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

17th
Sep 2015
John Males
@johnmales
Sep 17 2015 00:09
Thanks esz
I meant thanks @eszcloud :)
Ben West
@bewest
Sep 17 2015 00:12
openaps's temp basal and bolus commands take a json file as input
so it's the name/path to the json file
the json file should have the bolus/basal record structure in it, it's the same from the output from the other tools itself, { rate: 0.1, duration: 0 } for temp
assuming absolute
idea is you feed the basal/bolus tool a "program" to execute, in the form of a declarative json document, this facilitates openaps prepping and executing "recommendations" so that they are tracked
so some tool/use/report will prep a recommendation
the basal/bolus tool will enact the recommendation, which results in operational records on how the execution itself went
John Males
@johnmales
Sep 17 2015 01:09
@bewest ahah that is starting to make sense. So a report is generated, this is then used to create a recommendation, which is then used as input to the temp basal command - is this about right? I
Scott Leibrand
@scottleibrand
Sep 17 2015 01:46
Lots of reports are generated to represent data collected from pump. Further intermediate reports (like iob.json) are generated by external scripts, and then you use those to generate a recommendedtemp.json. The enact report then attempts to execute that recommendation and reports whether it was able to, and what actual rate was set.
Ben West
@bewest
Sep 17 2015 02:02
@johnmales exactly, in an automated loop, the input to the temp basal command is the output of the "recommend basal" command
John Males
@johnmales
Sep 17 2015 02:03
Ok, more homework when I get home. Thanks to all for the excellent explanations.
diabeticgonewild
@diabeticgonewild
Sep 17 2015 02:22
Some of the OpenAPS guide is jacked up, particularly for a Windows guide. Plus there is a bunch of information that is all over the place. I think some people put too much emphasis on structure and just tried to fit that in in places where it didn't belong instead of just making a more detailed write up about subjects that needed to be included.
Dana Lewis
@danamlewis
Sep 17 2015 02:31
PRs welcome :)
Ed Raskin
@msrcgm
Sep 17 2015 04:13
Just sent a PR regarding RPi2 setup to reflect my path.
Dana Lewis
@danamlewis
Sep 17 2015 04:13
Woot :)
Ed Raskin
@msrcgm
Sep 17 2015 04:13
Please excuse some of the formatting issues, still haven't figured out the finer points of GitHub text editor, but wanted to get the substance in before I forget what I did.
Dana Lewis
@danamlewis
Sep 17 2015 04:13
(And no worries on formatting! Great approach to just start)
Ed Raskin
@msrcgm
Sep 17 2015 04:27
@danamlewis @bewest @amazaheri and world, I have now completed through step 3.2 and am now stuck. How do I move forward an start using the tools given that I am not understanding "usage" "arguments" and "workflows" ... ;)
Dana Lewis
@danamlewis
Sep 17 2015 04:29
Scroll up, I think someone else was asking some of the same Qs earlier today and there were some links shared about where to find basics about the tools
Ed Raskin
@msrcgm
Sep 17 2015 04:42
I think you are referring to the comments/questions from @eszcloud ? I read those, unfortunately, my understanding -- or lack thereof -- is less that @eszcloud. I guess I am simply confused as to what I should do next after working through chapter 3. It seems like the guide basically moves from a step by step lego instruction book to: here are a bunch of pieces.
Dana Lewis
@danamlewis
Sep 17 2015 04:42
Oh yes. That's true :)
Next step is beginning to understand and make reports
Ed Raskin
@msrcgm
Sep 17 2015 04:43
How do I understand and make reports? :)
Dana Lewis
@danamlewis
Sep 17 2015 04:44
Delegating @scottleibrand To post some links since I'm only on my phone :) stay tuned
Ed Raskin
@msrcgm
Sep 17 2015 04:44
Will do, thank you!
Scott Leibrand
@scottleibrand
Sep 17 2015 04:45
have you read the "Sample use commands" part of https://github.com/openaps/openaps/blob/master/README.md ?
or are you past that?
Scott Leibrand
@scottleibrand
Sep 17 2015 04:46
there's also a similar help text for each openaps subcommand if you do something like openaps report -h
yeah, the book is the same content as the readme
Ed Raskin
@msrcgm
Sep 17 2015 04:47
I have loaded all tools onto the pi. I break down when it comes to the usage
Essentially once section/chapter 4 begins, I cannot understand what is going on.
