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

17th
Feb 2019
live4sw
@live4sw
Feb 17 14:13
Very basic question - when autotune runs nightly, is it basing its calculations on only the past 24 hours, or is it doing it over a longer period?
It looks like just 24 hours, so is it just ignoring anything prior, or is it averaging it with readings over an [x] day period? Just asking because my sensitivity has been high in the past two days (heavy workout + >15k steps/day) and was wondering how distortive that might be.
Martin Haeberli
@mhaeberli
Feb 17 14:44
Getting CC111x radio errors on one or two of our Edison rigs; reboot doesn’t fix. Suggestions? Old reflash procedure doesn’t seem to run on dev branch?
Scott Leibrand
@scottleibrand
Feb 17 15:45
Andrew: it’s iterative. It makes small adjustments from the previously autotuned settings based on each day’s new data.
(FWIW, if I ever address any of y’all like that without your handle, it’s because your Github/Gitter display name doesn’t match your username closely enough, (or it’s not unique enough) so Gitter won’t autocomplete your name on mobile.)
Dana Lewis
@danamlewis
Feb 17 15:50
@mhaeberli try resetting the Edison
Martin Haeberli
@mhaeberli
Feb 17 15:56
@danamlewis thx. As in reflashing ? I don’t think I otherwise know how to reset. I’ll also ask google
Scott Leibrand
@scottleibrand
Feb 17 15:57
@cluckj if you haven’t seen any regression from it, want to PR the re-addition of npm link?
Dana Lewis
@danamlewis
Feb 17 16:05
@mhaeberli no. Lift Edison up and resettle and replace the Edison on the explorer board. Physical reseat of the Edison component. Sometimes it wiggles lose and will be slightly separated from pins but not obviously off
(Sorry earlier typo - reseating not resetting :))
Jon Cluck
@cluckj
Feb 17 16:35
@scottleibrand I haven't seen any on the pi; the issue @rkresha was having was on the edison, though
live4sw
@live4sw
Feb 17 16:37
Thanks @scottleibrand That makes sense and is helpful, I just suggested a slight addition to the docs to reflect that.
Dana Lewis
@danamlewis
Feb 17 16:44
@live4sw haven’t seen the PR come through; can you check and make sure you hit the final green button to submit the PR?
(You may have committed to your branch; but still need to PR back to OpenAPS:master)
live4sw
@live4sw
Feb 17 16:47
Yup, just did it! I totally would have noticed that...eventually. thanks
Dana Lewis
@danamlewis
Feb 17 16:47
Great, see it now!
:+1:
Scott Leibrand
@scottleibrand
Feb 17 16:49
@live4sw thx!
Martin Haeberli
@mhaeberli
Feb 17 17:16
@danamlewis thx. I understand ‘reseating’
TranceCake
@TranceCake
Feb 17 19:51
Hi, another small question: In the docs it asks you to fill in a form if you have been looping for 3+ days. Is that full looping? because I'm currently in low-suspended-mode to learn
Dana Lewis
@danamlewis
Feb 17 20:59
Full looping, but if you’re doing maxiOB of 0 it still counts :)
PieterGit
@PieterGit
Feb 17 21:18
@jpcunningh did you or somebody else have a look at integrating xdripjs Lookout into oref0 0.7.x and oref0-setup.sh? Current dev only has
    echo "xdrip: will work with an xDrip receiver app on your Android phone"
    echo "xdrip-js: will work directly with a Dexcom G5 transmitter and will upload to Nightscout"
PieterGit
@PieterGit
Feb 17 21:39
@scottleibrand double check. am I right there should not be a reason why we install the packaged oref0 master (released version) with bin/openaps-packages.sh . I think oref0-setup.sh takes care of installing it. See proposed commit PieterGit/oref0@54a0f25
Jeremy Cunningham
@jpcunningh
Feb 17 21:45
@PieterGit, the xdrip-js option in oref0 is for Logger integration. I haven't gotten around to coding up a Lookout option.
PieterGit
@PieterGit
Feb 17 21:47
@jpcunningh first will try to get the ww stuff into shape, and put the Lookout integration on my (too long) list of todo. Feel free to beat me to it :smile:
Jeremy Cunningham
@jpcunningh
Feb 17 21:48
Great! Thanks, @PieterGit!
PieterGit
@PieterGit
Feb 17 21:48
didn't want to start coding stuff that might already have been done
Jeremy Cunningham
@jpcunningh
Feb 17 21:52
I just got our $40 pi setup to the point of making it the primary today. :grinning:
Scott Leibrand
@scottleibrand
Feb 17 22:09
@PieterGit installing the packaged version makes it possible to run things like oref0-setup without specifying the full path. Not sure if we still need that or not.
PieterGit
@PieterGit
Feb 17 22:12
but it will install several programs we are not using anymore, see https://github.com/openaps/oref0/pull/1036/files#diff-b9cfc7f2cdf78a7f4b91a753d10865a2
we're always calling /root/src/oref0/bin/oref0-setup.sh as far as I know, because we had troubles with the oref0-setup call with upgrades in the past
Jon Cluck
@cluckj
Feb 17 22:27
(^ that's how cas and I caught the linking issue, he used oref0-setup instead of calling the full path)
PieterGit
@PieterGit
Feb 17 22:39

