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

1st
Sep 2015
Dana Lewis
@danamlewis
Sep 01 2015 03:01
@channemann wow, finally getting a chance to look at your writeup. Nice work! and @bewest that's a really cool tool. Looks like a few chunks of text run over the space avail, is there a scroll option within a block, or do you just have to break it into multiple blocks in that case?
Ben West
@bewest
Sep 01 2015 03:04
for this, you have to break it up
it's @channemann's thing tweaked slightly
Dana Lewis
@danamlewis
Sep 01 2015 03:04
K. That's easily doable. And makes you really focus on making sure everything is bite sized chunks, which is nice :)
Ben West
@bewest
Sep 01 2015 03:04
yeah, sometimes constraints are nice
a lot of things people have been breaking up into separate steps can be combined
Dana Lewis
@danamlewis
Sep 01 2015 03:05
Ya, once we nail the writeup for the basic version and can iterate more about tests etc., this would be good. I like how it has your descriptions balanced in as an intro to next steps in some places.
Ben West
@bewest
Sep 01 2015 03:05
I've put several gists together you can just curl $gist | sudo bash to run lots of stuff in one go
ah, thanks, that was experiment
Dana Lewis
@danamlewis
Sep 01 2015 03:06
I think it really helps people vs just blindly copy and paste and not know what each part does, esp. if they're less familiar with certain concepts in coding
That's one of the things I want to do w/ our safety defaults, is make it so even if you just copy and paste lots of stuff, no go (or no go far) without testing and understanding.
Preaching to the choir, I know :)
Chris Hannemann
@channemann
Sep 01 2015 03:13
Thanks, @danamlewis; just to be clear, that's not yet verified for public consumption, just a brain dump ;-)
Dana Lewis
@danamlewis
Sep 01 2015 03:13
Understood :)
Not claiming to verify, either, just to like it :) and the direction everyone's heading in helping clarify all the chaos around the project!
Ben West
@bewest
Sep 01 2015 03:14
the install step can be changed to curl -s https://gist.githubusercontent.com/bewest/d5f0fc68c0b9ac49c146/raw/1bcc97d1ed8ddf2c17ee4c4e79f83ec2201a43ab/install_tools.sh | sudo bash
and that'll install everything in one go
Chris Hannemann
@channemann
Sep 01 2015 03:15
Is it weird that I find it fun to simply make the blocks flip a bunch?
Dana Lewis
@danamlewis
Sep 01 2015 03:15
It's therapeutic
Chris Hannemann
@channemann
Sep 01 2015 03:15
Quite
Ben West
@bewest
Sep 01 2015 03:15
no, I spent at least 3 hours doing that
Chris Hannemann
@channemann
Sep 01 2015 03:15
ha
Ben West
@bewest
Sep 01 2015 03:15
you can press esc
Dana Lewis
@danamlewis
Sep 01 2015 03:15
Reminds me of Prezi, except a little less vertigo-inducing because it's more structured
Ben West
@bewest
Sep 01 2015 03:15
and get the grids to line up in funny ways
Chris Hannemann
@channemann
Sep 01 2015 03:16
Can you scroll on a block?
Ben West
@bewest
Sep 01 2015 03:16
no
Dana Lewis
@danamlewis
Sep 01 2015 03:16
No, that's why you have to cut it into small chunks
Ben West
@bewest
Sep 01 2015 03:16
need to break those up ;-)
Chris Hannemann
@channemann
Sep 01 2015 03:16
No worries, just a demo
Ben West
@bewest
Sep 01 2015 03:16
yeah, surprised how good some of them look
just wanted to see mix of too small, too big, etc
you'll notice an input on one of them
thinkin was to walk people through the steps "name your openaps: "
name: ____
Chris Hannemann
@channemann
Sep 01 2015 03:17
Ben West
@bewest
Sep 01 2015 03:17
and then the rest of the instructions would update to include that name
lol
your ending was kind of anti-climactic
thought it needed something a little more final, lol
Dana Lewis
@danamlewis
Sep 01 2015 03:18
Ha, ya
Chris Hannemann
@channemann
Sep 01 2015 03:18
I like it
Dana Lewis
@danamlewis
Sep 01 2015 03:18
Except it's more like "confetti!!!" needed
Chris Hannemann
@channemann
Sep 01 2015 03:18
Wait, which one has input?