Scott Leibrand
@scottleibrand
Sep 17 2015 04:47
so, have you gotten to the point where you do openaps use to get the pump to do things?
scottleibrand @scottleibrand finds chapter 4
Scott Leibrand
@scottleibrand
Sep 17 2015 04:48
oh, that's what you linked me
Ed Raskin
@msrcgm
Sep 17 2015 04:49
No, I'm unable to actually use the tools or communicate with the pump.
Scott Leibrand
@scottleibrand
Sep 17 2015 04:50
ok. at this point you need to start figuring out how to translate these linux man pages / help docs into things you can actually run and get output from
Ed Raskin
@msrcgm
Sep 17 2015 04:51
Ha. I would love to...
Scott Leibrand
@scottleibrand
Sep 17 2015 04:51
read through the background until you get to the "init new openaps environment part"
then there are a bunch of commands that you can run that should actually do things
Ed Raskin
@msrcgm
Sep 17 2015 04:51
Sorry, I'm a lawyer not a coder, but I follow directions and generally have a few brain cells ....
Scott Leibrand
@scottleibrand
Sep 17 2015 04:52
yeah, I can see how these directions make little sense if you don't know how they're generally written
this kind of help page is second nature to those of us who use linux or other command-line tools
but once you get the basics figured out, it will make more sense what they're trying to tell you
so as you do each step in the step-by-step part under "init new openaps environment", look for the section up above describing the command you just ran and putting into context what it should be doing
also, it's not entirely clear, but if you do init new, you can skip the init existing part
Ed Raskin
@msrcgm
Sep 17 2015 04:56
Thanks :) I can help revise the directions and explain it in terms that make sense to me ... if that is something you are looking for.
Scott Leibrand
@scottleibrand
Sep 17 2015 04:57
yes please. :)
I was just discussing exactly that with Dana, and telling her I hope you get to the point where you can be the one submitting the PR to make this section of the docs more readable. ;-)
Ed Raskin
@msrcgm
Sep 17 2015 04:57
Ok, so I guess the first step AKA step 4.1 is to type in: openaps init myopenaps
Scott Leibrand
@scottleibrand
Sep 17 2015 04:59
yes. you should be starting in your home directory on the pi, usually denoted by just ~
if you're unsure, type cd and hit enter
Ed Raskin
@msrcgm
Sep 17 2015 05:01
Yes, my version of directions will problem include some steps that the core developers will find funny and unnecessary to have in written format (e.g. step one, turn on your computer. Step two, place fingers on little plastic keys with letters ... ;) )
Scott Leibrand
@scottleibrand
Sep 17 2015 05:01
lol
Ed Raskin
@msrcgm
Sep 17 2015 05:02
Screen Shot 2015-09-16 at 10.00.17 PM.png
Dana Lewis
@danamlewis
Sep 17 2015 05:02
:)
Ed Raskin
@msrcgm
Sep 17 2015 05:02
Ok. Here is what I get when I attempt 4.1
Scott Leibrand
@scottleibrand
Sep 17 2015 05:03
yeah, that part isn't documented yet either
Ed Raskin
@msrcgm
Sep 17 2015 05:03
Don't forget to breath either ...
Scott Leibrand
@scottleibrand
Sep 17 2015 05:03
you need to do what it suggests: is that clear?
Dana: "Step 1: open a cold Diet Coke. You'll need it." ;-)
Ed Raskin
@msrcgm
Sep 17 2015 05:08
I'm guessing its is asking for my github email and username. I entered that in and get the following
Screen Shot 2015-09-16 at 10.07.57 PM.png
Did I do that right?
Scott Leibrand
@scottleibrand
Sep 17 2015 05:09
perfect, yep.
Ed Raskin
@msrcgm
Sep 17 2015 05:11
Yay :) Step 4.1 complete? I am now in my the /myopenaps directory. What would be step 4.2?
BTW, I'm basically writing a new set of instructions as I go through each one of these steps ....
