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

18th
Oct 2016
Hannah Crabtree
@lollydaggle
Oct 18 2016 00:17
That alcohol trick for the "Button Error" was a lifesaver. Looks like I'm back
Dana Lewis
@danamlewis
Oct 18 2016 00:33
woot!
Tim Mecklem
@tmecklem
Oct 18 2016 00:40
Where should I go to ask a decocare cgm PagedData record parsing code question?
It’s pretty specific and probably has a straightforward answer. :)
Paul Martin
@Spazholio
Oct 18 2016 00:43
https://gitter.im/nightscout/decoding-dexcom ? I have no idea, I’m just guessing.
Pete Schwamb
@ps2
Oct 18 2016 00:43
Here is as good as any place. :)
Tim Mecklem
@tmecklem
Oct 18 2016 00:44
Cool, I choose @ps2’s answer :) (sorry @Spazholio)
Pete Schwamb
@ps2
Oct 18 2016 00:44
Ben is the guy you want, but he’s taking a well deserved break. We can try to fill in.
Paul Martin
@Spazholio
Oct 18 2016 00:44
I am so staggeringly offended, words can’t possibly express….
Tim Mecklem
@tmecklem
Oct 18 2016 00:44
Good for Ben. Everyone needs a break.
lol @Spazholio
Dana Lewis
@danamlewis
Oct 18 2016 00:45
@Spazholio make a new unicorn, it’ll make you feel better ;)
Tim Mecklem
@tmecklem
Oct 18 2016 00:45
ok, here goes: On https://github.com/openaps/decocare/blob/1713f27ecf951b0c04c8245cdb0e3cb2ceb09f76/decocare/cgm/__init__.py#L133 it’s checking for op > 0, which means that if the code encounters a 0x00 byte in parsing the middle of a page, it turns it into a glucose record with an sgv of 0.
Is that intentional? (If so, is that correct?)
Paul Martin
@Spazholio
Oct 18 2016 00:46
I can make unicorns?! Are you sure? I don’t think I even know which muscle to flex...
Tim Mecklem
@tmecklem
Oct 18 2016 00:48
Pete Schwamb
@ps2
Oct 18 2016 00:51
@tmecklem Does the eat_nulls method get rid of the zeros before that method is used?
If I was you and having questions like that, I’d probably wire up the python code to run through my data with some prints in it.
Tim Mecklem
@tmecklem
Oct 18 2016 00:52
It gets rid of the nulls at the beginning (really the end of the page, but it’s parsed backwards).
It doesn’t handle them in the middle of the page.
Pete Schwamb
@ps2
Oct 18 2016 00:52
Guessing there shouldn’t be 0’s elsewhere (outside of records)?
If it’s like history, the first byte is an event type opcode, and should never be 00; you just see zeros after the last record.
Tim Mecklem
@tmecklem
Oct 18 2016 00:53
My page of cgm data has ‘em. That’s what was keeping the results of the ported Swift code to match the Python code, since the Swift code was ignoring them in the middle of the page.
Pete Schwamb
@ps2
Oct 18 2016 00:54
Weird, maybe check the record length of the preceding record? Maybe it should be slurping up the 0?
Tim Mecklem
@tmecklem
Oct 18 2016 00:54
wait, now I’m second guessing that last statement.
I changed the page parsing Swift code to match the Python and the events matched up, but I don’t see any 0x00 bytes around the place where the mismatch begins.
Okay, they’re there. A few 0x00s in a row after one of the other single byte opcodes which happens to be a DataEnd record.
It seems there are a few of those 0x00 bytes after several of the DataEnd records. I wonder if that just means the pump overwrote the page there and put a few nulls after the DataEnd for good measure? I have some more reading to do regarding what to do with the DataEnd records I think.
Tim Mecklem
@tmecklem
Oct 18 2016 01:00
Oh, interesting! Everywhere a DataEnd record occurs (0x01), it’s always followed by 0x0000. I wonder if the DataEnd record should really be considered a 3 byte record including the op code.
Right now it looks like those extra bytes are getting parsed into two individual glucose records with sgv of 0. That doesn’t seem right.
Output of decocare for reference
  86   {
  87     "name": "GlucoseSensorData",•
  88     "date_type": "prevTimestamp",•
  89     "_tell": 968,•
  90     "sgv": 0,•
  91     "date": "2016-02-06T14:35:00",•
  92     "packet_size": 0,•
  93     "op": 0
  94   },•
  95   {
  96     "name": "GlucoseSensorData",•
  97     "date_type": "prevTimestamp",•
  98     "_tell": 967,•
  99     "sgv": 0,•
 100     "date": "2016-02-06T14:40:00",•
 101     "packet_size": 0,•
 102     "op": 0
 103   },•
 104   {
 105     "packet_size": 0,•
 106     "name": "DataEnd",•
 107     "date": "2016-02-06T14:45:00",•
 108     "date_type": "none",•
 109     "_tell": 966,•
 110     "op": "0x01"
 111   }
I’m not sure what to do with this.
Tim Mecklem
@tmecklem
Oct 18 2016 01:06
I don’t think those are supposed to be sgv records. I think eating them along with the other 0x00 bytes is the correct thing to do, but I’m not 100% sure.
Scott Leibrand
@scottleibrand
Oct 18 2016 01:06
can you test whether changing the DataEnd parsing code to expect 0x010000 breaks anything else?
Tim Mecklem
@tmecklem
Oct 18 2016 01:09
I just realized that approach won’t work because the opcode would be at the end of the record instead of the beginning, the way the page is parsed.
It would look like 0x000001 instead of 0x010000 to the parser.
So I don’t think treating DataEnd like a three byte record as easily workable as applying the eat_nulls approach within the page.
I could do some more tests on it and see if the output of various sample page data in the project changes by anything other than the absence of the 0 value sgv records.
Scott Leibrand
@scottleibrand
Oct 18 2016 01:12
:+1:
Tim Mecklem
@tmecklem
Oct 18 2016 01:12
I’m assuming that oref0 discards 0 value sgvs, right? (I think I remember reading that in the code)
Scott Leibrand
@scottleibrand
Oct 18 2016 01:14
i believe the check is to consider < 30 to be errors. not sure it will simply discard them in all cases, though.
Tim Mecklem
@tmecklem
Oct 18 2016 01:17
Okay. I’ll run some page decodes and check in with the results. Should I open an issue now to track the conversation or wait until I have more?
