibrand is mostly curious how good their irc implementation is
Ben West
@bewest
Feb 08 2016 21:34
ok test one:
simple one line
```
```
Scott Leibrand
@scottleibrand
Feb 08 2016 21:35
those were out of order, but three separate messages
Ben West
@bewest
Feb 08 2016 21:35
yup
ok, back to dpaste like it's the 90's again
Scott Leibrand
@scottleibrand
Feb 08 2016 21:36
;-)
Sulka Haro
@sulkaharo
Feb 08 2016 21:52
If anyone ever needs to test OpenAPS behavior with noisy CGM data, just have an xDrip and Dex receiver upload to same profile at the same time. :D
Jason Calabrese
@jasoncalabrese
Feb 08 2016 22:20
think I'd need to do that for a while, to get a feel for how xdrip smooths the data compared to dex
always bugged me how much they smoothed it, but for the kind of forecasting we need the smoothing
Jason Calabrese
@jasoncalabrese
Feb 08 2016 22:38
Screen Shot 2016-02-08 at 2.36.37 PM.png
first day of school with the edison/RL
Dana Lewis
@danamlewis
Feb 08 2016 22:38
looks excellent :)
Stephan
@MosiGitHub
Feb 08 2016 22:38
:+1:
Jason Calabrese
@jasoncalabrese
Feb 08 2016 22:38
still anxious for it to get home so we can charge it
can't believe I missed the 100 (+4)
Scott Leibrand
@scottleibrand
Feb 08 2016 22:47
:-)
Aaron Michelson
@Aaybob
Feb 08 2016 23:36
I'm getting a lot of fails while not out of range. Here's a typical log
PREFLIGHT OK
GATHER OK
{"temson":"Eventual BG 160>120, adj. req. rate:5.2 to maxSafeBasal:2.4, temp 2.375 >~ req 2.4000000000000004U/hr"}
No Action Required
PREFLIGHT OK
GATHER OK
{"temp":"absolute","bg":195,"tick":"+3","eventualBG":163,"snoozeBG":163,"reason":"Eventual BG 163>120, adj. req. rate:5.4 to maxSafeBasal:2.4, temp 2.375 >~ req 2.4000000000000004U/hr"}
No Action Required
PREFLIGHT OK
GATHER OK
{"temp":"absolute","bg":196,"tick":"+1","eventualBG":162,"snoozeBG":162,"reason":"Eventual BG 162>120, adj. req. rate:5.4 to maxSafeBasal:2.4, temp 2.375 >~ req 2.4000000000000004U/hr"}
No Action Required
PREFLIGHT FAIL
IF PREFLIGHT WAS OK THEN ENACT OR GATHER ERROR
PREFLIGHT FAIL
IF PREFLIGHT WAS OK THEN ENACT OR GATHER ERROR
PREFLIGHT OK
monitor/clock.json raised bytearray index out of range
Tra
And it comes back to working for a while then fails for a while
Aaron Michelson
@Aaybob
Feb 08 2016 23:48
It seems typicall to work for two hours, fail for four hours, work when I wake up and look at logs, stop working an hour after I go to bed.
But it eventually comes back to working and so I don't want to change things bacause this is my best result so far
but I'm not running the reset usb or mm-link warmups or two enacts...which of those would be a good idea to try first?
Scott Leibrand
@scottleibrand
Feb 08 2016 23:51
it's not enact, if preflight is failing. I would add the mm-warmup || reset-usb stuff to your preflight next