Use correct units when parsing BolusWizard & CallBGForPH records

[SebastienLussier]

As discussed in issue #8, the processing of these record was previously hardcoded to a specific type of unit depending on the model. For the context, we own both a 722 & 522 and we are using mmol/L but these two models were hardcoded to mg/dl.

Tested on my 722, on which I issued consecutive boluses using the wizard, switching the unit in between.
"openaps use pump iter_pump" correctly displayed the different records in the unit in which they were entered.

[SebastienLussier]

This also fixes #26

[scottleibrand]

Can you add a note here with what testing (beyond what is in #26) someone could do to validate the before and after behavior of this change? Thanks.

[SebastienLussier]

For me, the most noticable effect was that meals/treatments weren't showing on the BG graph in NS.

A more practical way to validate the issue is to set the pump to the "incorrect" unit based on the opposite of what is found in "decocare/models/__init__.py". You then bolus using the bolus wizard and enter the BG (ex: 6.2 mmol/L). The treatment should show up in the "Treatments" report and will look like this (incorrect BG in bold):

Time	Event Type	Blood Glucose	Insulin	Carbs	Notes
19:11	Meal Bolus	62 (BolusWizard)	0.70	19	Normal bolus with wizard. Programmed bolus 0.7 Delivered bolus 0.7 Percent delivered: 100% Food estimate 0.8 Correction estimate -0.1 Bolus estimate 0.7 Target low 65 Target high 75 Hypothetical glucose delta -50.4 Glucose was: 62
19:10	BG Check	62 (Finger)

Repeating the same process with my changes should yield the following:

Time	Event Type	Blood Glucose	Insulin	Carbs	Notes
10:27	Meal Bolus	5.4 (BolusWizard)	1.60	35	Normal bolus with wizard. Programmed bolus 1.6 Delivered bolus 1.6 Percent delivered: 100% Food estimate 1.7 Correction estimate -0.1 Bolus estimate 1.6 Target low 6.5 Target high 7.5 Hypothetical glucose delta -14.719999999999999 Glucose was: 5.4
10:27	BG Check	5.4 (Finger)

Thanks!

[ghost]

@SebastienLussier as written in ITB I'll test the PR in about 1 week once return back from my vacation, thanks for your investigation so far!

[scottleibrand]

To make this easier to test, I just pushed a SebastienLussier-fixes branch (https://github.com/openaps/decocare/tree/SebastienLussier-fixes), so you can just clone decocare, git checkout SebastienLussier-fixes, and install it with sudo python setup.py develop.

[ghost]

@scottleibrand thanks for pushing to a separate branch

[ghost]

@SebastienLussier @scottleibrand the fix is working and solves the issue, thanks very much!!!
I entered 120 as BG in BolusExpert and it's shown correctly

[Missdazzle]

Also checked this on an European Veo 754 with mg/dl - putting in the BG together with carbs (BolusExpert).
The fix is working perfectly! Thank you!

[PieterGit]

@SebastienLussier: It works for me for Event Type Meal Bolus, but it still fails for Event Type BG Check.
I just did a reinstall with the SebastienLussier-fixes branch, because maybe i already had a dev branch with 0.0.32-dev installed. I'll check later to see if the reinstall fixed the problem. Is there anything that would explain it to work for Meal Bolus but not for BG Check?

@Missdazzle : a European Veo with mg/dl? You sure you don't mean mmol/l?

[Missdazzle]

@PieterGit Oh yes, I´m sure! :)
Is that "BG Check" something that has to do with the Enlite? In that case I cant´t test it.

[scottleibrand]

So should I merge this to dev for further testing there? Or is there anything we need to fix first?

[ghost]

For me it's okay, what's about the 6.4U fix? As I understood it is not merged to dev yet...

[scottleibrand]

The 6.4U/h fix is in decocare 0.0.31 (and in both the dev and master branches), which is available via easy_install -U decocare but not via pip.

[scottleibrand]

@PieterGit 👍 or 👎 for merge to dev?

[PieterGit]

@scottleibrand : please merge to dev. The fix fixes at least the Meal Bolus records. I think it's a rig specfic thing because I installed dev first, and will debug later and ask @ecc1 or @SebastienLussier later or open a new issue/PR.

[lsandini]

Sorry to pop up so late about this, but using the same rig with openaps 0.6.0 dev4, the bolus wizard events are detected and reported correctly using a 754WW pump, but display to problem described above when I use a 715WW or 722WW pump, all in mmol/l.

Do I understand correctly that it should have been fixed long ago ?

My right was built last may and many times updated. Shouldn't the decocare part be up to date as well ?

[PieterGit]

@lsandini Current oref0 0.6.0 release installs decocare-0.0.31-py2.7.egg and then a branch checkout of 0.1.0-dev branch, see https://github.com/openaps/oref0/blob/master/bin/oref0-setup.sh#L520-L538
That's because I think @scottleibrand had issues with releasing/publishing a new decocare release.

I noticed this patch is on the dev branch, see also https://github.com/openaps/decocare/network

Can you test on a test rig if using the dev branch of decocare fixes your problem with a 715WW and 722WW pump. If so, we should ask for a 0.1.1-dev branch, or change the oref0-setup.sh to use the dev branch.

That is, it would of course be better to create, test and release a fresh decocare release with all relevant fixes included.

@ecc1 and @scottleibrand and @bewest : thoughts?

[lsandini]

It took me some time, sorry. I had to retrieve my 722WW first.

Unfortunately the problem persists with 0.6.0 master. I didn't try the 715WW yet, but it is probably the same.