Scott Leibrand
@scottleibrand
Sep 17 2015 05:12
read the other stuff until you get down to Managing devices
most of that section is also background. the last two commands are the ones that actually do stuff
the 665455 thing is a pump serial number: you'll need to replace that with yours if you want anything to work. :)
as I mentioned before, any command with a -h is usually a "show me what I can do with this command" help screen
basically a quick reference guide you'll want to be able to bring up later without having to go look for docs
Ed Raskin
@msrcgm
Sep 17 2015 05:15
It seems like step 4.2 is "Init existing git repo as openaps-environment" Can you please translate the following: If you already have a git repo which you would like to become a valid openaps environent, in the root of your repo, run ....
Scott Leibrand
@scottleibrand
Sep 17 2015 05:15
I mentioned earlier, you can skip that if you're creating a new one
Ed Raskin
@msrcgm
Sep 17 2015 05:15
How do I know if I already have a git repo that I would like to be come a valid openaps environment?
Sorry, I missed that ...
Scott Leibrand
@scottleibrand
Sep 17 2015 05:15
if you don't know what that means, you don't :)
Ed Raskin
@msrcgm
Sep 17 2015 05:15
Got it
Scott Leibrand
@scottleibrand
Sep 17 2015 05:16
I actually can't imagine anyone following this guide wanting to do that
either you start a new one, or you clone one off github
Ed Raskin
@msrcgm
Sep 17 2015 05:16
I'm also trying to be a good contributor and translate as much of this into non-tech as possible + understand more. Sorry for the zillions of questions.
Scott Leibrand
@scottleibrand
Sep 17 2015 05:17
exactly what we need. I could try, but it probably still wouldn't make sense. so best if the intended audience does it. :)
Ed Raskin
@msrcgm
Sep 17 2015 05:21
Ok, so step 4.2 is to basically use the openaps device add pump medtronic [serial number]
Scott Leibrand
@scottleibrand
Sep 17 2015 05:21
yes, and do the same for your cgm
Ed Raskin
@msrcgm
Sep 17 2015 05:22
Ok, eating an elephant one bite at a time here ... :)
Scott Leibrand
@scottleibrand
Sep 17 2015 05:22
hope you have a big BBQ pit
Ed Raskin
@msrcgm
Sep 17 2015 05:23
This message was deleted
Scott Leibrand
@scottleibrand
Sep 17 2015 05:24
yes. but delete that post off gitter
Dana Lewis
@danamlewis
Sep 17 2015 05:24
Bahhhhh don't post that ever on the internet
Just as a safety precaution
Scott Leibrand
@scottleibrand
Sep 17 2015 05:24
there's a ... to the right of your message on gitter: hit that, then delete
you have to mouse-over the timestamp to see it
the serial number is all anyone needs when in radio range of your pump to take full control of it and bolus you
Ed Raskin
@msrcgm
Sep 17 2015 05:26
done
Scott Leibrand
@scottleibrand
Sep 17 2015 05:26
that's why medtronic had to take the functionality out of 523 and later in the US
thx
we probably should take the example serial number out of the docs and replace it with that warning
Dana Lewis
@danamlewis
Sep 17 2015 05:27
Yes
Scott Leibrand
@scottleibrand
Sep 17 2015 05:27
pretty sure that # was never a real live pump, just one ben made available over the Internet for testing
Ed Raskin
@msrcgm
Sep 17 2015 05:27
I was using example numbers anyway ....
Scott Leibrand
@scottleibrand
Sep 17 2015 05:27
heh good
and that wasn't my password I just accidentally pasted into chat either. ;-)
Dana Lewis
@danamlewis
Sep 17 2015 05:28
But a good reminder to be careful in general with screenshots etc
Scott Leibrand
@scottleibrand
Sep 17 2015 05:28
you have no idea how many times I've had to change my work password after typing it into the wrong window. ;-)
Ed Raskin
@msrcgm
Sep 17 2015 05:28
*msrcgm goes forth and edits his working draft revised instruction manual .....
moving back to the 'openaps init myopenaps' -- what happened there so I can explain it in the text. I can deduce it interacted with my github account in some way because it wanted my username and email ... but what happened under the hood?
Scott Leibrand
@scottleibrand
Sep 17 2015 05:29
it didn't actually, yet...
you could've put anything you wanted in as your name and email
what it did was simply create a local git repo on the pi
openaps uses git under the hood
Ed Raskin
@msrcgm
Sep 17 2015 05:30
ok -- and a repo is just a kind of like a file folder that is formatted in a way that plays nice with github, yes?
