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

10th
Jan 2016
Tim Howard
@tghoward
Jan 10 2016 00:11
@bewest - thanks, will look at mm-stick diagnose thanks.
in the meanwhile, I've put the 33 pages I just grabbed here:
https://github.com/tghoward/decoding-carelink/tree/thowdat/logs/2016-01-09
JaysonEwer
@JaysonEwer
Jan 10 2016 00:13
@bewest After I used sudo, it did drop me into a text file to update commit log.
'''pi@Ewer-Open-APS:~/Documents/openaps $ sudo git pull backup master
warning: no common commits
remote: Counting objects: 9, done.
remote: Compressing objects: 100% (5/5), done.
remote: Total 9 (delta 1), reused 0 (delta 0)
Unpacking objects: 100% (9/9), done.
From /root/my_openaps_je
  • branch master -> FETCH_HEAD
  • [new branch] master -> backup/master
    Auto-merging openaps.ini
    Merge made by the 'recursive' strategy.
    openaps.ini | 8 ++++++++
    1 file changed, 8 insertions(+)'''
Tim Howard
@tghoward
Jan 10 2016 00:14
Actually, not sure `mm-stick diagnose' would make sense because the second time I had the problem, it was with Decocare 0.0.19 and reverting to Decocare 0.0.18 allowed the downloads to continue. As noted here:
bewest/decoding-carelink#156
Ben West
@bewest
Jan 10 2016 00:24
oy
@JaysonEwer sudo chown -R pi ~/Documents/openaps
mm-stick diagnose always makes sense :-)
Tim Howard
@tghoward
Jan 10 2016 00:26
:smile: OK, you got me there!
Ben West
@bewest
Jan 10 2016 00:27
if it doesn't work, you're going to have problems
maybe I rename it to mm-sanity
Tim Howard
@tghoward
Jan 10 2016 00:28
Love it. or mm-get-me-some-sanity
At any rate, I'll do my best to trap it and get more data if the error shows up again.
Ben West
@bewest
Jan 10 2016 00:29
I wrote up some instructions for https://github.com//bewest/openxshareble
on this one, I decided to make it reconfigurable wrt to serial AFTER adding it as device
Tim Howard
@tghoward
Jan 10 2016 00:30
I've been kind of assuming this is up on the GCM side of the pump as it seems like very few (no one?) folks are using enlite and so perhaps there was a message you've missed?
Ben West
@bewest
Jan 10 2016 00:31
hmmm
dunno
the CGM messages are usually quite distinct, in different location than pump history
this is not the y2k thing?
are all pages exhibiting the problem or only some?
mm-decode-history-page.py --model 723 <HistoryFile.data>
it dissects pages nicely
Tim Howard
@tghoward
Jan 10 2016 00:33
it's an iter_pump error, happened twice, in Dec (so before Y2K).
Ben West
@bewest
Jan 10 2016 00:33
but it cleared after a few days?
Tim Howard
@tghoward
Jan 10 2016 00:34
no, first time was with decocare 0.0.18 and that version allowed error to post but process to continue
the decocare update to 0.0.19 for some reason made it so the error killed the process
but, yes, it would go away after a few days
(I'm using the term 'process' loosely here, sorry about bad terminology)
Ben West
@bewest
Jan 10 2016 00:37
no worries
hmm
Tim Howard
@tghoward
Jan 10 2016 00:38
wow, cool command mm-decode-history-page.py
I'll try to go through them. :smile: see if I see anything
Ben West
@bewest
Jan 10 2016 00:42
looks like Sara6E
:-(
some kind of midnight record
I don't like that one
suspect it's different between 722 and 723 maybe
Tim Howard
@tghoward
Jan 10 2016 00:44
Is this what you just found?
pi@betacb ~/bcb-s $ mm-decode-history-page.py --model 723 ReadHistoryData-page-31.data
## START ReadHistoryData-page-31.data
ERROR month must be in 1..12 0000   0x0c 0x00                                  ..
Traceback (most recent call last):
  File "/usr/local/bin/mm-decode-history-page.py", line 6, in <module>
    exec(compile(open(__file__).read(), __file__, 'exec'))
  File "/home/pi/src/decoding-carelink/bin/mm-decode-history-page.py", line 141, in <module>
    main( )
  File "/home/pi/src/decoding-carelink/bin/mm-decode-history-page.py", line 119, in main
    records = find_records(stream, opts)
  File "/home/pi/src/decoding-carelink/bin/mm-decode-history-page.py", line 91, in find_records
    record = parse_record( stream, B, model=opts.model, larger=opts.larger )
  File "/home/pi/src/decoding-carelink/decocare/history.py", line 742, in parse_record
    record.parse( head + date + body )
  File "/home/pi/src/decoding-carelink/decocare/records/base.py", line 66, in parse
    return self.decode( )
  File "/home/pi/src/decoding-carelink/decocare/history.py", line 702, in decode
    return (dict(valid_date=date(*mid).isoformat()))
ValueError: month must be in 1..12
Ben West
@bewest
Jan 10 2016 00:44
yeah, I looked up those lines
I see comments in there telling me I have no idea what I'm doing there
seems specific to 722?