Dana Lewis
@danamlewis
Sep 01 2015 03:18
Or take a pic and post it with #OpenAPS or something ;) to spread the joy
so thinking is you start typing in your intended names
Chris Hannemann
@channemann
Sep 01 2015 03:18
Should immediately order you a pizza and then post your results 12 hours later ...
Ben West
@bewest
Sep 01 2015 03:18
maybe on the explanation screens
Dana Lewis
@danamlewis
Sep 01 2015 03:19
Only if it already knows your carb absorption rate and does 'eating soon mode' ;)
Chris Hannemann
@channemann
Sep 01 2015 03:19
Better coordinate with the delivery guy
Ben West
@bewest
Sep 01 2015 03:19
then after you type your name, the docs would render using the name you chose
Chris Hannemann
@channemann
Sep 01 2015 03:20
Yeah, that input is cool if you can get it to perpetuate
Ben West
@bewest
Sep 01 2015 03:20
so you could copy/paste without any confusing <replace-something-in-here>
Chris Hannemann
@channemann
Sep 01 2015 03:20
Right
Dana Lewis
@danamlewis
Sep 01 2015 03:20
Does it perpetuate? That would be AWESOME
Chris Hannemann
@channemann
Sep 01 2015 03:20
although, I suppose that small hurdle is sort of good
Dana Lewis
@danamlewis
Sep 01 2015 03:20
that's beginner's most common hangup is getting confused in subsequent steps about what the names are
Chris Hannemann
@channemann
Sep 01 2015 03:21
it would be ideal to make it perpetuate but also highlight where it is being inserted
Ben West
@bewest
Sep 01 2015 03:21
yeah, but that's the stumbling block on purpose now
but just to get basic monitoring stuff set up
Dana Lewis
@danamlewis
Sep 01 2015 03:21
Hmm.
Ben West
@bewest
Sep 01 2015 03:21
that's fine
don't have to complete the loop for them
Chris Hannemann
@channemann
Sep 01 2015 03:21
right
Ben West
@bewest
Sep 01 2015 03:21
getting-started should definitely be about wrangling/owning your data I think
Dana Lewis
@danamlewis
Sep 01 2015 03:21
Then some more steps or explanation pages to perpetuate then. "Remember my_open_aps_loop = (RANDOM NAME YOU MADE UP)"
Ben West
@bewest
Sep 01 2015 03:22
no, in the intro just hammer it in
repatedly
Dana Lewis
@danamlewis
Sep 01 2015 03:22
yea
Ben West
@bewest
Sep 01 2015 03:22
"this <foonameyou made up> is the "home" for your openaps"
all openaps commands need to be from inside that place
Chris Hannemann
@channemann
Sep 01 2015 03:23
I know I had some trouble realizing that the report names had .json in them
Sort of the next level of confusion
So that when you invoke a report, you invoke the resulting filename
Just not the way I think
Ben West
@bewest
Sep 01 2015 03:24
yeah, report name == filename
Chris Hannemann
@channemann
Sep 01 2015 03:24
It's obvious now, but before I was thinking "why is the report filename the process I go through to generate the report?"
Ben West
@bewest
Sep 01 2015 03:25
maybe rename invoke to something else
thing is it could be doing anything... making changes in the world (basal/dose) vs reading model, vs posting data to website....
so figured it's kind of like a spell you assembled
how would you feel about openaps report cast
Dana Lewis
@danamlewis
Sep 01 2015 03:28
That makes more sense at first glance esp. when you describe it as a spell :)
Chris Hannemann
@channemann
Sep 01 2015 03:28
really going D&D here, eh?