Paul Martin
@Spazholio
Oct 18 2016 01:17
I got unofficially diagnosed with gastroparesis today. I swear, I need a “delayed bolus” option on my pump. It would be incredibly helpful.
Dana Lewis
@danamlewis
Oct 18 2016 01:17
are you running AMA, @Spazholio ?
Darrell Wright
@beached
Oct 18 2016 01:18
@Spazholio wouldn't not bolusing and letting the close loop handle that help?
Paul Martin
@Spazholio
Oct 18 2016 01:18
@danamlewis 99% sure, yes. I mean, I was, and I chose that option yesterday when I re-set up the loop.
Scott Leibrand
@scottleibrand
Oct 18 2016 01:19
if you make your meal bolus smaller (but enter all the carbs) AMA will start high-temping as it sees digestion actually picking up
Paul Martin
@Spazholio
Oct 18 2016 01:19
@beached …somewhat. I already do a form of that since I eat LCHF. But not for if I want to have, say, a handful of pretzels.
So if I were to eat 50g worth of carbs (10u), you’re suggesting (hypothetically, of course) taking 5u directly with the pump, but then entering 50g worth of carbs into NS?
(corrected my question to actually include carbs)
Scott Leibrand
@scottleibrand
Oct 18 2016 01:21
or putting 50g into the bolus wizard and dialing down the bolus to 5U
Paul Martin
@Spazholio
Oct 18 2016 01:22
Silly question - you mean “bolus wizard” on the pump, and not BWP in NS? Or…?
Dana Lewis
@danamlewis
Oct 18 2016 01:22
correct. pump
Paul Martin
@Spazholio
Oct 18 2016 01:22
I don’t think I’ve ever used BWP in NS. Looks neat though. =)
Dana Lewis
@danamlewis
Oct 18 2016 01:22
but if you’re like me and don’t like that, the other method you suggested (bolus , and put carbs into NS somehow) also works
Paul Martin
@Spazholio
Oct 18 2016 01:23
So OpenAPS/NS is smart enough to say, “Hey, this guy ate 50g of carbs, but only gave himself 5u of insulin! I’ll make up the difference!” or something like that?
David
@davidkeddydb
Oct 18 2016 01:23
@danamlewis @scottleibrand Let's say you eat pizza, can you handle that the same way but increase the max basal rate allowed to handle for the bgs creap in the 4-5 hours after?
Paul Martin
@Spazholio
Oct 18 2016 01:23
I’ve always been a little iffy on how that aspect of it works, but since I don’t really eat a lot of carbs, it’s never been an issue.
Dana Lewis
@danamlewis
Oct 18 2016 01:23
yes BUT will do it in concert with actually observing deviations
so if you have no absorption because (insert reason), it’ll wait till you start rising
Scott Leibrand
@scottleibrand
Oct 18 2016 01:23
@davidkeddydb yes. :)
Paul Martin
@Spazholio
Oct 18 2016 01:24
Interesting.
Dana Lewis
@danamlewis
Oct 18 2016 01:24
@davidkeddydb you could, but I don’t increase my max basal like that on a temporary basis. but yes that max basal should be somewhat higher than normal basal to give you extra wiggle room for after meals, but don’t go crazy crazy with it because there’s safety implications of allowing the loop to go all-in on insulin.
David
@davidkeddydb
Oct 18 2016 01:26
@danamlewis I hear you. I almost have the whole meal thing worked out. I have had to increase sensitivity, change basals, it is all getting there. I was never at 70%+ in 70-170 prior so it is all coming together
Dana Lewis
@danamlewis
Oct 18 2016 01:26
:+1:
Darrell Wright
@beached
Oct 18 2016 01:27
It would be interesting to see the long term tracking of close loop with regard to long term complications. Those starting young and into adulthood when the onset often starts
Then again, I hope it becomes irrelivant
Dana Lewis
@danamlewis
Oct 18 2016 01:28
Yea, I expect studies for closed loop starting closer to diagnosis, to preserve insulin production. And studies to show reduction in standard deviation in addition to a1c helping. And studies of all the things ;) but probably a few years out from that, once closed looping is commerically available and becomes more common
:+1:
Terri Lyman
@tatigger
Oct 18 2016 01:29
I have some questions about the way the system is calculating basals, I set the range from 130 to 150, and the OpenAPS pill reads "BG:161,COB: 8,Dev:36,BGI: -6.95, ISF:65, Target, 140, 145-145 in range, no temp required, temp 0.45~req 0.45u/hr IOB:0.79u, Basal IOB -0.12" Why does the system say target 140, 145-145 in range? Earlier today it had the whole range when comparing....so like target 140, 130-150 in range
David
@davidkeddydb
Oct 18 2016 01:29
Anyone using IFTT to cancel temp target? I don't have that right. I have the set eating soon which is cool but my cancel isn't working
Dana Lewis
@danamlewis
Oct 18 2016 01:30
@tatigger 140 is the midpoint
Darrell Wright
@beached
Oct 18 2016 01:30
I just ordered a new pump instead of waiting. the veo should still be available in Canada and I am pretty sure the new 670g equiv isn''t what I want. At least not until they are able to allow a lower target as the data comes in
Dana Lewis
@danamlewis
Oct 18 2016 01:30
it’s predicting you’ll be 145, which is still in range
David
@davidkeddydb
Oct 18 2016 01:30
sorry, I skipped a T
Scott Leibrand
@scottleibrand
Oct 18 2016 01:31
@tatigger IIRC the 145-145 means both eventualBG and snoozeBG were 145. it's not telling you the target range there, it's telling you the eventualBG-snoozeBG range
Terri Lyman
@tatigger
Oct 18 2016 01:32
Oh, ok. Thanks : )
I knew you would know
Dana Lewis
@danamlewis
Oct 18 2016 01:32
so you’ll see those numbers change over time
you may decide to narrow your range a bit, if you get annoyed with it not recommending actions. i.e. 130-140, or 140-150.
Guess who has two thumbs, and finally started working on coding up the infamous oref0 issue openaps/oref0#99 ;)
Terri Lyman
@tatigger
Oct 18 2016 01:35
Yes, I was, and yes, I will : ) hahaha - I am usually much lower during the day, but I was being cautious, as my husband is out of town for 2 weeks. Not a good time for my other pump to crap out. But I like what I have been seeing - It keeps me at the top of that range - the very top. And moved me there quite easily this morning from a lower bg. You can't really tell that from observing the loop running all by itself, until you see how you actually respond. At least I can't.
