Cwa-app-ios: Number of risk encounters increasing although the number should be declining

[zopyx]

iOS 14.3 beta 4
CWA 1.5

CWA on 21.10.2020

CWA on 22.10.2020

Why is the counter of risk encounters increasing although no new encounters happened since 8 or 9 days?

---

Internal Tracking ID: EXPOSUREAPP-3407

[Ein-Tim]

@zopyx
Since you were one of the Beta Users who had the problem with iOS 14.2 Beta 3, I would assume that because of this 7 (?) days the App didn't download new keys (=was not working correctly) something in the counting went wrong.
So or so could you please share your Exposure Log?

[zopyx]

How can I share the Exposure Log?

[Ein-Tim]

(In German because I saw that the Language from CWA is German):

Please go into Settings, then click "Begegnungsmitteilungen" -> "Aktiv" -> "Begegnungsüberprüfungen" -> scroll all the way down to the bottom -> "Überprüfungen exportieren"

[zopyx]

https://www.dropbox.com/s/qd57z4v1sj5x9dz/ExposureChecks-2020-10-22.json?dl=0

[Ein-Tim]

Okay so from the Exposure Log I can tell that the 9 Encounters are right and the ENF recorded them.
Why the App don't show the correct number of days since the Encounter, I do not know.
My assumption is that this is an Issue because of the not working iOS 14.2 Public Beta 3.
Maybe @thomasaugsten can take a deeper look into this.

[thomasaugsten]

Days since Encounter is coming directly from the API. We are not sure how they exact calculate this. If they calculate from encounter time than this maybe can happen. But they can also change the calculation during the betas which is also very likely.

[zopyx]

Not a real-world problem for me, I know how to interpret the data :)

[ndegendogo]

Another possible explanation:
it was indeed 9 encounters on that day. But they reported their infection / uploaded their DEK on different days.
So their DEK are in different bundles and you downloaded them on different days. On the day of the first match you knew only part of the truth.
The message says: days since the (last) encounter. Not: days since download, or since match.

[heinezen]

Hello everyone,

Thanks for all the feedback. This issue has been mirrored to the internal Jira with ticket ID EXPOSUREAPP-3407. The devs have been notified and will now investigate the problem. Status updates will be posted in this thread.

Best Regards,
CH

---

Corona-Warn-App Open Source Team

[daimpi]

@zopyx
Ok. I looked at your data and you're right there seems not to be any mystery: it looks like your last encounter (not necessarily a red one) was on Oct 13th (from the info displayed in CWA). But some ppl you encountered before that only uploaded their diagnosis keys (DKs) on the 21st so they were available for you to download on the 22nd.

Regarding

Not a real-world problem for me, I know how to interpret the data :)

In your OP you wrote:

Why is the counter of risk encounters increasing although no new encounters happened since 8 or 9 days?

As there is no mystery why this happens what is your question and/or proposal?

[dsarkar]

Dear @zopyx, dear community,

Thank you all very much for your contributions.
Our understanding is now that this is not a bug. Within the next few days, we will publish in the blog on the website https://www.coronawarn.app an explanation regarding this issue. Please, stand by and we will let you know, once the explanations are published. Then, @zopyx, if applicable you might close this issue.

Best wishes,
DS

Corona-Warn-App Open Source Team

[bufemc]

It seems the CWA or the underlying API is counting incorrectly the days since the last (c19-positive) contact:

This might be related or not at all. I know a person who got on Thursday 29th Oct this:

But today, on 2nd Nov this:

"For visual checking":
[TH] FR SA SU [MO]

Why is it on Thursday 7 days, but on Monday, that's 4 days after, just 9 days. It should be 11 days, or if we take into account that only full 24 hours are counted at least 10 days, but not 9. The person even joked whether the CWA app just counts working days ;)

IFF this is another bug feel free to move it..

[daimpi]

@bufemc can you share this persons EN log here (just rename the ending from .json to .txt) then we can take a closer look.