along with bunch of different guesses
Tim Howard
@tghoward
Jan 10 2016 00:46
huh
Ben West
@bewest
Jan 10 2016 00:47
maybe bad merge somewhere
?
thought these were renamed/fixed
hmm
Tim Howard
@tghoward
Jan 10 2016 00:49
that ReadHistory page 31 caught it, if that helps at all ... I'll get the direct link to the file
Ben West
@bewest
Jan 10 2016 00:50
yeah, I see page 31 is troubesome
good find
Ben West
@bewest
Jan 10 2016 00:50
nice
well at this point my tool is not being superbly helpful because it's crashing
Tim Howard
@tghoward
Jan 10 2016 00:51
:smile:
Ben West
@bewest
Jan 10 2016 00:52
bewest@bewest-MacBookPro:~/src/decocare$ git diff
diff --git a/decocare/history.py b/decocare/history.py
index 0f01626..c57ee96 100644
--- a/decocare/history.py
+++ b/decocare/history.py
@@ -699,7 +699,10 @@ class Sara6E(Model522ResultTotals):
   def decode (self):
     self.parse_time( )
     mid = unmask_m_midnight(self.date)[0:3]
-    return (dict(valid_date=date(*mid).isoformat()))
+    try:
+      return (dict(valid_date=date(*mid).isoformat()))
+    except ValueError, e:
+      return (dict(error_date=mid))

 _confirmed.append(Sara6E)

bewest@bewest-MacBookPro:~/src/decocare$
ah interesting
Ben West
@bewest
Jan 10 2016 00:58
so record 24 looks ok
bewest/decoding-carelink#156
what does your csv say happened at that time?
Alin Artiom Kenibasov
@ktomy
Jan 10 2016 00:59
Hi everyone! When uploading treatments to NS, all records are duplicated each time they are uploaded. I can see a deduplicating tool which didn't work for me for the first run. Should I try to make it work, or there is another mechanism that eliminates duplicates (i.e. on NS side api does not insert duplicates in db in certain conditions)?
Ben West
@bewest
Jan 10 2016 00:59
after that it starts to look fairly bogus
Tim Howard
@tghoward
Jan 10 2016 00:59
I'll dig it up
Jason Calabrese
@jasoncalabrese
Jan 10 2016 01:01
@ktomy there a cull tool that uses the last OpenAPS from NS and filters out everything before that
Also on the NS dev as of last night we do upsets for treatments so dupes only waste bandwidth
Ben West
@bewest
Jan 10 2016 01:02
hmm, I need to update
for some reason, the cull thing stopped working for me on my laptop
did a fresh install basically
think the arguments changed or something
Alin Artiom Kenibasov
@ktomy
Jan 10 2016 01:02
aha, so i'll update NS so no debug will be needed :)
thank you @jasoncalabrese
Ben West
@bewest
Jan 10 2016 01:05
@tghoward needs to be on 12/23/15 in the csv
Tim Howard
@tghoward
Jan 10 2016 01:06
@bewest my csv doesn't have that date. I'll see if I can upload to the carelink site and get a fresh csv. Might be a while
Ben West
@bewest
Jan 10 2016 01:06
maybe that date is in older files though
oh 31
hmm
page 32 looks ok
hah, and it starts with Sara6E
Jason Calabrese
@jasoncalabrese
Jan 10 2016 01:07
@ktomy you still want to look into it depending on how much your pulling from the pump, it could be a lot to send to NS each loop
Tim Howard
@tghoward
Jan 10 2016 01:07
True - i could perhaps go back a bunch more pages and try to find the first occurrence
Ben West
@bewest
Jan 10 2016 01:07
the problem occurs before that though
I think there's only 32 pages?
Tim Howard
@tghoward
Jan 10 2016 01:07
Can I get more with mm-ReadHistory?
