Round scaled sensor values

[fisher-alice]

Now that the code-dot-org/johnny-five is updated to the latest release from upstream, this PR re-implements rounding scaled sensor values proposed as Change #1 in this investigation.

When I first investigated this last customization, I ran an App Lab program to see if sensor values were already displayed as integer values using the updated johnny-five package, and they are. In addition to this observation, I also noted that in lib/sensor.js within eventProcessing, once the median is computed, it is rounded and assigned to roundMedian at line 122 but is not rounded in johnny-five-deprecated. Thus, I thought that this third customization was not necessary.

However, after running the tests in PlaygroundComponentsTest.js in @code-dot-org, I saw that the setScale tests failed which helped me understand that the scaling occurs after the value is reached from the eventProcessing function.

I confirmed that this rounding is necessary with the following program.

Before the customization:

before-update-scaled.mp4

After the customization:

after-update-scaled.mp4

I looked into why the sensor values in the johnny-five-deprecated package are still displayed as integer values despite not having the median rounded. The comment above the function median explains that to reduce the noise in sensor readings, collected samples from the sensor are collected and then sorted. The value at the center is selected for the final value emitted and because there are many duplicate numbers, the result is almost always an integer even where there are an even number of values in the sample.
However, I am happy to see they added the Math.round just in case!

I posted a PR to update the johnny-five version to 2.1.0-cdo.1. However, I will first merge this change, and then update the version of the package.

Links

jira ticket - Re-add rounded scaled sensor values

Testing

I ran the @code-dot-org unit tests in files that import johnny-five. In particular, PlaygroundComponents.js had a handful of unit tests that specifically tested the setScale function.

[sanchitmalhotra126]

Just curious - do we have a standard way of tagging Code.org specific updates in this repo? (is it just adding a comment with // Code.org:?)

[fisher-alice]

Looking at prior PRs that added Code.org customizations, @islemaster seemed to consistently use this convention, as seen here.

[sanchitmalhotra126]

nit: round inline? Fn.constrain(Math.round(mapped), ...)

[fisher-alice]

Thanks! Just updated.

[bencodeorg]

LGTM -- does this mean that all code-dot-org circuit playground tests now pass while using your updated johnny-five fork? Or is "all code-dot-org circuit playground tests" not such an easy distinction to make?

[fisher-alice]

LGTM -- does this mean that all code-dot-org circuit playground tests now pass while using your updated johnny-five fork? Or is "all code-dot-org circuit playground tests" not such an easy distinction to make?

All code-dot-org circuit playground tests within code-dot-org/apps/test/unit/lib/kits/maker/boards/circuitPlayground now pass while using the updated johnny-five fork locally - there are 9 of them. Thanks!