Scott Leibrand
@scottleibrand
Sep 17 2015 05:30
in future if you wanted to you could push your openaps git repo up to github, but few of us do
exactly
with a special .git subdirectory in it for tracking all the history
git was actually a tool that predated github
github is just the "cloud" git repository
Ed Raskin
@msrcgm
Sep 17 2015 05:31
got it. trying to put that kind of basic information into the explanation document.
Scott Leibrand
@scottleibrand
Sep 17 2015 05:32
the reason for using git in openaps is so you can have a full copy and log of everything your openaps has ever done
great for auditing. I think you lawyers have a few similar tools. :)
Ed Raskin
@msrcgm
Sep 17 2015 05:36
Ha. Yes, they cost thousands of dollars and they generally overcharge us lawyers because we don't understand what they are selling us.
Scott Leibrand
@scottleibrand
Sep 17 2015 05:36
so you just need to go install git instead. ;)
Ed Raskin
@msrcgm
Sep 17 2015 05:52
ok, I already have a page of instructions. I am just typing those instructions in word, do you want to see work in progress, if so -- what is the best way to share that working document?
Scott Leibrand
@scottleibrand
Sep 17 2015 05:53
you should fork the git book and edit it there
Dana Lewis
@danamlewis
Sep 17 2015 05:53
Or create a gist if you want us to review prior to a PR
Scott Leibrand
@scottleibrand
Sep 17 2015 05:53
or if you're not ready for that yet, yeah, gist
John Males
@johnmales
Sep 17 2015 05:57
Is it known whether the 640g has all the remote functionality disabled? Just thinking of the future.
Ed Raskin
@msrcgm
Sep 17 2015 05:58
So, it will be pretty different from the current version ... I don't want to hack it up if no one likes my edits. What is gist?
PS, I do google this stuff before I ask. I did, and I just don't get it ....
Dana Lewis
@danamlewis
Sep 17 2015 05:59
@msrcgm Look to top or top right of GitHub, should be link for gist. Kind of like a notepad for GitHub and you can give us the url to review it
No pr required, just create new, you can save it as secret so only those w url will see it
Heading offline for the evening, night!
Ed Raskin
@msrcgm
Sep 17 2015 06:03
got it. Here is the url for my secret gist -- calling it Chapter 4 https://gist.github.com/msrcgm/0fafb4bd8445687c2906
goodnite. Thanks @danamlewis "How do I ... thank you ;)
Dana Lewis
@danamlewis
Sep 17 2015 06:07
:)
Ed Raskin
@msrcgm
Sep 17 2015 06:33
So I am at the point where I use "use" the "use" commands, save reports, etc. -- but ... to what end? It looks like the next stages of instructions in that regard are still in progress?
Ben West
@bewest
Sep 17 2015 06:39
@msrcgm , idea is the use commands allow practicing the report commands
report commands "save" how you want to use device, and then can be put in very simple script to run the "loop"
basically the rough idea is to configure bunch of reports
then run in them all in order
Ed Raskin
@msrcgm
Sep 17 2015 06:47
ok, stupid question -- but how do the reports turn into a script to run an increased basal to deal with a high blood sugar? In other words, I can pull tons of information from the pump, I can also tell it to increase the basal from .6 to .9 u/hr starting at 1:00 a.m. and ending at 2:00p.m., but how is that closing the loop? I know there is a very simple logical connection, but where is the algorithm calculation, and how do I make that happen based on CGM glucose?
Ben West
@bewest
Sep 17 2015 06:47
good question
Ed Raskin
@msrcgm
Sep 17 2015 06:48
hahaha
Ben West
@bewest
Sep 17 2015 06:48
I covered this a bit earlier, but we try to decompose all the steps into isolated, discrete steps
so idea is from high level perspective:
  • get updated data from devices
  • run algorithms to get predictions, recommendations
  • run "enact" scripts to make needed changes
openaps is toolkit to practice "using" things
we layer things into reports
so we configure "reports" to automatically run stuff (we practice with use)
then we put all the reports in special order in very simple script
Ed Raskin
@msrcgm
Sep 17 2015 06:50
I guess I am missing in the instructions where the algorithm outputs predictions and how to "calibrate"
Ben West
@bewest
Sep 17 2015 06:51
very good question
Ed Raskin
@msrcgm
Sep 17 2015 06:51
that algoritim with -- I presume -- pump settings
Ben West
@bewest
Sep 17 2015 06:51
so if we imagine some program that runs every 5 minutes
Ed Raskin
@msrcgm
Sep 17 2015 06:51
that I read from the pump and "use" in a "report" linked together with the prediction...