Dana Lewis
@danamlewis
Oct 18 2016 01:35
(the “issue” is where we’ve been logging notes and ideas about it as a feature, similar to how we sketched out advanced meal assist and auto-sens, as a hint for anyone who ever wants background on a big feature like that, looking through the closed “issues” you’ll often find good details)
@tatigger no judging, ever, on what you set your limits! Just wanted to point that out :) as a prediction for a future question :)
Terri Lyman
@tatigger
Oct 18 2016 01:46
LOL, @danamlewis! You are definitely one of my heroes
Paul Martin
@Spazholio
Oct 18 2016 01:47
@danamlewis Jumping back to my hypothetical scenario earlier - if I were to eat 50g, put it in the Bolus Wizard, but only give myself a 0.1u bolus (instead of 5u), would the system know that when it detects absorbtion/a rise in BG, it needs to give a larger temp basal?
Because I just ate 20g of cookies, and I’m itching to test out a new way to play with this. =)
Dana Lewis
@danamlewis
Oct 18 2016 01:54
So personally, I still like to do at least part of a meal bolus. But you might be interested in the unannounced meal testing approach: https://docs.google.com/document/d/1Y5a1X16-7BFhQFP5EYXejRKKZ9HhD0zBXBd1aPT5E34/edit?usp=sharing
(yes, if it knows about the carbs it’ll AMA to help give the amount of insulin needed, varying based on carb absorption/actual changes in BG, capped by your max basal, etc.)
Jason Calabrese
@jasoncalabrese
Oct 18 2016 01:59
it will also wait for a rise, and probably be too slow
Paul Martin
@Spazholio
Oct 18 2016 02:01
See, that’s the part I was concerned about.
Dana Lewis
@danamlewis
Oct 18 2016 02:02
see the note in unannounced testing. some people “too slow” and “too high” (i.e. 150) ; others would be happy peaking after a meal only at 180, etc.
so YDMV and it’s always worth testing & figuring out what works for you
Scott Leibrand
@scottleibrand
Oct 18 2016 02:04
also see the notes in the unannounced meal testing doc about doing eating soon before the normal scheduled meal time
Paul Martin
@Spazholio
Oct 18 2016 02:04
I pretty much mentally append that last sentence to everything said here. =)
Dana Lewis
@danamlewis
Oct 18 2016 02:04
what, YDMV, or eating soon?
because I append eating soon recommendations to every conversation ;)
Paul Martin
@Spazholio
Oct 18 2016 02:05
@scottleibrand That doesn’t do me any good if I get hungry and want a couple cookies while watching TV.
Scott Leibrand
@scottleibrand
Oct 18 2016 02:05
if I don't beat you to it. ;-)
Paul Martin
@Spazholio
Oct 18 2016 02:05
@danamlewis The YDMV bit. Because nothing is truer than that. =)
Scott Leibrand
@scottleibrand
Oct 18 2016 02:05
@Spazholio yeah. for those it's a question of how big the spike is gonna be without insulin, and whether a temp basal would be able to bring it down fast enough
Dana Lewis
@danamlewis
Oct 18 2016 02:05
:+1:
Paul Martin
@Spazholio
Oct 18 2016 02:05
@scottleibrand I use the hell out of the temp targets anyway. Works really well so far.
Dana Lewis
@danamlewis
Oct 18 2016 02:06
:+1:
Paul Martin
@Spazholio
Oct 18 2016 02:06
Oh, I had a question about the square/dual-wave bolus issue. The system still handles that….sub-optimally, right?
Dana Lewis
@danamlewis
Oct 18 2016 02:07
you can’t set a temp if a dual or square is running
Paul Martin
@Spazholio
Oct 18 2016 02:08
Hmmm. So I suppose it doesn’t really matter how it handles it since it can’t do anything about it anyway. Ok, that more or less answers my question. =)
Tim Mecklem
@tmecklem
Oct 18 2016 02:44
I ran all of the glucose pages through before and after eating nulls in the middle of pages. Now I’m trying to find a diff tool that A) doesn’t crash, and B) displays the differences intelligently :)
Scott Leibrand
@scottleibrand
Oct 18 2016 02:45
git diff --word-diff=color ?
Once you IFTTT it's hard to stop!
Paul Martin
@Spazholio
Oct 18 2016 02:50
…well that’s a helluva thing. I have GOT to get into IFTTT, I think.
Dana Lewis
@danamlewis
Oct 18 2016 02:51
Like eating soon mode ;), it is addictive
Paul Martin
@Spazholio
Oct 18 2016 02:52
*sigh* So…are there directions on setting this stuff up? =)
Dana Lewis
@danamlewis
Oct 18 2016 02:53
Yep. @logichammer wrote them up yesterday :) I meant to approve his PR but got distracted with 99.
Paul Martin
@Spazholio
Oct 18 2016 02:54
Mind dropping a note in here when you approve it? I’d love to read it. =)
Dana Lewis
@danamlewis
Oct 18 2016 02:55
Go check out the open PRs to OpenAPS docs, should be the top open one :)
That way you don't have to wait for me :)
Tim Mecklem
@tmecklem
Oct 18 2016 02:56
Results of the analysis: the only thing modified on any of the parsed cgm pages when the code is modified to eat nulls is the absence of the double sgv: 0 entries and it adjusts the adjacent DataEnd record’s timestamp back by 10 minutes since there aren’t two empty sgv entries in front of it when relative timestamping is applied.
The rest of the records are left untouched.
Scott Leibrand
@scottleibrand
Oct 18 2016 02:56
:+1:
Anyone else here have MDT CGM and could help test @tmecklem's fix?
Tim Mecklem
@tmecklem
Oct 18 2016 02:58
Let me spin up a branch and push it to github.
Perhaps of interest, not all of the pages had those two null bytes after the DataEnd op byte. It might be pump model specific, but it doesn’t affect the page decoding on the pages that don’t have the double null bytes.
Tim Mecklem
@tmecklem
Oct 18 2016 03:34
Okay, PR submitted: openaps/decocare#11 (and another unrelated one at openaps/decocare#10)
I’m off to bed. Thanks for helping me understand the code better tonight!
Dana Lewis
@danamlewis
Oct 18 2016 03:38
Nice work @tmecklem !
Tim Mecklem
@tmecklem
Oct 18 2016 03:41
:thumbsup: glad to be able to (hopefully) contribute back to the project. The work you all have done has enabled my wife and another good friend to loop. We’re all very grateful.
