nooooo, not Diasend basals again

[jebeck]

Something funky is happening where the data is not "continous" (according to the basalutil) even though it looks continuous, results in the lack of a basal:bolus ratio. Placeholder for @jebeck to look into it.

[kentquirk]

I love the GIF. :)

On Tue, May 6, 2014 at 9:30 PM, Jana Beck notifications@github.com wrote:

Something funky is happening where the data is not "continous" (according to the basalutil) even though it looks continuous, results in the lack of a basal:bolus ratio. Placeholder for @jebeck https://github.com/jebeck to look into it.

[image: oh no not again]https://cloud.githubusercontent.com/assets/1588547/2897679/211f8570-d587-11e3-80c8-99340dd0954f.gif

— Reply to this email directly or view it on GitHubhttps://github.com/tidepool-org/hub/issues/87 .

Kent Quirk VP of Engineering, Tidepool

Tidepool is an open source, not-for-profit effort to build an open data platform and better applications to reduce the burden of Type 1 Diabetes.

[cheddar]

@jebeck fwiw, I ran into more problems yesterday trying to handle suspends properly. I'm going to re-write segmentutil to generate the data structure that we spoke about before and then just have a conversion layer from that structure into what it is currently pushing out.

So, if your investigation turns up that something is funky in segmentutil, it will hopefully be un-funkified with these changes.

[jebeck]

OK @cheddar, good to know. It's possible this is actually something weird in basalutil (to do with the test that was recently failing after we removed squashing), but I still do have to investigate.

[jebeck]

Looking at my own data, this may or may not be related, but there are regular gaps (this pattern of a missing segment from 2-6 am occurs over and over again in my data from April 19th back in time):

@cheddar At first I thought this was related to new segmentutil, but then I was looking at it on current blip-devel to verify, and it's the same there. I suspect it's a regression for Diasend data that happened with the changes for suspends. And I suspect, but do not know for certain, that it's the same problem that I opened this issue with, except that the missing segment that occurs there is much smaller.

Given that this results in incorrect display of the actuals stream of basal data, I'd like to elevate to P1.

[cheddar]

Where are you suspecting the regression is? In the parsers or in segmentutil? Have you looked at the actual data being returned from the tidepool platform to verify that the basal rate actually does exist?

[jebeck]

Short answer: yes, verified. Long answer via e-mail because heaps of PHI.