Ben West
@bewest
Sep 17 2015 06:52
idea is to run bunch of openaps reports
first, it will gather data from devices
last it will report everything to eg power BI/nightscout
in the middle, we'll run some openaps utilities to generate recommendations, and then enact basal change
but for now, we rely on training to train people to write these themselves, for legal reasons :-(
Ed Raskin
@msrcgm
Sep 17 2015 06:54
I am having flashbacks to law school with your socratic method responses ;)
Ben West
@bewest
Sep 17 2015 06:54
hahaha I take that as compliment
yes, we have to rely on first-principles
promise to give as much help as needed, we want successes
but it's a responsibility, we have to ask people to write their own
then we can compare/review etc
Ed Raskin
@msrcgm
Sep 17 2015 06:56
I have the first principles -- see e.g. my ability to use/run reports and save them, but I can't even begin to write some sort of conditional algorithm. I think I've officially arrived at my roadblock. Let's discuss when I see you on Friday?
Ben West
@bewest
Sep 17 2015 06:56
yes
it's ok
yes
you have gotten very far
no worries :-)
you will be doing this soon
keep playing around with the uses
you want to feel comfortable using the uses to do stuff
Ed Raskin
@msrcgm
Sep 17 2015 06:58
hahaha ... I don't feel like it. All I have done is is build a calculator that allows you to punch in numbers but never get any answers. The add, subtract, multiply and divide buttons don't work ....
Ben West
@bewest
Sep 17 2015 06:59
for this project we have to focus very carefully
we're changing the insulin that goes into someone, for better or worse
Ed Raskin
@msrcgm
Sep 17 2015 06:59
I guess the problem is that there is nothing interesting to do with the uses .... I can read basal profile a and suspend the pump ... :sparkles:
Ben West
@bewest
Sep 17 2015 06:59
big responsibility there, as everyone knows, there's an active question as to whether we should be allowed to even do this
I testified to us copyright office in defense of this, the 1201 hearings
that's a federal hearing
the prevailing argument is that we should not do this at all
this risk is too great
it's tricky
lot of different stakeholders
and everyone is honestly, genuinely for the most part trying to protect, even the vendors I've met
Ed Raskin
@msrcgm
Sep 17 2015 07:01
I totally get that ... plus its my kid, its not me, so I'm not exactly going to hook a death machine onto the guy written by a dad that doesn't know what the eff the "cd" stands for "cd myopenaps" without being really effing sure it works 1000% of the time ...
Ben West
@bewest
Sep 17 2015 07:02
yes, you are doing great
really
I spent 5 years discovering how to talk to the medtronic
I will recommend a little patience, something I don't often do
you are doing all the right things
it's hard
some of it is hard on purpose, it's great training, but we are invested in you, you will succeed
lot of this is just matter of vocabulary
Ed Raskin
@msrcgm
Sep 17 2015 07:07
I'm fine with patience, but I do better when I understand where I am going. It feels a little bit like the process is set up like a test to prove worthiness. If that's the case, I'm fine with the developers essentially having a litmus test of tasks to perform (I'm thinking indiana jones and the last crusade here ...) to ensure that they aren't handing over the holy grail to those that will use it to unintentionally kill themselves or their kids ....
Ben West
@bewest
Sep 17 2015 07:07
when we meet, we'll run through lot of how this works
it's not to prove worthinesss, lot of it is limited basically on what the tool can do
I made the tool to do certain things and not do other things
Ed Raskin
@msrcgm
Sep 17 2015 07:08
worthiness is the wrong word
Ben West
@bewest
Sep 17 2015 07:09
I had a lot of constraints during the design
Ed Raskin
@msrcgm
Sep 17 2015 07:09
A level of fundamental knowledge is a better word
Ben West
@bewest
Sep 17 2015 07:09
right, it has certain demands
but I'm confident you're meeting them
you've been playing with the use commands
I can walk you through setting up a loop that reports dexcom + medtronic to nightscout
then you have full blown monitoring
then we walk you through considerations on adding predicting/monitoring
there's simply no software that automates it yet
but it's the same tools/process
in our little community already, we have at least 3 different ways of generating potential doses
I decided to be agnostic
force user to say exactly what they want
Ed Raskin
@msrcgm
Sep 17 2015 07:11
While makes sense. But at the end of that road (assuming the road is defined by the building an open aps document, only gets one to the point of being able to read data and send commands. My issue is that I don't have the mathematical know how to generate a prediction and send a command based on a prediction ...