any ideas how to debug:

# oref0-mmtune
mmtune: WW
2019/02/17 23:37:46 cannot connect to CC111x radio on /dev/spidev5.1
2019/02/17 23:37:46 cc111x: no response

I tried reflashing the firmware (and even installed mraa again). flashing the firmware shows the leds, but mmtune / Go libs can't seem to find the freshly installed TI chip

# ls -al /dev/spidev5.1
crw------- 1 root root 153, 0 Feb 17 23:33 /dev/spidev5.1
That's good for an explorer board . Isn't it?
Eric
@ecc1
Feb 17 23:03
You're sure you flashed with the SPI version of subg_rfspy firmware (not UART)?
Eric
@ecc1
Feb 17 23:05
that's because it's not seeing the firmware version string it expects from the firmware on the cc1110
PieterGit
@PieterGit
Feb 17 23:07
Eric
@ecc1
Feb 17 23:10
probably ERD's version. If you use @ps2 's, you have to checkout the v1.0 version or earlier from his repo
PieterGit
@PieterGit
Feb 17 23:11
ok, i used spi1_alt2_EDISON_EXPLORER_US_STDLOC.hex from ERD's version
did I just blow up a explorer board?
# ./ccprog -p 19,7,36 erase
Using pins: DC=19, DD=7, RESET=36
Erasing chip.
This code is only tested on CC1110. Unsupported chip id = 0x00.
Chip erase failed.
# ./ccprog -p 19,7,36 write spi1_alt2_EDISON_EXPLORER_US_STDLOC.hex
Using pins: DC=19, DD=7, RESET=36
This code is only tested on CC1110. Unsupported chip id = 0x00.
PieterGit
@PieterGit
Feb 17 23:16
after a reboot of the rig it looks better...
# ./ccprog -p 19,7,36 erase
Using pins: DC=19, DD=7, RESET=36
Erasing chip.
Chip erased.
# ./ccprog -p 19,7,36 write spi1_alt2_EDISON_EXPLORER_US_STDLOC.hex
Using pins: DC=19, DD=7, RESET=36
Intel Hex file: spi1_alt2_EDISON_EXPLORER_US_STDLOC.hex
Writing 2769 bytes to flash....
but it still doesn't mmtune
It's a Explorer Block v1.2 from 2017. Will debug again later this week. thanks for the help so far. I checked pump.ini and medtronic_freq. all look ok to me.
Jon Cluck
@cluckj
Feb 17 23:20
does it respond to any of the subg_rfspy tools?
Eric
@ecc1
Feb 17 23:20
There's a low-level command in the cc111x/cmd/probe Go directory you can build and run with "./probe" to check that it's seeing the radio chip OK.
tepidjuice
@tepidjuice
Feb 17 23:43
Will running ./oref0-runagain.sh run the npm global-install command needed for the dev branch? I've always manually updated my dev rig wondered if it is automated now?
Jon Cluck
@cluckj
Feb 17 23:48
runagain.sh isn't currently working on the dev branch