Dana Lewis
@danamlewis
Oct 18 2016 03:43
:heart: us too :)
That's meant to be a heart as a statement, and an "us too" as in we are very grateful to the community, too :)
John Males
@johnmales
Oct 18 2016 05:16
What is the MDT CGM issue that @tmecklem is looking at?
Dana Lewis
@danamlewis
Oct 18 2016 05:18
openaps/decocare#11
@johnmales ^
John Males
@johnmales
Oct 18 2016 05:20
Thanks. Happy to test, hopping on a plane back to Sydney and will look at it in a day or so. I wonder whether this may be related to the no updates to cgm values issue as well. Thanks @danamlewis @tmecklem :)
diabeticgonewild
@diabeticgonewild
Oct 18 2016 05:32
@Spazholio I have severe gastroparesis but they are pretty sure that mine is not related to T1D as my autonomic neuropathy started around the time I was diagnosed with T1D. I have a rare disease that caused it called autoimmune autonomic ganglionopathy.
Marijuana (if it's not legal in the state you're in get Marinol) plus cutting out carbs helps. I was diagnosed with severe gastroparesis at age 18.
I have a gastric pacemaker and the battery was dying so I had it replaced last Monday.
I pretty much can eat whatever I want now.
But I have severe diarrhea and gut pain/cramping to go along with it. That's much worse than my gastroparesis at this point of time.
I wouldn't take the diagnosis too seriously. Just find a GI doctor that specializes in motility disorders...
diabeticgonewild
@diabeticgonewild
Oct 18 2016 05:37
I'm actually vaporizing marijuana right now (vaping saves herb...). My vaporizer finally charged up...
But I definitely don't feel well. Hence herb...
diabeticgonewild
@diabeticgonewild
Oct 18 2016 05:43
Thanks intravenous immunoglobulin
And I am blazed now.
And watching Harry Potter.
Tim Street
@tim2000s
Oct 18 2016 05:44
@scottleibrand @danamlewis Dumb question I know, but determine-basal.js is the only component of oref0 that determines tbr and takes inputs to do so? The rest of the OpenAPS toolkit is about providing the right data for determine-basal to use?
(Sorry, digging in to Loop and comparing the LoopDataManager and DoseMaths with determine-basal)
Scott Leibrand
@scottleibrand
Oct 18 2016 05:50
@tim2000s that's correct
diabeticgonewild
@diabeticgonewild
Oct 18 2016 05:52
I have to go asleep. This stuff is powerful.
David Young
@wgs4
Oct 18 2016 05:58
i get this error message trying to rerun the oref0 setup script
[master (root-commit) bc5ce09] initializing openaps 0.1.5
 1 file changed, 0 insertions(+), 0 deletions(-)
 create mode 100644 openaps.ini
/home/pi/myopenaps initialized
/home/pi/src/oref0/ already exists; pulling latest
fatal: unable to connect to github.com:
github.com[0: 192.30.253.112]: errno=No route to host

Couldn't pull latest oref0
pi@rpi ~ $
hmm maybe i dont have internet connection on the Pi right now.. cant seem to ping google from it
Scott Leibrand
@scottleibrand
Oct 18 2016 06:30
yeah, looks like that's the problem. you might check ifconfig, and try a sudo dhclient if you don't have an IPv4 address
Hristina Dimova
@xpucuto
Oct 18 2016 07:38
I remeber someone was looking for pumps with motor errors so he can test with them. I saw one in the Pay-it-Forward group but can not remeber the person. Here is a link
William Fallon
@inform880
Oct 18 2016 12:09
Does anyone have a fix for the '?AD' alarms? We have them turned off, but it's notifying my website anyway. I dove into the javascript plugin code but I'm not completely sure what needs to change.
fallon225
@fallon225
Oct 18 2016 12:27
2016-10-18 11.28.10.png
Is that still currently this issue? nightscout/cgm-remote-monitor#1402
Paul Martin
@Spazholio
Oct 18 2016 12:55
@diabeticgonewild I had a question about the whole gastroparesis thing whenever you're around.
David
@davidkeddydb
Oct 18 2016 12:55
Screenshot 2016-10-18 08.54.04.png
My BG target on my 722 at 12am-7am is 105-105. Why am I getting a target of 124 here?
Chris Oattes
@cjo20
Oct 18 2016 13:00
not looked at the screenshot, do you have AMA enabled?
David
@davidkeddydb
Oct 18 2016 13:01
Yes, I do
Chris Oattes
@cjo20
Oct 18 2016 13:01
if so, when it detects increased sensitivity, it ups your target
David
@davidkeddydb
Oct 18 2016 13:02
Thank you. That is exactly what is happening as the ISF is 86 or something and normally it is set at 70.
Chris Oattes
@cjo20
Oct 18 2016 13:03
:)
David
@davidkeddydb
Oct 18 2016 13:12
@cjo20 If it changes the target bg based on a higher or lower ISF, do you ever reach your intended bgs? I have only been on for really 3 weeks full time with it and I am still working out ISF, basals and eating soon but I have it just about locked down.
Chris Oattes
@cjo20
Oct 18 2016 13:12
seems to work :)
David
@davidkeddydb
Oct 18 2016 13:13
:smile: That works for me.
Chris Oattes
@cjo20
Oct 18 2016 13:14
I can't remember the exact reasoning behind it. I think it's that it doesn't adjust your basal rates along with the ISF. Changing the target up is equivalent to lowering your basal rate.
David
@davidkeddydb
Oct 18 2016 13:17
I understand. I am just wondering in my case, not that it is huge, but how I don't end up 19 higher than my intended bgs. Will know here shortly.
Chris Oattes
@cjo20
Oct 18 2016 13:17
So just changing your ISF doesn't change how much basal you're getting. If you're more sensitive to insulin, you're also more sensitive to your basal insulin
so if it doesn't change the target, it'll keep giving the same basal rate as if you have normal sensitivity. But you're more sensitive, so it'll drop you. So instead, it raises the target, which means OpenAPS says "oh, you're below the target, i'll reduce insulin to get you up there". The amount it reduces it by is enough to match your increase in sensitivity
David
@davidkeddydb
Oct 18 2016 13:21
Well changing your ISF does in that on the calc to target it would be a higher calculated temp basal or lower based on the ISF right?