Ben West
@bewest
Sep 17 2015 07:12
right, that's the plugins
I'll teach you how to play around with plugins, there's a very nice "agp" one that is harmless
basially the DIYPS folks, dana and scott came up with the DIYPS algorithm
so it's possible to add that as plugin/device to openaps
then use that same as medtronic/dexcom
so it's same deal with openaps use/report
you can practice with use
save the config with report
then add it to very very simple loop.sh
there's a built-in tool that runs the whole thing every 5 minutes
just getting familiar with openaps use will go long way
I'll show how you can add plugins
so you can choose who/what to trust yourself
I created a suite where there are lots of isolated tools
I'll show you how they can connect together
you won't need to know how to write new tool
Ed Raskin
@msrcgm
Sep 17 2015 07:15
I can't wait to get together on Friday and learn more. I guess, the issue is that the DIYPS algorithm is private, right? So I've gotta write my own algorithm?
Ben West
@bewest
Sep 17 2015 07:15
no
it's actually all "out there"
in fact, there are two potentially competing versions "out there"
I'm agnostic, I'm the plumbing guy
Ed Raskin
@msrcgm
Sep 17 2015 07:16
And by out there --- in the matrix?
Ben West
@bewest
Sep 17 2015 07:16
sure, as in you can do either or
I just make the matrix work
I let others decide what goes on it
Ed Raskin
@msrcgm
Sep 17 2015 07:17
You are the architect ....
Ben West
@bewest
Sep 17 2015 07:17
and I will teach you how to choose what you want in your version
Ed Raskin
@msrcgm
Sep 17 2015 07:17
And where to get it ....? and how each one works ....?
Ben West
@bewest
Sep 17 2015 07:18
well
some assembly required
but you'll find the authors will be helpful
it won't be just me
most people that write software are very interested in other people using it
Ed Raskin
@msrcgm
Sep 17 2015 07:18
'till Friday then, Ben the great and powerful!
Ben West
@bewest
Sep 17 2015 07:19
pfff
:-)
Ed Raskin
@msrcgm
Sep 17 2015 07:19
When do you get in?
Do you need a pick up from the airport/anything at all?
Ben West
@bewest
Sep 17 2015 07:19
thursday night
sounds like ali has got me
I'm arriving just after 5
Ed Raskin
@msrcgm
Sep 17 2015 07:20
Ok -- you staying at the disneyland hotel on Thursday?
Ben West
@bewest
Sep 17 2015 07:21
I think I'm staying with the magnificent and generous ali
Ed Raskin
@msrcgm
Sep 17 2015 07:24
Ok, I'm sure he will take great care of you. Enjoy your time with him. Any objections to getting together at my office on Friday (or Ali's office/home/whatever) so that we are not in a very uncomfortable coffee shop with questionable WIFI? I would like to be able to have all tools available and easy to use during our get together and maybe do some work together ... Thanks for being generous with your time and patient with a beginner like me. You are too kind.
Jason Calabrese
@jasoncalabrese
Sep 17 2015 07:27
posted over at /public, but it anyone here is interested here's the presentation we've been working on
Ed Raskin
@msrcgm
Sep 17 2015 07:27
I'm signing off ... thanks @bewest @danamlewis @scottleibrand @amazaheri for helping me get this far.
Jason Calabrese
@jasoncalabrese
Sep 17 2015 07:27
we have 3 sessions, this is the 2nd, Wes and James are doing the first on more community/history topics
Ed Raskin
@msrcgm
Sep 17 2015 18:06
A general "cron" question -- what is the significance of using cron in a predictive/contingent environment like a closed loop? I am specifically referring to the very last part of Chapter 4 "Further Integration". I cannot think of think of a scenario where I want to increase or decrease a temp basal based solely (or even partially) on the time of day. I'm clearly missing something and seek to understand ....