Dana Lewis
@danamlewis
Sep 01 2015 03:28
HP
Ben West
@bewest
Sep 01 2015 03:28
dunno
or cook
Dana Lewis
@danamlewis
Sep 01 2015 03:28
::insert geek thing of your choice::
Ben West
@bewest
Sep 01 2015 03:28
like a recipe book
Dana Lewis
@danamlewis
Sep 01 2015 03:28
hm
Ben West
@bewest
Sep 01 2015 03:28
cook works well
literally bake the file
Chris Hannemann
@channemann
Sep 01 2015 03:29
invoke works, I just needed a plain english explanation
Ben West
@bewest
Sep 01 2015 03:30
dunno, I googled/thesaurized around the web for awhile on that one
Chris Hannemann
@channemann
Sep 01 2015 03:30
the key is that openaps invoke is doing something and then generating a report about what it did
Ben West
@bewest
Sep 01 2015 03:30
someone else was looking at similar semantics and invoke had bitter complaints
I like cook more and more
Chris Hannemann
@channemann
Sep 01 2015 03:31
I was thinking about it as a two-step process, so I expected something more like openaps invoke <some_process> <output_report.json>, but since the report is tied to the process, the <some_process> becomes redundant
I don't think invoke is any worse than cook or bake or conjure
Just my own thought process causing a blockage; I'm good with it now
Dana Lewis
@danamlewis
Sep 01 2015 03:33
Right but I agree about some plain english explanations (and this is where analogy about cooking/baking might help)
Jason Calabrese
@jasoncalabrese
Sep 01 2015 03:43
@bewest thats nice, makes me feel guilty for not doing enough with my pumps
Dana Lewis
@danamlewis
Sep 01 2015 03:51
@jasoncalabrese as you've been doing all gazillion things on NS? :)
if you fill in the blanks, the docs update themselves
Nathan Racklyeft
@loudnate
Sep 01 2015 05:43
Cool!
Chris Hannemann
@channemann
Sep 01 2015 16:19
Hmmm, still not perpetuating for me, @bewest
But I'm down with the concept; very cool
Chris Hannemann
@channemann
Sep 01 2015 16:28
Battery pack question: it seems that most battery packs do not permit simultaneous charging and discharging, which means you must kill the system and either plug it into another battery or the wall and restart when you want to recharge your battery.
I found one that works—TeckNet® POWER BANK 9000mAh USB External Battery Backup Pack—but I was wondering if there were any others out there.
Rachel Sandlain
@audiefile
Sep 01 2015 18:55
I got this one since it was suggested on the docs on gitbook. If the batter isn't dead it will charge and discharge at the same time in my experience.http://www.amazon.com/dp/B00M6V0R2C/ref=wl_it_dp_o_pC_nS_ttl?_encoding=UTF8&colid=2OYKR43UGE0YB&coliid=IC4EHVFRTC117&psc=1
Ben West
@bewest
Sep 01 2015 19:09
apparently it will do some scrollbars for code snippets, which is nice:
blob
Chris Hannemann
@channemann
Sep 01 2015 19:14
Thanks, @audiefile; I thought that one might work, but couldn't find any verification.
Rachel Sandlain
@audiefile
Sep 01 2015 19:16
:) FYI it's pretty heavy (a lot more than I expected anyway) so it still might not be the best option for mobile use
Chris Hannemann
@channemann
Sep 01 2015 19:24
Not sure how much better one can do at the moment given general energy density limitations of batteries
Rachel Sandlain
@audiefile
Sep 01 2015 19:27
Maybe you can't. It was a lot heavier than I anticipated but it also has a lot larger capacity than the other battery packs I've used
Chris Hannemann
@channemann
Sep 01 2015 19:34
Oh wow, you're right; just checked and it is nearly twice the weight of the one I'm using. 12,000 mAh @ 8.96 oz vs. 9,000 mAh @ 4.64 oz
Rachel Sandlain
@audiefile
Sep 01 2015 19:35
That's a huge difference! Good to know I'm not crazy though
Chris Hannemann
@channemann
Sep 01 2015 19:41
Better start a weightlifting program ...
