1.8.9.1 DBG version accidentally deployed. Implications of installing corrected version?

[eyarzebinski]

The 1.8.9.1 dbg version was accidentally deployed in Bagamoyo to at least 3 of the 5 tablets within the last few weeks. I just alerted @JackMostow and he is wondering the implications of what happens to student progress if we change the version again to the correct 1.8.9.1 sw version. Is student progress completely wiped out and students who get the corrected version will have to start over and retake the placement test? Or does something else happen?

[eyarzebinski]

@judithodili you might be interested in this discussion, feel free to unsubscribe if not.

[kevindeland]

They will have to re-take the placement test.

[JackMostow]

Is it necessary to uninstall the current (dbg) version before reinstalling the sw version?

[kevindeland]

For their purposes, yes they will have to uninstall.

[JackMostow]

@eyarzebinski and @judithodili -

1. What should we do about the debug version running in at least 4 if not all 5 tablets in Bagamoyo?
   a. Exploit data from activities we might not get otherwise?
   b. Revert to the SW version to see the effect on the monthly test?

2. What should we do about the absence of any data yet from R52H1015Y3D and R52JA0LWRBM?
   a. Ask Mwita why?

SAMSUNG FIELD TESTING shows:

Bagamoyo PIXEL GLASS 5B02000273 running the Debug version
starting with PERF_RoboTutor_release_dbg.1.8.9.1_2018.06.11.17.56.39_5B02000273.json
and as recently as PERF_RoboTutor_release_dbg.1.8.9.1_2018.07.04.16.46.06_5B02000273.json

Bagamoyo PIXEL TREE 5B02000315 only running the release version, at least through PERF_RoboTutor_release_sw.1.8.9.1_2018.06.05.18.06.44_5B02000315.json
-- but the debug version didn't start till 2018.06.11, so we won't know till newer data uploads.

SAMSUNG - R52JA0N89EW = bagamoyo_1 running the Debug version
starting with RoboTutor_release_dbg.1.8.9.1_2018.06.11.17.43.04_52009f57fe7e943d.json
and as recently as PERF_RoboTutor_release_dbg.1.8.9.1_2018.07.02.17.55.59_52009f57fe7e943d.json

SAMSUNG - R52JA18WFEY = Bagamoyo_2 running the Debug version
starting with PERF_RoboTutor_release_dbg.1.8.9.1_2018.06.11.17.08.55_52009e8afef694ad.json
and as recently as PERF_RoboTutor_release_dbg.1.8.9.1_2018.07.02.17.51.12_52009e8afef694ad.json

SAMSUNG - R52JA18WXND = Bagamoyo_3 running the Debug version
starting with PERF_RoboTutor_release_dbg.1.8.9.1_2018.06.11.17.52.30_52000266fecfa4e5.json
and as recently as PERF_RoboTutor_release_dbg.1.8.9.1_2018.07.04.17.01.45_52000266fecfa4e5.json

In Mugeta:

SAMSUNG - R52H1015Y3D has no data, just ._icon from 2/10/2018.

SAMSUNG - R52JA18WZ1K's newest data is PERF_RoboTutor_release_sw.1.8.9.1_2018.06.13.15.05.44_5200f3feeab0a4e9.json

SAMSUNG - R52JA13ACET's newest data is PERF_RoboTutor_release_sw.1.8.9.1_2018.05.31.16.16.17_520002a4fa36a46b.json.

SAMSUNG - R52JA0LWRBM has no data, just ._icon from 2/10/2018.

SAMSUNG - R52J808XDLY's newest data is
PERF_RoboTutor_release_sw.1.8.9.1_2018.05.14.16.09.11_52031e8aec4574d9.json

SAMSUNG - R52JA1GETSJ's newest data is's newest data is
PERF_RoboTutor_release_sw.1.8.9.1_2018.06.18.15.46.33_520048c4fef2a46d.json

[judithodili]

I don’t think we can do much with insights from the kids accessing content that they are not cognitively ready for. What are the animal names of the tablets from Mugeta that are not syncing? I know Tembo is one of them... already contacted Mwita ... the sync folder was accidentally switched off. -- Regards, Judith Odili Uchidiuno www.judithu.com

[JackMostow]

For future reference, please see Tablet_Name_Mapping.xlsx:

Only one tablet in Mugeta sent no data:
SAMSUNG - R52JA0LWRBM | Mugeta_Tembo

SAMSUNG - R52H1015Y3D | Farasi is back at CMU. I moved its folder to IN-HOUSE TESTING to prevent future confusion.

At minimum we can get reliability information from kids who aren't ready for activities they attempt. So if kids are aiming too high, they'll test activities that other kids haven't gotten to, especially their feedback and scaffolding.

If we restrict attention to activities that kids perform satisfactorily, we should still get usable information, e.g. about step duration, though it may not be representative of better-prepared kids.

Options for the Bagamoyo tablets:

a. Revert to the non-debugger version.

b. Change to a better-instrumented version that's identical from the kid's viewpoint.
+: same benefits as a., plus additional information
?: can we do it without restarting the kid from scratch?

c. Change to a non-debugger version with new and modified activities to kid-test them.

Thoughts?

[judithodili]

I’m already working on getting the Tembo data. I’d vote for reverting the bagamoyo back to the non debugger version - I don’t think the debugger is comprehensible enough for the kids to make any informed decisions on what activities to try. -- Regards, Judith Odili Uchidiuno www.judithu.com

[JackMostow]

@kevindeland -

1. Can someone else, e.g. Octav, incorporate Shiven's code for logging crashes? If not, when can you?
2. What new or modified activities are ready for prime time? I'd like to deploy them to at least one tablet.

[JackMostow]

@kevindeland - How hard would it be to salvage the student models (i.e. position in each content area matrix) from the dbg verson of 1.8.9.1 to use in a functionally identical version that logs crashes, so that installing it doesn't make the kids start over?
Since the activities are the same, we don't have the issues of student models non-comparable across different versions.

[kevindeland]

Will be addressed in #414. Closing.