Alin Artiom Kenibasov
@ktomy
Jan 10 2016 01:08
@jasoncalabrese yes, I know, but it decreased this task priority :)
Ben West
@bewest
Jan 10 2016 01:08
only max pages available
max pages is 32 I think
openaps use pump read_current_history_pages
oh, 36 for me!
that might cover it
get those pages :-D
your logs are filled with temp basals :-)
Jason Calabrese
@jasoncalabrese
Jan 10 2016 01:10
I had lots of problems with dupes when going in and out of network
Ben West
@bewest
Jan 10 2016 01:10
500 carbs + ?
it's fun to watch it go down using the dedupe tool though
Tim Howard
@tghoward
Jan 10 2016 01:13
Hmm, guessing this means I've reached the end of my available pages
### PUMP MODEL: `ReadPumpModel:size[64]:data:'723'`
<class 'decocare.commands.ReadHistoryData'> {'page': 36}
CRITICAL:decocare.session:this seems like a problem
response: ReadHistoryData:size[1024]:[page][36]:data[2]:
hexdump:
JaysonEwer
@JaysonEwer
Jan 10 2016 01:13
@bewest YEP, everything seems to be working properly now after fixing the permissions on the new repository.
Ben West
@bewest
Jan 10 2016 01:13
0x0d 0x00?
nice
JaysonEwer
@JaysonEwer
Jan 10 2016 01:14
I proceeded to working through some of the example reports and aliases over at https://github.com/openaps/docs/blob/master/docs/Log-clean-analyze-with-openaps-tools/using.md
Ben West
@bewest
Jan 10 2016 01:14
you can get history with openaps use pump iter_pump_history 4 to get last 4 hours of history
JaysonEwer
@JaysonEwer
Jan 10 2016 01:14
yep
Ben West
@bewest
Jan 10 2016 01:14
er, iter_pump_hours maybe
do you have a dexcom share?
JaysonEwer
@JaysonEwer
Jan 10 2016 01:15
I do, but i've been using my old dexcom G4 platinum receiver with my Pi, is it recommended to use the share?
I was going to setup a Bitbucket account to backup my stuff.
Ben West
@bewest
Jan 10 2016 01:16
@tghoward pretty sure Ian50 is to blame
Tim Howard
@tghoward
Jan 10 2016 01:16
:smile:
Ben West
@bewest
Jan 10 2016 01:17
I see two places where it definitely goes off the rails after Ian50
page 15 has problem as well
Tim Howard
@tghoward
Jan 10 2016 01:23
put the last few pages up on git. page 34 and 35 also have some funny records
But they don't go back enough in time to match up with my csv
Ben West
@bewest
Jan 10 2016 01:25
ok
problem goes away if body_length = 30 on Ian50
Tim Howard
@tghoward
Jan 10 2016 01:26
Would it help to get a csv for the correct time (for page 31)?
Ben West
@bewest
Jan 10 2016 01:27
yeah
several records in here, would be nice to fix names
Tim Howard
@tghoward
Jan 10 2016 01:27
Ok. It might take me some time, but I'll work on it.
sounds good.
Ben West
@bewest
Jan 10 2016 01:27
definitely has something to do with these sensor alarms
btw, you can try searching for glucose history to test against as well
the engineers at medtronic included a nifty search by date feature :-)
which apparently I did not yet include in the openaps driver
oh, no I did
filter_glucose_date
will return the pages holding glucose data for your data
Ben West
@bewest
Jan 10 2016 01:42
try bewest/decoding-carelink#166 @tghoward :-)
Alin Artiom Kenibasov
@ktomy
Jan 10 2016 01:47
:) please don't send me to documentation in your first phrase :) My question is: I have a basal (from profile) of 0.2/h, bg is in target range. dia is set to 2h, and iob is calculated to 0 (shouldn't it be something like 0.025?) and when I "ask" for suggested, it returns me 0.05u/h for 30m, shouldn't it return me my "default" basal which I have now?
Do I do something wrong or do I understand wrongly the logic of how it works?
Scott Leibrand
@scottleibrand
Jan 10 2016 01:50
What is the avgdelta?
What are the dates in your pumphistory.json? 2000 or 2016?
Alin Artiom Kenibasov
@ktomy
Jan 10 2016 01:53
in pumphistory i don't have a string "2000-"
Scott Leibrand
@scottleibrand
Jan 10 2016 01:54
k good. You have the latest decocare with the y2k16 bug fixed then.
JaysonEwer
@JaysonEwer
Jan 10 2016 01:54
@bewest I've got my Backup all tucked away over at Bitbucket. All two reports. :-D I've still got a lot of work to do, but i'm making progress!!
Alin Artiom Kenibasov
@ktomy
Jan 10 2016 01:56
last delta was about -2, i can't see it in clear text, just a second will search more
Scott Leibrand
@scottleibrand
Jan 10 2016 01:57
Can you paste the determine-basal output?
as detailed in that new doc, it shows all the inputs and reasons for what it did.