Rachel Sandlain
@audiefile
Sep 01 2015 19:46
Nah I'll just get a second one. No need for free weights now
Chris Hannemann
@channemann
Sep 01 2015 19:53
:+1:
Oskar Pearson
@oskarpearson
Sep 01 2015 20:43
Heya all. I'm getting to grips with the source code and how the system functions. I was wondering if someone would be able to tell me how https://github.com/openaps/openaps-js and the openaps-agp modules are used, and what their overall "place" is in the system?
Oskar Pearson
@oskarpearson
Sep 01 2015 20:49
Overall, my personal intention is to work towards a closed loop system for my wife. We don't really have any requirement for nightscout's current-glucose-levels sharing functionality, but we do have a requirement for sensible graphing and visualisation of glucose patterns, plus any functionality nightscout would give us WRT closing the loop
I see there's some talk yesterday about implementing default algorithms. I am still lacking a bit of context, so I was wondering if those algorithms are mean to work towards closing the loop, or whether they are something else entirely :)
Dana Lewis
@danamlewis
Sep 01 2015 20:51
Yes they are :0
*:)
Oskar Pearson
@oskarpearson
Sep 01 2015 20:51
Awesome
Dana Lewis
@danamlewis
Sep 01 2015 20:51
Nightscout is good for the visualization part you mentioned. We use the iob-cob branch, but it's outdated, and our goal in the next few weeks is to get it up to speed and work on some other visualization options people could use to see when the loop is running
@amazaheri has a different NS branch that's also a good view for now if you wanted to check that out (he's also running #OpenAPS for his son)
Have you joined the openaps-dev google group yet?
If not, that's the first thing to do, and I have a (very outdated) "getting started" message in a thread there that's a good first start. We have had some updates & great progress (thanks to @loudnate @bewest @channemann @amazaheri @scottleibrand &many others) behind the scenes on making parts more easy to fit together, but not quite ready (next few weeks likely) to roll out a new getting started suggestion to the group just yet. So if you're not itching to do it right this moment, there should be some more coming soon in the next month or so.
Rachel Sandlain
@audiefile
Sep 01 2015 21:01
The super simple answer is that opens has a bunch of code that can talk directly to certain medtronic pumps and cgms, and to dexcom cgms. It can then generate some simple reports based off of that communication. opens-js is a group of javascript programs that uses openaps (and it's reports) to do stuff (figure out iob, figure out a temp basal)
Chris Hannemann
@channemann
Sep 01 2015 21:06
@oskarpearson If you have an appropriate Medtronic pump and Medtronic or Dexcom CGM, I would highly recommend installing the openaps tools and getting familiar with how they function. Understanding the use and report commands as well as how vendors, devices, and aliases work will go a long way towards wrapping your head around openaps-js or whatever other method you use to build the actual looping script.
Oskar Pearson
@oskarpearson
Sep 01 2015 21:06
Thanks for the helpful welcome! I've joined the dev group, yep. And I've started working through the documentation and historical posts. I'll look for that set specifically.
We've just received the 540G (which I understand doesn't work atm). But we have the previous 722 and sensor kit available for testing. I wanted to play previously, but was very wary of doing something stupid that'd lock out our only pump
(the firmware on the pump allow setting of basals too, so it seems like an ideal option)
I've got openaps tools talking to the pump - but haven't played with the report command yet.
Dana Lewis
@danamlewis
Sep 01 2015 21:10
Where are you geographically?
Oskar Pearson
@oskarpearson
Sep 01 2015 21:11
Thanks for clarifying the relationship between the JS and the openaps code
London, UK.
Dana Lewis
@danamlewis
Sep 01 2015 21:12
Ah, yes, ya'll have different pumps to play with
Oskar Pearson
@oskarpearson
Sep 01 2015 21:15
While I'm pestering you all: I read that some people have multiple carelink receivers / raspberry pi's around the house, so that they are more likely to get reception. Would each pi run it's own copy of the code, and be responsible for setting the new basal value based on the results of the openaps reports? Or does the decision making happen in one place ("in the cloud") and each pi just tries to relay that decision?
I'm guessing that each pi runs it's own routines. And we need to make sure that we run the same algorithm / decision making code on all pis
Since the pi sets a temporary basal that overrides any previous basal, the risks are low.
Dana Lewis
@danamlewis
Sep 01 2015 21:16
Yes for multiple receivers, or for having one set with a battery in a small bag that travels around for you. Yes, each runs it's own copy. Whichever reads and picks up gets to play and set a temp basal if it thinks it should change something.
Oskar Pearson
@oskarpearson
Sep 01 2015 21:17
(it's not like pi1 would bolus 10 and pi2 would bolus another 10)
Great
Dana Lewis
@danamlewis
Sep 01 2015 21:17
Correct. Also, never ever ever build it to allow boluses ;)
but yes for basals just being re-set
Paul Andrel
@stavlor
Sep 01 2015 21:17
heh that could end badly
Oskar Pearson
@oskarpearson
Sep 01 2015 21:18
So, if the intention is "just" closing the loop, I can ignore setting up the nightscout software entirely, except as a visualisation tool
Dana Lewis
@danamlewis
Sep 01 2015 21:18
big part of openaps is designing for comm failure - ie you totally lose comm with the system after a command is issued. that's why we only do 30 min temps and it's always "if i get no more data after this, this is the safe thing to do"
Yes, but it's easy and you might as well, because you'll already have your cgm data in a mongodb which can be used for openaps.
Oskar Pearson
@oskarpearson
Sep 01 2015 21:18
Setting the basal doesn't require the data from the nightscout system
Rachel Sandlain
@audiefile
Sep 01 2015 21:18
temp basals also last for 30 min so even if you step away at the time the script runs the previously set temp basal will just keep running
Correct. Nightscout is independent but also awesome.
Oskar Pearson
@oskarpearson
Sep 01 2015 21:19
Great
Dana Lewis
@danamlewis
Sep 01 2015 21:19
True. But if you want to collect your data in the cloud to analyze later, it's the eaasiest that I know of right now to do it that way.
Rachel Sandlain
@audiefile
Sep 01 2015 21:20
Plus the pebble is super helpful in simple awareness of bg and trends, even without the whole remote monitoring aspect
Chris Hannemann
@channemann
Sep 01 2015 21:20
@danamlewis how many systems do you have set up?
Oskar Pearson
@oskarpearson
Sep 01 2015 21:20
Thanks for all your assistance and guidance. Even with my silly questions :) We both appreciate it
Dana Lewis
@danamlewis
Sep 01 2015 21:21
@channemann 4 Pis :) two of which are bedside, one tends to be dev and lives on our desk or floats around living room, and one that's my OTG version always in a bag ready to use with a battery.
@oskarpearson no silly questions, and we're glad to have more folks getting going :) so anything we can do to help let us know!
Rachel Sandlain
@audiefile
Sep 01 2015 21:21
@danamlewis and everyone what does your physical set up look like? do you use nightscout? do you use multiple cgms?
Chris Hannemann
@channemann
Sep 01 2015 21:21
Do you run extension USBs on the bedside ones?
Dana Lewis
@danamlewis
Sep 01 2015 21:22
@audiefile I have NS but don't look at it regularly, but when I want to look overnight and see when it high and low temped, it's great. And I have two CGMs but don't really need them, it's more about habit now.
Chris Hannemann
@channemann
Sep 01 2015 21:22
I'm only running one, totally mobile; sits bedside at night with the battery charging and the CL stick connected to a USB extension.
Rachel Sandlain
@audiefile
Sep 01 2015 21:22
I have. I found a retractable one. fully extended it ends up in the middle of the bed and tends to read well
Dana Lewis
@danamlewis
Sep 01 2015 21:22
@channemann yes, one carelink is in the middle of bed under mattress, the other is right off the side of the bed
Chris Hannemann
@channemann
Sep 01 2015 21:22
Sounds about right
Dana Lewis
@danamlewis
Sep 01 2015 21:23
@esteward was telling me he only has one, and it reads fine. Not sure why I have so much difficulties, but my pump changes sides depending on where my site is frequently, so maybe that impacts it
Chris Hannemann
@channemann
Sep 01 2015 21:23
My battery, pi, CL stick, and dex receiver all go in this case
I just ran a quick CL range test today, actually: https://gist.github.com/channemann/0ff376e350d94ccc9f00
Dana Lewis
@danamlewis
Sep 01 2015 21:24
The other perk to having one middle and one on bedside is that it's near door to room and when I walk by, it picks up easily even if I'm not intentionally doing it, so it regularly uploads from there too
Rileylink will be great. (and because 4 is the magic number, we got 4 of those, too) but it's a ways a way from being easily usable.
Chris Hannemann
@channemann
Sep 01 2015 21:25
@loudnate did some range testing on the RL; he essentially couldn't get out of its range!
Rachel Sandlain
@audiefile
Sep 01 2015 21:26
I'm pretty addicted to my pebble so I'll probably keep up with NS but looking at logistics. I'm not sure I want to have two cgms to keep accurate though.
Dana Lewis
@danamlewis
Sep 01 2015 21:26
I still like my pebble for DIYPS and even overnight quick glance. 2 cgms aren't a big deal to keep accurate if you're calibrating at same time. @audiefile
Paul Andrel
@stavlor
Sep 01 2015 21:27
@audiefile why would you need two cgms?
Dana Lewis
@danamlewis
Sep 01 2015 21:27
Some people like to stagger calibrations. So you could "restart" a second cgm and not lose data. Also good if you want to run 2 complete offline loops (re carelink range issues) @stavlor
Oskar Pearson
@oskarpearson
Sep 01 2015 21:28
Oh yeah. The Rileylink - I had a question there too, just to verify my understanding. It lets you pull data from a pump and push it to Nightscout, right? Would the IOS app then have to handle calculating basals to push back to the pump? And would it have functionality to do that?
Dana Lewis
@danamlewis
Sep 01 2015 21:28
@oskarpearson You would still needs the OpenAPS loop stuff, it just replaces the carelink USB and Pi and boosts connectivity range by about 10x :)
That's why it's not really ready yet, still needs work for people to even be able to use to read-only easily, no one's added it to a loop setup yet
Rachel Sandlain
@audiefile
Sep 01 2015 21:30
@stavlor I had been using two because I didn't have a share receiver and both my openaps and nightscout setup require the cgm be connected to it via usb
Pete Schwamb
@ps2
Sep 01 2015 21:30
I wouldn’t overstate the connectivity range of the RL. I still have it drop packets and don’t understand why. One theory I have is that the OOK radio encoding that the medtronic pump uses is a very poor choice. It’s easily susceptible to interference.
Sometimes it’s great, and sometimes not.
I’d like to spend some more time myself doing some rigorous tests of various parameters that might help make it more consistent.
Ali Mazaheri
@amazaheri
Sep 01 2015 21:31
nice discussion, as @danalewis mentioned i have a custom webAPI running on Azure and hence why we added sendtempbasal_azure.js to post to end point. i then store the information on azure sql which triggers push notifications to phone and my band, also power bi for reporting and updating NS treatment collection using Azure logic app. end to end solution been working solid for past 2 months and code is on repo.
Dana Lewis
@danamlewis
Sep 01 2015 21:31
True, @ps2. like with all OpenAPS loop stuff, everything has to be designed to assume faulty communication, so that's true for carelink as well as rileylink
Chris Hannemann
@channemann
Sep 01 2015 21:31
Come on, Pete; gotta jump in here and spoil the party ;-)
Ali Mazaheri
@amazaheri
Sep 01 2015 21:32
@ps2 @bewest and i are planning some testing when he is in town mid september
Rachel Sandlain
@audiefile
Sep 01 2015 21:32
@stavlor I just got a share receiver so I could plug that into my pi and use bluetooth for nightscout
Oskar Pearson
@oskarpearson
Sep 01 2015 21:35
One problem with the RileyLink from my side is that the UK 722 appears to use the 868.35 Mhz frequency instead of the (USA?) 916.50 frequency https://www.medtronicdiabetes.com/sites/default/files/library/support/carelink_usb_user_guide.pdf page 3 - specifications
Pete Schwamb
@ps2
Sep 01 2015 21:36
The RL can be built with an 868 Mhz antenna.