Btw: I would expect to see "10 days since the last exposure" (not 11) on the second screen b/c of this problem: https://github.com/corona-warn-app/cwa-documentation/issues/433

[bufemc]

I gave the instructions as above:

Please go into Settings, then click "Begegnungsmitteilungen" -> "Aktiv" -> "Begegnungsüberprüfungen" -> scroll all the way down to the bottom -> "Überprüfungen exportieren"

However I wonder why it seems there is no MatchCount other than 0? Am I blind? The app shows 2 infections 9..11 days before.

This is the full JSON output:

exposure_log.txt

[daimpi]

@bufemc

However I wonder why it seems there is no MatchCount other than 0?
Build" : "18A393

Your friend is unfortunately still using iOS 14.0.1 which is affected by https://github.com/corona-warn-app/cwa-app-ios/issues/1106, that's why there are no matches in the log. That is o/c unfortunate, b/c even if they update right now to iOS 14.1, matches will not appear retroactively and only new checks will show the correct match count. Could you ask them to update and share a screenshot + log once the checks have been performed after the update to iOS 14.1?

[heinezen]

@bufemc I put your JSON output in a file because it was filling up most of the issue page. Hope that's okay for you.

Regards,
CH

---

Corona-Warn-App Open Source Team

[bufemc]

Yeah that's fine. As wished the person did upgrade iOS now.. and I will just put the MatchCounts >= 1 here:

       {
          "Hash" : "7493205AA9537DD485CEB19B33C9346E6F4473733C2BC8A25EF6284E93DA073D",
          "MatchCount" : 1,
          "KeyCount" : 26166,
          "AppBundleIdentifier" : "de.rki.coronawarnapp",
          "Timestamp" : "2020-11-03 17:50:40 +0100"
        },

        {
          "Hash" : "064B723E5EDD2EB8CD4C36FAA67FE839A486451748A5521B447606795867A061",
          "MatchCount" : 1,
          "KeyCount" : 18645,
          "AppBundleIdentifier" : "de.rki.coronawarnapp",
          "Timestamp" : "2020-11-03 17:50:38 +0100"
        },

BTW, will only the underlying API know when the contagious persons were met?

And the CWA app shows 11 days now:

[Ein-Tim]

BTW, will only the underlying API know when the contagious persons were met?

@bufemc, yes only the ENF (Exposure Notification Framework) knows this.

[daimpi]

@bufemc your screenshot from today is consistent with the screenshot from Nov 02, both indicate that the last encounter happened on Oct 23. The only deviation is the first screenshot from Oct 29 which indicates that the encounter already happened on Oct 22. It's unclear to me where this discrepancy comes from tbh 🤷.

BTW, will only the underlying API know when the contagious persons were met?

As @Ein-Tim already mentioned above: all the matching of keys and risk calculation is currently done within the Exposure Notification Framework (ENF) and the result is then exposed to CWA via the API. The output CWA gets from ENF looks like this (CWA currently only queries ExposureSummary).

[ndegendogo]

@bufemc once a day the ENF API is called to perform the matching procedure. The result (number of matches, days since last exposure) are then displayed as-is in your risk status.
These two numbers stay the same till the next check procedure. If you look closely, your screenshot of 29 Oct indicates that the checking procedure was performed yesterday. So, actually, yesterday at time of check it was 9 days since the exposure.

I agree that this is a bit confusing.
Also I am not sure how exactly ENF calculates / rounds these days (I have asked this question somewhere but yet have to find the link to my question; answer not given so far).

[Ein-Tim]

There is a similar Issue open in the Documentation Repo: https://github.com/corona-warn-app/cwa-documentation/issues/433
(@ndegendogo in this Issue you also asked

how exactly ENF calculates / rounds these days)

---

@dsarkar You stated a week ago that there will be a blog post Regarding this. When will the post get published?

[ndegendogo]

@Ein-Tim Thanks for the link 😀

Actually, I expect that the answer to my additional question may take a while, because maybe they have to ask G/A

[ndegendogo]

@Ein-Tim and I am always amazed how fast you find related tickets and you have an excellent overview over the tickets
👍👍👍

[thomasaugsten]

Because of the update to ENF v2 we can close this