Chris Oattes
@cjo20
Oct 18 2016 13:24
ISF affects corrections, not the basal rate itself
So when calculating "I need to get from X to Y", it uses ISF
but changing the ISF doesn't change the underlying basal rate
Paul Martin
@Spazholio
Oct 18 2016 13:24
Wait, ISF doesn't affect basals?
At all?
Chris Oattes
@cjo20
Oct 18 2016 13:25
so you set a 1u/hr basal rate on your pump with an ISF of 30
Then you change your ISF to 25. What change does taht make to your basal rate?
David
@davidkeddydb
Oct 18 2016 13:26
Underlying yes, but ISF does affect the temp basal that gets thrown in the enact, right?
Chris Oattes
@cjo20
Oct 18 2016 13:26
yes, ISF is used in calculating temp basals
Paul Martin
@Spazholio
Oct 18 2016 13:26
@davidkeddydb is asking what I was wondering about.
Ahhhh, ok. That makes more sense.
Chris Oattes
@cjo20
Oct 18 2016 13:26
but if you don't change the target BG, OpenAPS doesn't know it needs to a temp basal when it detects you are more sensitive
David
@davidkeddydb
Oct 18 2016 13:27
@cjo20 Right. So if the ISF gets jacked up from your normal, say 70 up to 101, and your target is 90 shouldn't it just change the insulin amount suggested for the temp basal vs changing the target?
Chris Oattes
@cjo20
Oct 18 2016 13:28
davidkeddydb: if your BG is 90 and your target is 90, what temp basal does it set?
David
@davidkeddydb
Oct 18 2016 13:28
@cjo20 it would set whatever my default is for that time assuming no iob
Chris Oattes
@cjo20
Oct 18 2016 13:29
but you know you're more sensitive than usual, so what will happen if it gives you your default basal?
David
@davidkeddydb
Oct 18 2016 13:29
you go lower than target
Chris Oattes
@cjo20
Oct 18 2016 13:30
OpenAPS bases decisions on where it thinks you are going to be relative to the target. The assumption is that the default basal will keep you flat
There are two ways you can deal with the drop caused by default insulin dropping you when you're more sensitive
either you specifically tell it to reduce the basal amount, or you raise the target by the amount it will drop you
Raising the target will cause it to give you less insulin. Because of the way the code works, all you have to do is multiply the target by a ratio in one place, and then it just works
Dealing with manually lowering the basal rates will involve (possibly lots) more code
David
@davidkeddydb
Oct 18 2016 13:34
so I get that the temp basal suggested has to go lower but why does the target ever have to change? The target should stay at 90, ISF should change based on AMA if higher ISF greater reduction in basal if lower ISF then lesser. I am sure it is a coding on the multiple targets, just trying to wrap my head around it.
Chris Oattes
@cjo20
Oct 18 2016 13:36
If you do it that way, the system has to care whether an ISF is modified by AMA or not
because ISF isn't fixed for a given basal rate. It's just as valid to have an ISF of 30 or 25 or 20 with the same basal rate, it depends on the person
Trying to think of a decent analogy
Chris Oattes
@cjo20
Oct 18 2016 13:42
If you know your oven is hotter than it should be, say you set it to 180C and it ends up at 200C. There are two ways you can fix it, one is to fix the thermostat so that it reads the correct value again, which involves ordering parts and taking the oven apart. Alternatively, you can just set the oven to 160C and it'll be 180C, which is where you wanted to be, and it took less work to get
there
The "fix the termostat" option is equivalent to adjusting default basal rates, the "turn it down a bit" is equivalent to changing the target BG
@davidkeddydb does that make any more sense?
David
@davidkeddydb
Oct 18 2016 13:46
@cjo20 It does.
Chris Oattes
@cjo20
Oct 18 2016 13:54
This is the point where Scott wakes up and tells me I'm wrong :P
David
@davidkeddydb
Oct 18 2016 14:06
@cjo20 When using ama does your set ISF on your pump ever matter, or does it only matter when you are off-line? (I don't think Scott sleeps)
Chris Oattes
@cjo20
Oct 18 2016 14:06
yes, it matters. AMA adjusts the ISF you set
David
@davidkeddydb
Oct 18 2016 14:09
So are most taking the cues that AMA is giving and adjusting ISF based on that?
Chris Oattes
@cjo20
Oct 18 2016 14:09
AMA adjusts what you have set to try and account for changes in sensitivity
David
@davidkeddydb
Oct 18 2016 14:11
@cjo20 I have always managed pretty hands-on so letting Openaps do the work is still a little bit of a challenge
Chris Oattes
@cjo20
Oct 18 2016 14:12
I watched it like a hawk for the first few weeks
David
@davidkeddydb
Oct 18 2016 14:14
My edison rig flakes out ever so often with communication with the pump/TI stick more so than the RPI so my mobility with it hasn't been 100%. Looking forward to new explorer board.
Matt Pressnall
@logichammer
Oct 18 2016 14:32
Nice, @danamlewis , I like who you are sending the messages to... :)
albertotrg
@albertotrg
Oct 18 2016 15:21
Still looking for an old Medtronic pump for openAPS... no luck on Medwow (Western Europe) neither on my community (Spain).
Any idea about where to try? Thanks
Hardwick1973
@Hardwick1973
Oct 18 2016 15:44
@albertotrg I secured a pump in the UK using http://www.medwow.com/ - people proved to be a little unresponsive but I got there in the end. I wouldn't pay much attention to the prices; most people were open to an offer. It did take me about 4 weeks though! Best of luck
AdrianLxM
@AdrianLxM
Oct 18 2016 15:53

Hi, @ceben80 and @C-Ville asked for a line in cron to power down the edison at a certain battery level. I'm not familiar with the edison loop, but this should do:

awk '/batteryVoltage/{if($2<=3200)system("sudo shutdown -h now")}' ~/myopenaps/monitor/edison-battery.json

or more in line with other commands:

cat ~/myopenaps/monitor/edison-battery.json | grep batteryVoltage | awk '{if ($2<=3200)system("sudo shutdown -h now")}'
Matthias
@ceben80
Oct 18 2016 15:55
@AdrianLxM works perfect :tada: - thanks Adrian, awesome :clap:
I am inserting this into my cron
but will define 3200 as shutdown level regarding to this in voltage.c
#define BAT_FULL        4180
#define BAT_NORMAL      3680
#define BAT_LOW         3400
#define BAT_CRIT        3250
#define BAT_DEAD        2950
* * * * * awk '/batteryVoltage/ {if ($2<=3200)system("sudo shutdown -h now")}' ~/myopenaps/monitor/edison-battery.json
Chris Oattes
@cjo20
Oct 18 2016 15:58
You're fine with that, as long as the edison-battery.json updates again before that cron job runs
I'd force edison-battery.json to be refreshed immediately before running the shutdown
Paul Martin
@Spazholio
Oct 18 2016 16:00
@danamlewis Remember last night how we were talking about entering the full number of carbs into the BW, but then adjusting the amount of insulin given downward, and AMA would pick up the slack? What if I wanted to put all of the carbs into NS somehow, but give ZERO insulin? I'm about to eat a protein-heavy meal, and instead of my usual temp target trick, I was wondering if there's a way to give no insulin in advance, but stil put my rig "on alert" that I may have carb absorption happening in the near future?
Hardwick1973
@Hardwick1973
Oct 18 2016 16:01
Picture1.png
I am looking for some help tuning my rig following several successful nights of PLGM looping. I have increased maxIOB to 1.2, set the target range on my Medtronic 522 pump to 140-150 and max basal to 1.0. The chart shows the algorithm trying to compensate for a high BG but it never quite did enough to get down to the target range. I'm monitoring things closely so happy to increase these levels but the chart suggests that the algorithm is operating within limits set and yet not driving levels down below 180s
Matthias
@ceben80
Oct 18 2016 16:04
@cjo20 that's a good idea :+1: thx
* * * * * cd /home/edison/myopenaps && openaps report invoke monitor/edison-battery.json && awk '/batteryVoltage/ {if ($2<=3200)system("sudo shutdown -h now")}' ~/myopenaps/monitor/edison-battery.json
Scott Leibrand
@scottleibrand
Oct 18 2016 16:10
@Spazholio the two ways to alert oref0 to upcoming meals are with eating soon mode and/or then (pre-)entering carbs.
If you do those, then yes, it will be more aggressive and respond to any positive deviation accordingly.
Paul Martin
@Spazholio
Oct 18 2016 16:10
By "pre-entering carbs" do you mean doing a "Carb Correction" Event Type? If not, how does one do that?
Scott Leibrand
@scottleibrand
Oct 18 2016 16:12
Probably. "Enter carbs into NS" somehow. We don't differentiate based on event type. If it makes the little carb indicator on NS, and you're configured to download ns-meal-carbs in your loop, then it'll pull them in.
Paul Martin
@Spazholio
Oct 18 2016 16:14
Ok, cool. And the latest setup script takes advantage of ns-meal-carbs? I can't check right now, but I JUST re-set up the loop with all the bells and whistles activated.
Scott Leibrand
@scottleibrand
Oct 18 2016 16:15
@cjo20 @davidkeddydb as you suspected, your speculation about autosens does require some correction. ofef0 does adjust your basals based on the autosens ratio. In fact, that is probably the most important thing it does most of the time (except right after a meal). The way we also adjust the target based on autosens ratio is an additional adjustment to make sure you actually get close to your real target when running sensitive or resistant.
@Spazholio yeah, it is part of oref0-setup.
One thing that autosens target adjustment compensates for is that the autosens basal adjustments are only used for setting temps, not for calculating IOB. That would be hard due to feedback loops, so instead we calculate IOB based on the regularly scheduled basals, and make an additional adjustment to the target. Also, you can disable that if you want, either by setting a manual temp target, or in preferences.json.
albertotrg
@albertotrg
Oct 18 2016 17:44
@Hardwick1973 thanks for the suggestion! I'll keep trying on http://www.medwow.com/ but for the moment no one even responded me
Dana Lewis
@danamlewis
Oct 18 2016 17:44
@albertotrg many people in the US use Craigslist; not sure if it works in Europe. But also perhaps posting an “ask” in local community groups might help?
albertotrg
@albertotrg
Oct 18 2016 17:57
Thanks @danamlewis. Craiglist is not used here. I asked in my local community but nobody keeps old pumps...we got pumps from our National Health System which upgrades them eventually (collecting the old ones). I'll keep trying on Medwow!
Dana Lewis
@danamlewis
Oct 18 2016 17:58
:+1: or ask on Facebook, someone may be willing to ship you one :) from elsewhere
Paul Martin
@Spazholio
Oct 18 2016 18:01
You could try over in https://www.reddit.com/r/diabetes as well. I got one from there previously.
David
@davidkeddydb
Oct 18 2016 18:10
@scottleibrand So if the ISF from autosens is always much higher or lower then the set pump ISF, does it make sense to go and adjust ISF on the pump itself, or not bother? I am still tweaking to get my settings right. I am close but I bottom out from time to time post prandial as my timing with pre-bolus and eating soon is still a work in progress...but I am no longer going to 230 mg/dl which is good.
Scott Leibrand
@scottleibrand
Oct 18 2016 18:12
@davidkeddydb don't automatically adjust just ISF. think about whether you're running high/low after meals or when fasting. Having mistuned basals will actually affect autosens more than ISF, so adjust basals first, or adjust them proportionally together the same way autosens does.
David
@davidkeddydb
Oct 18 2016 18:13
:+1:
Eric
@ecc1
Oct 18 2016 18:49
The dev branch of https://github.com/ecc1/decocare has a fix for the 6.4+ U/hr temp basal issue. PR pending for openaps/decocare.
Dana Lewis
@danamlewis
Oct 18 2016 18:50
fantastic! thanks @ecc1
Matthias
@ceben80
Oct 18 2016 18:58
Yeah that's great @ecc1 what exactly needs to be done for updating :question:
Scott Leibrand
@scottleibrand
Oct 18 2016 19:11
see the PR: looks like this is a partial fix that's not yet safe to loop with (if your max basal is > 6.4U/hr)
you can of course test it while @ecc1 works on adding the decoding fix as well
you'd need to checkout his branch and install it, as we do for the dev branch here: https://github.com/openaps/docs/blob/master/scripts/quick-src.sh#L11-L15
Sandra Keßler
@SandraK82
Oct 18 2016 19:23
@ecc1 @scottleibrand i was confused as the decoding was strange.. had only a few looks last week and was not able to resolve the error
Eric
@ecc1
Oct 18 2016 19:26
Just committed the decoding fix too
Sorry for overlooking the decoding at first
I was lulled into complacency by the fact that the read_temp_basal command worked ...
Scott Leibrand
@scottleibrand
Oct 18 2016 19:28
thx. @SandraK82 it should be ready for you to test now
Sandra Keßler
@SandraK82
Oct 18 2016 19:29
Ah nice one, that's was my error... @ecc1 did not consider that it was highest bit only...
Matthias
@ceben80
Oct 18 2016 19:31
Thanks @scottleibrand
Scott Leibrand
@scottleibrand
Oct 18 2016 19:31
openaps/decocare#12 is the link to the PR to comment on after you've tested the fix
Eric
@ecc1
Oct 18 2016 19:31
@SandraK82 Yes, other bits in that byte are used to distinguish percentage vs. absolute temp basal type.
Sandra Keßler
@SandraK82
Oct 18 2016 19:33
Thank you very much @ecc1
Diabetic_Dude
@diabetic_dude_twitter
Oct 18 2016 19:41
hello, great to be here again. just wanted to get some input, would this be a preferred rasberry pi 3 starter kit ? if not, could you please send me a link to one on amazon that is??? https://goo.gl/9bAKKA
Scott Leibrand
@scottleibrand
Oct 18 2016 19:47
that one would work fine. I got the Pi 2 version of that back before the 3 was released. you won't need most of the extra wires or LEDs, but it's still cheaper than buying the components individually
Darrell Wright
@beached
Oct 18 2016 19:48
@scottleibrand there is some really neat stuff on the RPI front that gets linux out of the stack and lets you make a single program device. The TI USB ACM rfspy looks supported too, along with networking.
Scott Leibrand
@scottleibrand
Oct 18 2016 19:51
that seems like a lot of work. what is the benefit?
Darrell Wright
@beached
Oct 18 2016 19:52
not sure it is. You still have the high level libraries. So connect to service and upload data, talk to pump over usb acm connection
but loose the setup. it becomes a 4M image
let variables
less
Scott Leibrand
@scottleibrand
Oct 18 2016 19:54
if you can get that working without completely rewriting everything for the new platform, that'd be great. most single-purpose boards want you to rewrite things in their language though
Darrell Wright
@beached
Oct 18 2016 19:54
i've seen C++ for one, but not sure about parts. There's Windows IOT and then there is a pascal dev env that looks like it has a lot
Scott Leibrand
@scottleibrand
Oct 18 2016 19:55
all of our stuff is in javascript and python. I've never seen a board that supports both without a full OS
Darrell Wright
@beached
Oct 18 2016 19:55
I would prefer a reference design in C++ as it works on every device out there these days.
yeah. I've seen micropython on some esp8266's but it's not full python
Diabetic_Dude
@diabetic_dude_twitter
Oct 18 2016 19:57
@jlucasvt I am interested in your set up
I need most of those pieces anyways, and worse case, I could just change out the pi2 for a pi3
Sandra Keßler
@SandraK82
Oct 18 2016 20:00
@beached i'm planning to do something like this, without the need for rewriting every patch through the usage of transformation
scottleibrand @scottleibrand looks forward to seeing that. :)
Darrell Wright
@beached
Oct 18 2016 20:02
Like if everyone used a common language like c++ for the common libraries(interacting with rileylink, mmcommander..., uploading to NS, the glucose models) it would mean a lot of the work of porting to new phones could be reduced
or go/rust. something compilable on everything
im just thiking client side, not NS side.
@SandraK82 what do you mean by transformation? not sure I understand
Scott Leibrand
@scottleibrand
Oct 18 2016 20:06
@beached yeah, the challenge is getting everything ported the first time, and making sure it all works. most of our testing was done by people like me who haven't touched C++ since high school.
Sandra Keßler
@SandraK82
Oct 18 2016 20:07
I work in software engineering and developed a system with my team to parse source languages to extended uml notation and generators for different target languages (transformations). Each target languages/platforms specific toolkits can be used this way and very good code can be generated
Scott Leibrand
@scottleibrand
Oct 18 2016 20:08
nice! is that tool available open source or commercially?
Sandra Keßler
@SandraK82
Oct 18 2016 20:10
It's currently total closed source as this generates our income currently. We are discussing options how we can provide this toolkit to this community without providing to much insight how the internals work
Darrell Wright
@beached
Oct 18 2016 20:10

@scottleibrand there are compilers for d/go/rust on like every platform now too but yeah I understand.

@SandraK82 ah I thought you might have meant that but wasn't sure. that is neat. How do you get around the regression to the least common denominator in the languages. Like Java lacks operator overloading so you are stuck with methods like add( lhs, rhs ) or is that accounted for too and in languages like c++ would output operator+( lhs, rhs )... Or C# using interface IDisposable and languages with deterministic destruction?

Scott Leibrand
@scottleibrand
Oct 18 2016 20:11
@SandraK82 is your tool generally available on a commercial basis yet?
Sandra Keßler
@SandraK82
Oct 18 2016 20:13
@scottleibrand only internally used on consulting basis for client projects as the specific parsers and generators have to be customized for each project
Scott Leibrand
@scottleibrand
Oct 18 2016 20:14
If you could turn that into a more generalized software solution that didn't require consulting, that could be huge
but that may be an impossibly hard problem, I don't know
Sandra Keßler
@SandraK82
Oct 18 2016 20:15
@beached this is solved through uml notations. There are generic concepts for everything you can do in software... therefore you just need to get the model right and generation will be easy
Scott Leibrand
@scottleibrand
Oct 18 2016 20:16
I wonder if those kind of transformations might be useful as a form of static analysis to identify bits of code that might not behave as intended...
Darrell Wright
@beached
Oct 18 2016 20:17
That's true. I guess once you have an AST you have choices
Scott Leibrand
@scottleibrand
Oct 18 2016 20:18
I could also imagine some IDE integrations that allow you to write code in any language (even using different syntax on different lines if you want) and then converts it into a generic representation on the back end, and presents it to you in whatever language you prefer for reading and editing...
Darrell Wright
@beached
Oct 18 2016 20:18
@scottleibrand I find that funny with javascript. reasoning with that language where type punning happens all the time is interesting.