The same chip will do 868.
Oskar Pearson
@oskarpearson
Sep 01 2015 21:54
Thanks ps2 - I'll have a look then!
Oskar Pearson
@oskarpearson
Sep 01 2015 22:52
I'm presuming it's unnecessary
Chris Hannemann
@channemann
Sep 01 2015 22:53
@bewest I'm stepping through the docs and that readme I posted the other day to verify a clean install, but I'm getting an error during the openaps install:
pi@raspberrypi ~ $ sudo easy_install -Z openaps
Searching for openaps
Reading http://pypi.python.org/simple/openaps/
Download error on http://pypi.python.org/simple/openaps/: [Errno -2] Name or service not known -- Some packages may not be found!
Reading http://pypi.python.org/simple/openaps/
Download error on http://pypi.python.org/simple/openaps/: [Errno -2] Name or service not known -- Some packages may not be found!
Couldn't find index page for 'openaps' (maybe misspelled?)
Scanning index of all packages (this may take a while)
Reading http://pypi.python.org/simple/
Download error on http://pypi.python.org/simple/: [Errno -2] Name or service not known -- Some packages may not be found!
No local packages or download links found for openaps
error: Could not find suitable distribution for Requirement.parse('openaps')
Ben West
@bewest
Sep 01 2015 22:54
:eyes:
Chris Hannemann
@channemann
Sep 01 2015 22:54
This is after a (from what I can tell) successful sudo apt-get install python python-dev python-setuptools python-software-properties python-numpy
Ben West
@bewest
Sep 01 2015 22:56
hmm
so you're connected to internet
I just deleted one of the files...
there's no openaps on this?
Chris Hannemann
@channemann
Sep 01 2015 22:56
nope
Ben West
@bewest
Sep 01 2015 22:56
ok
Chris Hannemann
@channemann
Sep 01 2015 22:56
totally clean
Ben West
@bewest
Sep 01 2015 22:56
try again, I just deleted one of the files
not sure it makes a different
Chris Hannemann
@channemann
Sep 01 2015 22:56
same
Ben West
@bewest
Sep 01 2015 22:56
think @loudnate told me one of the builds doesn't work
but that doesn't seem to be the issue here
maybe
easy_install -ZU setuptools
sudo
dunno what it's problem is, I'm clearly listed there
Chris Hannemann
@channemann
Sep 01 2015 22:58
pi@raspberrypi ~ $ sudo easy_install -ZU setuptools
Searching for distribute
Reading http://pypi.python.org/simple/distribute/
Download error on http://pypi.python.org/simple/distribute/: [Errno -2] Name or service not known -- Some packages may not be found!
Reading http://pypi.python.org/simple/distribute/
Download error on http://pypi.python.org/simple/distribute/: [Errno -2] Name or service not known -- Some packages may not be found!
Couldn't retrieve index page for 'distribute'
Scanning index of all packages (this may take a while)
Reading http://pypi.python.org/simple/
Download error on http://pypi.python.org/simple/: [Errno -2] Name or service not known -- Some packages may not be found!
Best match: distribute 0.6.24dev-r0
Adding distribute 0.6.24dev-r0 to easy-install.pth file
Installing easy_install script to /usr/local/bin
Installing easy_install-2.7 script to /usr/local/bin
Installing easy_install-2.6 script to /usr/local/bin

Using /usr/lib/python2.7/dist-packages
Processing dependencies for distribute
Finished processing dependencies for distribute
Oskar Pearson
@oskarpearson
Sep 01 2015 22:59
@channemann It seems likely to me that your DNS is not working.
Nathan Racklyeft
@loudnate
Sep 01 2015 22:59
Yep it's your ATT internet, remember?
Tether from your iPhone and you'll be golden.
Oskar Pearson
@oskarpearson
Sep 01 2015 23:00
That can happen if you switch over from wifi to ethernet (happened to me today, on an unrelated case). I presume the same is the case for switching from ethernet to wifi. Reboot would fix that.
Chris Hannemann
@channemann
Sep 01 2015 23:00
That was an issue before with my old router, but I hadn't seen it in a long time (switched routers)
Oskar Pearson
@oskarpearson
Sep 01 2015 23:00
try run "ping www.google.com"; and see if it works?
Chris Hannemann
@channemann
Sep 01 2015 23:00
Ping works
I'll switch over to my phone here in a sec
Oskar Pearson
@oskarpearson
Sep 01 2015 23:01
and does ping pypi.python.org work?
Chris Hannemann
@channemann
Sep 01 2015 23:02
indeed
Oskar Pearson
@oskarpearson
Sep 01 2015 23:04
We can probably diagnose it, but it's going to involve some hairy use of strace and/or tcpdump. It'd also share a bunch of information about your network config and so forth with me / other people, since you'd need to put the network dumps / strace output somewhere. Unless you're familiar with those tools.
Ben West
@bewest
Sep 01 2015 23:06
or just curl https://pypi.python.org/simple/openaps/
Oskar Pearson
@oskarpearson
Sep 01 2015 23:07
That's assuming that the package is served from there, rather than some-mirror-server-somewhere.example
Chris Hannemann
@channemann
Sep 01 2015 23:07
Looks like it's working now that I tethered to my phone
Oskar Pearson
@oskarpearson
Sep 01 2015 23:07
:thumbsup:
Chris Hannemann
@channemann
Sep 01 2015 23:07
So how come I can get all the other stuff (numpy, etc) and then it fails at easy_install?
Ben West
@bewest
Sep 01 2015 23:08
someone in the middle is trying to optimize things for you
Chris Hannemann
@channemann
Sep 01 2015 23:08
"optimize"
Oskar Pearson
@oskarpearson
Sep 01 2015 23:09
@channemann it might be that python.org or your ISP is trying to point your downloads at a local mirror. The address they are using for the mirror may be invalid / bogus. So it can connect to pypi.python.org ok, which then sends easy install a redirect to the mirror.
The address referred to by the mirror doesn't exist, so the download doesn't work
Chris Hannemann
@channemann
Sep 01 2015 23:10
Ah, interesting
Ben West
@bewest
Sep 01 2015 23:10
python wouldn't do that themselves
Chris Hannemann
@channemann
Sep 01 2015 23:10
Right
Ben West
@bewest
Sep 01 2015 23:10
it's more likely someone in the middle is using old version of DNS or something
Chris Hannemann
@channemann
Sep 01 2015 23:10
Well, I'll go ahead and blame AT&T because that's easy
Oskar Pearson
@oskarpearson
Sep 01 2015 23:11
And... probably right :)
Chris Hannemann
@channemann
Sep 01 2015 23:11
I'm sure they would say "but look, you got it to work by using your AT&T phone!"
Oskar Pearson
@oskarpearson
Sep 01 2015 23:16
So, I'm busy starting to hack some code in the https://github.com/openaps/openaps repo. I notice, though, that there's no license file, so I'm not sure of Copyright status. I don't really much worry about the license status, but I am wary of contributing to this sort of project without some sort of appropriate disclaimer of implied warranty or similar. I've checked the dev mailing list, and don't see any reference to a license.
Any previous discussions I should be aware of?
Same applies for openaps/openaps-js
There are lots of potential bugs in loop.sh that I'm looking to clean up
Chris Hannemann
@channemann
Sep 01 2015 23:32
Don't really have the background on that, @oskarpearson. I'm guessing @bewest and @scottleibrand can provide some thoughts on the matter.