Alin Artiom Kenibasov
@ktomy
Jan 10 2016 01:57
{"delta":-4,"glucose":117,"avgdelta":-3.6666666666666665}
{"duration":0,"rate":0,"temp":"absolute"}
{"iob":0,"activity":0,"bolusiob":0}
{"max_iob":30,"type":"current","dia":2,"current_basal":0.2,"max_daily_basal":0.45,"max_basal":1.5,"min_bg":120,"max_bg":130,"sens":250}
-4 :) it was only in output, not stored :)
Scott Leibrand
@scottleibrand
Jan 10 2016 01:58
And the suggested.json? (The stored file)?
Alin Artiom Kenibasov
@ktomy
Jan 10 2016 01:58
{"temp":"absolute","bg":117,"tick":-4,"eventualBG":106,"snoozeBG":106,"reason":"Eventual BG 106<120, no temp, setting 0.048U/hr","duration":30,"rate":0.05}
Scott Leibrand
@scottleibrand
Jan 10 2016 01:59
Ok, so that is exactly like the example in my doc. Avgdelta is driving deviation.
If you switch to dev it had more details on the output like in my doc that weren't in the master version.
*has
Alin Artiom Kenibasov
@ktomy
Jan 10 2016 02:00
ok, so let me read the documentation and if i'll have questions I'll post them, as I understand now, it was trying to prevent low
Scott Leibrand
@scottleibrand
Jan 10 2016 02:07
Yes. There is also a question of whether your IOB is correct. Check that both pumphistory and clock.json are showing correct times.
Tim Howard
@tghoward
Jan 10 2016 02:23
@bewest record 24 is "JournalEntryPumpLowReservoir"
bewest/decoding-carelink#156
Added a table bracketing that time here: bewest/decoding-carelink#166
huh. Ian50 is "BolusNormal"
no, it's not, Ian50 is "ChangeSensorSetupConfig2" !!
Dana Lewis
@danamlewis
Jan 10 2016 02:27
Heh - this (http://www.bestprogramminglanguagefor.me/q) suggests programming languages to learn based on what you want to do. Went down a number of paths in it with openAPS mindset (since several ppl ask what should they learn, and I was curious if this would be something different than what we recommend based on existing code base). Python kept being recommended.
Tim Howard
@tghoward
Jan 10 2016 02:27
that's a weird one I had thought might be the funny records on the earlier dataset
@danamlewis , Interesting. Bash seems to have been most important for me :smile:
But I guess bash isn't a programming language.
Scott Leibrand
@scottleibrand
Jan 10 2016 02:30
It kinda is.
Ben West
@bewest
Jan 10 2016 02:30
@tghoward it was earlier, earlier, until I discovered the engraving I posted on the issue was missing --model 723 flag, and I added it
edited it
@tghoward you'r confirming it's ChangeSensorSetupConfig2?
it does look like the "changetime" command because there are two in a row
are there two or one in the csvs?
Tim Howard
@tghoward
Jan 10 2016 02:36
@bewest the csvs have this pattern:
ChangeSensorSetupConfig2
ChangeSensorSetup2
ChangeSensorSetupConfig2
Jason Calabrese
@jasoncalabrese
Jan 10 2016 02:36
went to Fry's to get a cable, walked out with an edison, so now I don't have to wait for the one I ordered
figure I'll need 2 anyway
Tim Howard
@tghoward
Jan 10 2016 02:36
with lots of raw data attached to the ChangeSensorSetupConfig2
bewest/decoding-carelink#166
Rachel Sandlain
@audiefile
Jan 10 2016 02:41
Did it just jump in the basket @jasoncalabrese ? That happens to me a lot in microcenter
Scott Leibrand
@scottleibrand
Jan 10 2016 02:42
Lol
Ben West
@bewest
Jan 10 2016 02:43
what's after that?
next pump record (is CalBGForPH at 11:42:39?)
Tim Howard
@tghoward
Jan 10 2016 02:44
12/23/15 11:42:39 BGCapturedOnPump
12/23/15 11:42:47 BolusNormal
12/23/15 11:42:47 BolusWizardBolusEstimate
12/23/15 11:42:47 UnabsorbedInsulin
12/23/15 11:42:47 UnabsorbedInsulin
Ben West
@bewest
Jan 10 2016 02:44
ah ok
Tim Howard
@tghoward
Jan 10 2016 02:45
I see now that Sarah6E has the date error
Ben West
@bewest
Jan 10 2016 02:49
mm-decode-history-page.py --model 723 logs/2016-01-09/ReadHistoryData-page-*.data | grep -A 50 -B 20 Ian50 | less can be interesting
allows seeing everything around all Ian50's
with my patch I don't see any errors
are you getting cgm pages as well?
I think `list_cgm.py is only available in the repo
:-(
Tim Howard
@tghoward
Jan 10 2016 02:56
Not sure what you are asking. If I get all the Ian50 as described above, I do get a ValueError: month must be in 1..12 at the very end
well ... as the last thing, the error probably stopped mm-decode-history-page.py from running
Ben West
@bewest
Jan 10 2016 02:58
hmm
in the decocare repo
git remote -v
or
Tim Howard
@tghoward
Jan 10 2016 03:00
pi@betacb ~/src/decoding-carelink $ git remote -v
origin    git://github.com/bewest/decoding-carelink.git (fetch)
origin    git://github.com/bewest/decoding-carelink.git (push)
Ben West
@bewest
Jan 10 2016 03:01
ah
do git fetch
er
git fetch origin
Tim Howard
@tghoward
Jan 10 2016 03:01
which branch
Ben West
@bewest
Jan 10 2016 03:02
git checkout tghoward/wip-mystery-data
Tim Howard
@tghoward
Jan 10 2016 03:03
nice. got it
Ben West
@bewest
Jan 10 2016 03:03
ah, great
Tim Howard
@tghoward
Jan 10 2016 03:03
need to npm install or something?
Ben West
@bewest
Jan 10 2016 03:03
don't think so
try the decoding thing again
you might need sudo python setup.py develop but probably not
Tim Howard
@tghoward
Jan 10 2016 03:04
wow. That ran through with no error!
Ben West
@bewest
Jan 10 2016 03:06
:-)
I think problem is solved, not sure
Tim Howard
@tghoward
Jan 10 2016 03:06
Pushing my luck and checking for Sarah6E :smile:
Alin Artiom Kenibasov
@ktomy
Jan 10 2016 03:07
as it is most critical part and I cannot find an example, could you please tell me if this string is correct?
openaps report add enact/enacted.json JSON pump set_temp_basal enact/suggested.json
Scott Leibrand
@scottleibrand
Jan 10 2016 03:09
Ya, that is identical to what I have.
Alin Artiom Kenibasov
@ktomy
Jan 10 2016 03:09
thank you
Scott Leibrand
@scottleibrand
Jan 10 2016 03:09
Be sure to test it manually etc. etc.
Alin Artiom Kenibasov
@ktomy
Jan 10 2016 03:09
but please, don't put this string in documentation.
OF COURSE!!!
Scott Leibrand
@scottleibrand
Jan 10 2016 03:09
exactly our thinking.
Alin Artiom Kenibasov
@ktomy
Jan 10 2016 03:10
also from the samples in documentation, I would suggest splitting actual invoking of enact and suggest in terms of aliases
Tim Howard
@tghoward
Jan 10 2016 03:11
@bewest This is fabulous. So this patch is active? Do I need to do anything to make sure it applies to iter_pump ?
Scott Leibrand
@scottleibrand
Jan 10 2016 03:12
@ktomy Would love to discuss better ways to structure it. Maybe best to submit a PR with an idea for how to change it.
Alin Artiom Kenibasov
@ktomy
Jan 10 2016 03:13
yes, i'll try to helpo you with documentation imediately after I will finish the manual loop and will see that it works, but I will not try to do anything before completely understanding the system
Scott Leibrand
@scottleibrand
Jan 10 2016 03:14
Good plan.
Ben West
@bewest
Jan 10 2016 03:19
@tghoward it should just work
Tim Howard
@tghoward
Jan 10 2016 03:24
Thank you so much for spending this time on this. Very much appreciated. I'm thrilled. And also off to bed! :smile:
Thomas Emge
@ThomasEmge
Jan 10 2016 03:40
@ktomy @scottleibrand I have discovered the very same line ;) but I get a bunch of errors. Does that mean there is nothing to adjust - as the reason in suggested.json indicates?
Scott Leibrand
@scottleibrand
Jan 10 2016 03:41
Right. Your alias should grep for duration or rate and only run the enact report if it's present.
Ben West
@bewest
Jan 10 2016 03:43
it's better to use json tool
btw, anyone have a share receiver?
good way to cut the cords :-)
Jason Calabrese
@jasoncalabrese
Jan 10 2016 03:44
looking forward to that
any idea what it will take to get it to work on the edison?
Scott Leibrand
@scottleibrand
Jan 10 2016 03:45
Yeah, true. Might be useful to document how to use json to check suggested.json.
Jason Calabrese
@jasoncalabrese
Jan 10 2016 03:45
it won't just install and work right?
Ben West
@bewest
Jan 10 2016 03:46
yeah, I think it will, actually
Jason Calabrese
@jasoncalabrese
Jan 10 2016 03:46
that would be awesome
Ben West
@bewest
Jan 10 2016 03:47
I use this alias check-recommendation-and-enact = ! bash -c "json -f oref0-predict/oref0.json && test -n \"$(json -f oref0-predict/oref0.json duration rate | tr -d ' ')\" && (echo \"OK\" && openaps enact ) || echo \"No recommendation to send\""
depends on bluez version
it needs bluez 5 of some flavor
Jason Calabrese
@jasoncalabrese
Jan 10 2016 03:47
going to try getting connected to a RL first I think
but it that gets stuck, this is next
Thomas Emge
@ThomasEmge
Jan 10 2016 03:47
I see, I got sidetracked by what appeared in the terminal - as that contained a "duration" element
Jason Calabrese
@jasoncalabrese
Jan 10 2016 03:48
just need to get the kids dinner and put them to bed
Scott Leibrand
@scottleibrand
Jan 10 2016 03:48
Ah, yeah. That's the currently running temp.
Alin Artiom Kenibasov
@ktomy
Jan 10 2016 03:57
so first tests passed, it detected need of temp, enacted it, then detected lowering bg and stopped temp
more iterations and tests to be done
Ben West
@bewest
Jan 10 2016 04:07
the check-recommendation-and-enact alias above check the duration and rate, and only does the temp basal if both are set
if it's missing it just prints a message saying no recommendation
Scott Leibrand
@scottleibrand
Jan 10 2016 04:49
:+1: @ktomy
Dragan Maksimovic
@dm61
Jan 10 2016 05:18
I have made an attempt to explain enacting better in the using-oref0-tools section, and have opened a PR to dev of the docs. Here is my version of that section: https://github.com/dm61/docs/blob/dev/docs/Build-manual-system/Using-oref0-tools.md
Scott Leibrand
@scottleibrand
Jan 10 2016 05:18
@dm61 excellent job on that PR. does my comment make sense?
Dragan Maksimovic
@dm61
Jan 10 2016 05:22
makes sense; I'll modify that subsection
Scott Leibrand
@scottleibrand
Jan 10 2016 05:22
thx
Alin Artiom Kenibasov
@ktomy
Jan 10 2016 05:28
this is exactly what I was talking about earlier. Is it ok to put enact in documentation so anyone can just copy-paste it without understanding what it actually does?
Scott Leibrand
@scottleibrand
Jan 10 2016 05:29
see my comment on the PR. :)
Dana Lewis
@danamlewis
Jan 10 2016 05:29
No. We do not want something copy-paste. See @scottleibrand Comment in PR
Scott Leibrand
@scottleibrand
Jan 10 2016 05:29
jinx. :)
Dana Lewis
@danamlewis
Jan 10 2016 05:31
@ktomy This is also why the PR process is good, so we can continue to have discussions about what is good to include, etc. (And sometimes things evolve, but probably not this particular one.) These docs have come a looooooong way in the past 6 months :)
Alin Artiom Kenibasov
@ktomy
Jan 10 2016 05:33
yes, I know, documentation now is a loooot better than several months ago, actually now it exists :) and this documentation makes the difference and makes people to dare
Dana Lewis
@danamlewis
Jan 10 2016 05:33
LOL.
Scott Leibrand
@scottleibrand
Jan 10 2016 05:33
heh. existing is good. :)
Dana Lewis
@danamlewis
Jan 10 2016 05:34
@channemann And @bewest Get major props for making it exist :)
Alin Artiom Kenibasov
@ktomy
Jan 10 2016 05:36
I remember the time when @bewest posted his mm-* scripts , it was almost impossible to do anything, kilos of debug output, raw dumps, exceptions everywhere... :)
now I am not scared of running mm-something anymore :) At least all debug info is hidden
Alin Artiom Kenibasov
@ktomy
Jan 10 2016 05:45
good night, it's enough for today for me, it is 7:43AM here :) Tonight will be the first night of unter-test manual loopung. Thank you all again!
Dana Lewis
@danamlewis
Jan 10 2016 05:45
Have a good one!
Jason Calabrese
@jasoncalabrese
Jan 10 2016 07:59
used the cc-debugger and I think it worked, now to get the edison going
Jason Calabrese
@jasoncalabrese
Jan 10 2016 08:19
waiting 2 minutes for edison to boot, and since I've been warned I'm going wait even longer
Ali Mazaheri
@amazaheri
Jan 10 2016 17:52
looks like Share backend has some issues, anyone else problem connecting?
Dana Lewis
@danamlewis
Jan 10 2016 17:53
Yep servers are down
Super glad we plug receiver into Pi regardless for failsafe
Ali Mazaheri
@amazaheri
Jan 10 2016 17:54
nice
Dana Lewis
@danamlewis
Jan 10 2016 17:56
Hope everyone else continues to plug into their setup for backup offline mode if needed, too.