Here is my guess and tell me if I am wrong -- my cron will be "every five minutes" look at cgm glucose report and make a decision per algorithm? Is that the only cron purpose?
Or am I off?
Rachel Sandlain
@audiefile
Sep 17 2015 18:08
In my limited understanding cron allows you to automatically schedule scripts to run and do stuff
so in this case people develop a loop script to go through the same steps and react similarly based off the data
cron lets you run the script automatically, say every 5 or 10 minutes
(by react similarly I mean in the way that (in general) you calculate a correction bolus the same way every time but you get different results because you have different data (bg, iob))
Ed Raskin
@msrcgm
Sep 17 2015 18:11
Ha! I guess I figured it out. The way the instruction is written makes it sound like the cron component is more "complicated"/significant than "check the dexcom cgm glucose report file every five minutes in order to make a dosing decision". Its just the way my brain works, the step seems important but proportionately insignificant to the rest of the script
Which is still a mystery to me.
Rachel Sandlain
@audiefile
Sep 17 2015 18:12
sort of. cron lets you step out of the process. otherwise it's a manual system in that you have to push a button every time you want something to happen
the loop script is what actually does the work
Dana Lewis
@danamlewis
Sep 17 2015 18:23
@msrcgm It's significant because it's automation. If you've mucked up a step or a calculation, automating that to repeat is why there has to be all kinds of safety checks and settings to make sure there's caps on what you're automating. Because even if you're automating something correctly, stuff still happens. All about safety for all scenarios.
Ed Raskin
@msrcgm
Sep 17 2015 18:29
Right, but all that "stuff" is not really part of the cron "task" (i.e. its significant, but its not were the magic happens, its simply creates a moment in time for the magic to take place). It creates a time for everyone to convene and reconvene, but it has no say in what will actually happen at this recurring five minute meeting. That's what I mean by its "insignificant". Its a calendar invite -- getting people together is great -- you can't have magic without the meeting, but its what comes out of the meeting that is significant... ;)
eszcloud
@eszcloud
Sep 17 2015 18:37
From the issue that I reported a couple of days ago, I have a carelink file as well as 32 pages of info from a pump all on bitbucket. Any idea how I can connect that private repo to someone who can use it for addressing that issue or who that person/s might be?
Sulka Haro
@sulkaharo
Sep 17 2015 18:45
LOL I got a message from someone who had talked with a Medtronic rep. 640G will get Connect within the next 12 months (no ETA), but the rep suspected "the Nightscout people will probably crack the system before that". Promising! :)
Also just learned there's a new version of the sensors coming for even the Veo, which has a new design for the under-the-skin component, which is apparently much more reliable. Same tech as in 640G, which we heard has proven to be much better than the old Enlite.
Chris Oattes
@cjo20
Sep 17 2015 19:04
do we know how the 640G data is encrypted?
Dana Lewis
@danamlewis
Sep 17 2015 19:05
@sulkaharo Not sure if it's the same thing, but my friend has done trials for the Enlite 3 sensor and thinks it's getting closer to dex
also getting ready to eat lunch with someone on a 670g for a trial :)
Chris Oattes
@cjo20
Sep 17 2015 19:06
I want to do a trial for 670g, there aren't any studies nearby though :(
Scott Leibrand
@scottleibrand
Sep 17 2015 19:14
@msrcgm I think you are understanding cron correctly, yes. The significance is all related to whatever you tell your cron to actually run. If you have something screwed up in your script, you could have it repeatedly high- or low- temp you when it shouldn't, and if you're not watching and it does that for hours, you could end up quite low or high as a result.
Another alternative to cron is what they call a "daemon", which runs continuously, in this case looping through BG and IOB checks until it decides something needs done. Same issues with doing it that way, of course.
Ben West
@bewest
Sep 17 2015 19:19
cron simply runs a "job" when configured
so the job is up to us to write individually then share/review/compare, eventually there will be an openaps tool that helps with that too, but we are early days still
the cron config just tells cron when to run the job, which is set to every five minutes
it does "get all the data to make a dosing decision", but that's a fair number of steps to make sure everything is sane and recent
lot of people working on different steps/adding/improving/documenting
Ed Raskin
@msrcgm
Sep 17 2015 20:07
Sounds like I get it :clap: