feat(stats): calculate packet rates

[sergeymatov]

No description provided.

[qmonnet]

I don't have the full context here, and I didn't review in-depth - but assuming the PR is ready for review, it seems to me like this misses unit tests to validate the processing, in particular to validate the Smooth implementation.

[qmonnet]

Looks good overall, although I haven't been as far as checking the smoothing algo. Thanks for adding the unit tests!

I've got some minor nitpicks, they should all be trivial to address.

I'd welcome a bit more of context in the commit descriptions - I don't like empty commit descriptions. In something like “No more double-counting for packets”, for example, I'd love to find an explanation of the problem we solve: how comes that we had double-counting, and how do we solve it? Or in “smoothing instead of derivative”, it would be nice to mention why we want to switch from the latter to the former. It is more work, but it makes life so much easier when someone looks at the Git history later to try to understand a change.

[qmonnet]

What's SG, I suppose it's “Savitzky-Golay”? Please expand the first time you use it in the comments.

Suggested change

	// Push the cumulative snapshot into the SG derivative filter
	// Push the cumulative snapshot into the Savitzky-Golay (SG) derivative filter

[sergeymatov]

Here as well, this comment is not exists in latest revision of the file :)

[qmonnet]

Here, it doesn't matter, I'd still like you to expand the acronym the first time you use it in the file, whether it's this comment or another. There definitely are occurrences of this acronym in your latest revision.

[sergeymatov]

@qmonnet thanks for the review. In general we are saving Savitsky-Golay derivative calculation cause in particular it's should give same results using different approximation formula. The reason we picked smoothing instead - during tests it showed more stable results, so there is no algorithm decisions was made but based on practical usage.

[qmonnet]

Thanks for the context ❤️. Don't hesitate to add it to the commit descriptions next time, this is where I search for it first when I want to understand a code change.

[qmonnet]

I'd still like the SG acronym to be expanded once, but this is not a blocking issue. PR seems good from my side otherwise. Thanks!