Ali Mazaheri
@amazaheri
Jan 10 2016 17:58
:+1:
Jason Calabrese
@jasoncalabrese
Jan 10 2016 18:00
Don't see how anyone could use a service like that, no redundancy
Completely at their mercy
Would be good if you could point to their EU service or something at times like that
Ali Mazaheri
@amazaheri
Jan 10 2016 18:45
I miss our android uploader 😊 old schoolz
Dana Lewis
@danamlewis
Jan 10 2016 18:46
Ya. Now that we are home, pulling out our 3rd receiver to start testing @bewest openxshareble
Ben West
@bewest
Jan 10 2016 18:49
cut those cords :-)
Ali Mazaheri
@amazaheri
Jan 10 2016 18:49
lol
started our G5 receiver
till now we were using only phone and nightscout to feed Jarvis
plan B, part of our DR planning
@bewest was blaming Kylo before confirming Dex backend is down lol
Ben West
@bewest
Jan 10 2016 18:54
Kylo?
Ali Mazaheri
@amazaheri
Jan 10 2016 18:57
yeah look at my twitter feed
Scott Leibrand
@scottleibrand
Jan 10 2016 19:10
pi@rpi3 ~/openaps-dev $ openaps vendor add openxshareble
No module named openxshareble
openxshareble doesn't seem to be an importable python module
Scott Leibrand
@scottleibrand
Jan 10 2016 19:16
oh, duh. dependency install instructions are in the README.
Eric
@ecc1
Jan 10 2016 19:16
I'm trying to communicate with my G4 receiver over BLE (using some code in Go). What exactly causes the G4 receiver to stop blinking its BT icon? Is it just the authentication step (writing serial# to the appropriate characteristic)? Because the API I'm using doesn't have an explicit pair() function, and what I've read of BLE made me think pairing wasn't used. But the Java code (xdrip) and Python code (openxshareble) both call some pair() function, so I'm confused (as usual).
Scott Leibrand
@scottleibrand
Jan 10 2016 19:19
pi@rpi3 ~/openaps-dev $ openaps use share list_dexcom
Traceback (most recent call last):
...
File "/home/pi/src/Adafruit_Python_BluefruitLE/Adafruit_BluefruitLE/platform.py", line 38, in get_provider
from .bluez_dbus.provider import BluezProvider
File "/home/pi/src/Adafruit_Python_BluefruitLE/Adafruit_BluefruitLE/bluez_dbus/provider.py", line 29, in <module>
import dbus
ImportError: No module named dbus
Ben West
@bewest
Jan 10 2016 19:23
hmm
Scott Leibrand
@scottleibrand
Jan 10 2016 19:24
ah, yeah, let me try those
Ben West
@bewest
Jan 10 2016 19:24
sudo apt-get -y install libusb-dev libdbus-1-dev libglib2.0-dev libudev-dev libical-dev libreadline-dev
should be able to see DEXCOMRX in sudo hcitool lescan
@ecc1 I don't think you should need to call Pair
when I'm using openxshareble with openaps, I see the BT icon on the receiver only when openaps is reading data from the receiver
rest of time, it's absent
Ben West
@bewest
Jan 10 2016 19:30
yeah, oin my toolchain, I got errors if I did not Pair, so I connect/Pair, if not paired, do the auth step, do everything else, and then disconnect
Scott Leibrand
@scottleibrand
Jan 10 2016 19:39
compiling bluez is slow. we should package that if it's not already.
assuming we actually need the latest version. 5.23 or thereabouts is available as a package.
Eric
@ecc1
Jan 10 2016 19:40
@bewest Also, are you using some kind of UART emulation over BLE on the receiver (just like its USB-serial interface)? Trying to understand the use of the various characteristics it exports.
Scott Leibrand
@scottleibrand
Jan 10 2016 19:52
@bewest even after installing all prereqs I'm still getting ImportError: No module named dbus
Eric
@ecc1
Jan 10 2016 19:56
I'd try installing python-dbus and maybe python-dbus-dev if needed
Scott Leibrand
@scottleibrand
Jan 10 2016 19:58
that worked for that error, now:
File "/home/pi/src/Adafruit_Python_BluefruitLE/Adafruit_BluefruitLE/bluez_dbus/provider.py", line 31, in <module>
from gi.repository import GObject
ImportError: No module named gi.repository
sudo apt-get install python-gifixed that
Alex
@SuicidalTissue
Jan 10 2016 20:04
Hi, i'm about to start running an closed loop and I'm just getting to the software part. With oref0, will i need to create a script to execute commands or does the system auto-run?
Scott Leibrand
@scottleibrand
Jan 10 2016 20:05
once you have a "loop" alias that does everything, you'll need to schedule it in cron
read through all the docs: it walks you through it all
Eric
@ecc1
Jan 10 2016 20:06
@scottleibrand python equivalent of whack-a-mole
Scott Leibrand
@scottleibrand
Jan 10 2016 20:06
heh ya
Alex
@SuicidalTissue
Jan 10 2016 20:06
Brill, hopefully will be looping within a week!
Scott Leibrand
@scottleibrand
Jan 10 2016 20:07
:+1:
so now the Pi's BLE all seems to be working, but I can't seem to get it to pair with the receiver
Scott Leibrand
@scottleibrand
Jan 10 2016 20:12
the receiver just sits in pairing mode (blinking BT icon)
I configured the receiver's serial number...
the end of the list-dexcom output looks like:
None
paired? 0
pairing...
paired
[UUID('00001800-0000-1000-8000-00805f9b34fb'), UUID('00001801-0000-1000-8000-00805f9b34fb'), UUID('00001804-0000-1000-8000-00805f9b34fb'), UUID('0000180a-0000-1000-8000-00805f9b34fb'), UUID('f0aca0b1-ebfa-f96f-28da-076c35a521db')]
finding service
SERVICE None
Searching for UART devices...
[
  {
    "mac": "FF:F3:8F:02:29:46",
    "name": "DEXCOMRX",
    "advertised": [
      "00001800-0000-1000-8000-00805f9b34fb",
      "00001801-0000-1000-8000-00805f9b34fb",
      "00001804-0000-1000-8000-00805f9b34fb",
      "0000180a-0000-1000-8000-00805f9b34fb",
      "f0aca0b1-ebfa-f96f-28da-076c35a521db"
    ]
  }
]
if I try to run any commands to retrieve data, I get AttributeError: 'NoneType' object has no attribute 'ReadRecords'
Scott Leibrand
@scottleibrand
Jan 10 2016 20:18
@bewest how do I make the Pi pair with the Share?
Scott Leibrand
@scottleibrand
Jan 10 2016 20:26
tried removing and readding the share device: no change
Eric
@ecc1
Jan 10 2016 20:28
Do you have the receiver serial number set somewhere? That's how it authenticates.
Scott Leibrand
@scottleibrand
Jan 10 2016 20:32
yes, I did that with configure
Scott Leibrand
@scottleibrand
Jan 10 2016 20:56
another thing I can't figure out: how to do bluetooth tethering from the Pi. I can pair it with my iPhone, but all the instructions I can find online for actually using the iPhone's bluetooth hotspot are written for older versions of bluez and don't work on newer versions (and the directions didn't seem too solid even on those older versions, fwiw)
does anyone know how we might be able to get that working?
Ben West
@bewest
Jan 10 2016 21:24
it should pair itself when you use iter_... after using configure to set the serial
how many 'share's are there in your environment?
I'm not sure how it deals with multiple devices
fwiw, bluez 5.x is first versions with working ble
sudo hciconfig hci0 up
two dexcom in my environment....
not sure what will happen, not looking good
Ben West
@bewest
Jan 10 2016 21:30
hmm
time to hit up #bluez channel in irc
Scott Leibrand
@scottleibrand
Jan 10 2016 21:32
Ok, I'll try again with only one nearby.
Ben West
@bewest
Jan 10 2016 21:39
mine pairs fine, actually
I think I ran list_dexcom a few times, eventually, it only responds with one, even though lescan shows two
then iter_foo started working
and it looks like my data ;-)
should probably work on the debug out put of that
@ecc1 yes, the characteristics correspond to in/out endpoints like you might see on a UART
hmmm
Ben West
@bewest
Jan 10 2016 21:44
the protocol is the same as the usb uart, except all writes are prefixed with 0x01 0x01
Alin Artiom Kenibasov
@ktomy
Jan 10 2016 22:30
@bewest openaps/oref0#53 can you please confirm that I am not wrong? Thank you.
Alin Artiom Kenibasov
@ktomy
Jan 10 2016 22:58
@danamlewis in your new page when you explain how determine-basal works, could you please elaborate more on "fourth line" of logged inputs? For me at least it is not very clear. i.e. for bolussnooze:0.1 , does it mean that in 5 minutes bg should be changed by a previous bolus by ISF * 0.1 ?
Scott Leibrand
@scottleibrand
Jan 10 2016 23:09
Bolussnooze is just "how much of the IOB should we ignore for purposes of bolus snooze"
what you're describing is insulin activity.
Kevin Wells
@wkcwells
Jan 10 2016 23:57
Hi all. Working on uploading from openaps to Nightscout per https://github.com/openaps/docs/blob/master/docs/Automate-system/vizualization.md.
The upload report errors out with 'No such file or directory’. I believe that I have setup all the required precursor reports properly, but haven’t found explicit documentation on this. All the referenced files are there.
Any guidance would be much appreciated.