Sandra Keßler
@SandraK82
Oct 18 2016 20:19
@scottleibrand Most probably it would make a huge impact, however I wasted nearly 8 years with my business partner in trying to figure out the general solution... there are special concepts that are exclusive to each other
Scott Leibrand
@scottleibrand
Oct 18 2016 20:20
@SandraK82 I wonder if doing it at the time the code is being written might be a more tractable problem than trying to solve the general problem of transforming arbitrary existing code reliably
Sandra Keßler
@SandraK82
Oct 18 2016 20:21
Dynamic typing is no real problem
Scott Leibrand
@scottleibrand
Oct 18 2016 20:21
the IDE plugin could give feedback as you write / checkin on things that are ambiguous, and suggest improvements...
Darrell Wright
@beached
Oct 18 2016 20:23
@scottleibrand outputting to javascript is done a lot too eh. So I can program in C++ and the output platform can be javascript. Adobe did this with lightroom I believe. But in your case wouldn'tt typescript help a lot? I believe resharper does some of this too and migrated js to typescript
Scott Leibrand
@scottleibrand
Oct 18 2016 20:23
it might be really powerful to be able to write in a high-level language, and have it automatically translated on the fly into something lower level and higher performance without all the manual work
@beached I am not familiar enough with the existing solutions. I don't even use an IDE
Sandra Keßler
@SandraK82
Oct 18 2016 20:24
The way our system handles it is via parse exception if a concept is encoutered that does not match any input rule. If it gets parsed it will be abstracted and regenerated with better matching concepts
Scott Leibrand
@scottleibrand
Oct 18 2016 20:25
I'm an ops guy, but I do watch how much time our engineers spend fighting with the peculiarities of languages they have to work in for a particular project. being able to write in any language you need without really having to learn the language could be a huge productivity boost
Sandra Keßler
@SandraK82
Oct 18 2016 20:28
We currently have a long term project with backends in c, java, js, perl including tests and databases in sqlite, files, oracle and it runs on embedded, micro, mobile, desktop and server environments
Jeremy M Lucas (GetRileyLink.org)
@jlucasvt
Oct 18 2016 20:29
@diabetic_dude pm me your address and I'll send the parts to you.
Diabetic_Dude
@diabetic_dude_twitter
Oct 18 2016 20:32
@jlucasvt trying to figure out how to pm now
sorry
Scott Leibrand
@scottleibrand
Oct 18 2016 20:33
on the web interface, click the profile picture and click Chat Privately
I don't know if you can initiate a PM from the mobile app, but once you've started one you can continue it
Jeremy M Lucas (GetRileyLink.org)
@jlucasvt
Oct 18 2016 20:33
Email me is ok too. GetRileyLink@gmail
Diabetic_Dude
@diabetic_dude_twitter
Oct 18 2016 20:33
I sent it
Jeremy M Lucas (GetRileyLink.org)
@jlucasvt
Oct 18 2016 20:34
Cool got it. I'll package it up tonight
Diabetic_Dude
@diabetic_dude_twitter
Oct 18 2016 20:40
@jlucasvt thanks can you check the PM for me one more time I had a question for you
Darrell Wright
@beached
Oct 18 2016 20:43
@SandraK82 cross platform is interesting enough, but cross platform and environment is a fun twist eh. Like even something like proper temp file creation and removal isn't standard. Its rudimentary but can be a problem if done unsafely.
Sandra Keßler
@SandraK82
Oct 18 2016 20:46
Yes indeed. If you can concentrate on those issues it's much easier. Most of the software is "painted" and tested in java ( my fastest language for coding )
Darrell Wright
@beached
Oct 18 2016 21:22
Do you find antlr helps with the parser/generator code?
Gustavo
@bustavo
Oct 18 2016 22:01
@tmecklem on the MDT CGM issues… I also use MDT and have had this problem since forever… I’ve been able to “””fix it””” by calibrating at least 1 hour before the calibration alarms… if I do it very close to the calibration request when flipping page I stop getting data… I need to calibrate again for data to start appearing again… I updated decocare yesterday and it seemed to solve the problem… if you want me to help test your changes I can help just point me to the repo
Paul Martin
@Spazholio
Oct 18 2016 22:16
Does anyone know the specific method one needs to upload @cjo20’s battery monitor status to NS? Does it need additional info other than the JSON that’s output? Which endpoint needs to be used?
Jason Calabrese
@jasoncalabrese
Oct 18 2016 22:53
there's an extra --uploader option for ns-status that takes the json from the tool
I was planning on clean this up some, but this is what I'm doing https://github.com/jasoncalabrese/indy-e1i/blob/master/openaps.ini#L58-L69
Darrell Wright
@beached
Oct 18 2016 22:59
Looking at oref0 determine-basaljs and curious. Is rt.mealAssist a string that encapsulates a structure?
Jason Calabrese
@jasoncalabrese
Oct 18 2016 23:05
it was just for display/debugging
Darrell Wright
@beached
Oct 18 2016 23:08
so the return setTempBasal is where I should look?
and the logic of setting rate, profile, currenttemp...
Paul Martin
@Spazholio
Oct 18 2016 23:26
Wait, is there a way to enable Meal Assist to show up in a NS pill?
Tim Mecklem
@tmecklem
Oct 18 2016 23:36
@bustavo what problem are you seeing? Do you get sgv 0 values and that breaks you somehow?
Gustavo
@bustavo
Oct 18 2016 23:37
@tmecklem yes, when flipping to a new page there is no data present on new page… I need to recalibrate whenever that happens and then after 3-5 more readings the data starts appearing again on the latest page...
@tmecklem I’ve been able to make it not happen by calibrating earlier than required by the pump alarms ( about 1 hour before the alarms )
@tmecklem if I calibrate after the calibration reminder alarm goes off then no data gets on newest page… I am on a 722 pump
@tmecklem I also had that problem when reading data from a 754 pump
Scott Leibrand
@scottleibrand
Oct 18 2016 23:39
@bustavo do you have reason to think that the new page issue is the same as the sgv 0 issue?
Gustavo
@bustavo
Oct 18 2016 23:42
@scottleibrand I’ve played around with the records hash on decoding-carelink/decocare/cgm/init.py and have seen the sgv 0 … thought it could be related?
@scottleibrand maybe its not… yesterday I had the new page issue and upgraded to latest decocare … then the data was being extracted with a few missing data points
Gustavo
@bustavo
Oct 18 2016 23:59
@tmecklem as I remember.. the last time I changed the records I also started getting the 10 minutes off problem.. I left things as they were before my changes as I did not want to tamper with the dates on blood glucose values… but your assumption of the two missing records makes sense… Yesterday when I upgraded decocare and the data started appearing again, I could notice that there were 2 records missing between the old page and the new one...