Fix function erroneously declared as static

[thomas-moulard]

Functions declared in headers should never be static.
In the previous implementation, statisticsDataToString was duplicated
in each compilation unit as the function was both defined in the header
and declared as static.

[mm318]

How about make it a member function of StatisticData?

[codecov]

Codecov Report

Merging #37 into master will not change coverage.
The diff coverage is 16.66%.

@@           Coverage Diff           @@
##           master      #37   +/-   ##
=======================================
  Coverage   42.04%   42.04%           
=======================================
  Files          26       27    +1     
  Lines         880      880           
  Branches      506      506           
=======================================
  Hits          370      370           
  Misses         50       50           
  Partials      460      460

Flag	Coverage Δ
#unittests	42.04% <16.66%> (ø)	⬆️

Impacted Files	Coverage Δ
..._collector/src/moving_average_statistics/types.hpp	100% <ø> (+71.42%)	⬆️
..._collector/src/moving_average_statistics/types.cpp	16.66% <16.66%> (ø)
...ollector/linux_process_memory_measurement_node.hpp	100% <0%> (ø)	⬆️
...ollector/linux_process_memory_measurement_node.cpp	33.33% <0%> (ø)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update cf9bb2f...f79f702. Read the comment docs.

[thomas-moulard]

It's generally better to avoid fixing bugs and refactoring in the same PR. I don't have a strong opinion one way or another (except that a header called types.hpp should not have functions defined). My goal here was to try to get rid of the partial hit on the function. I thought it was because of the issue with static but it seems I was wrong. We can still merge this I guess as this is a step in the right direction (?)

edit: that being said I just dropped the spuriousstd::to_string in this function, so whatever works for me.

edit2: the tests relies on a char-for-char diff of the string so I can't get rid of std::to_string, back to only moving that into a cpp file.

[mm318]

The .hpp file should not be here.

[thomas-moulard]

https://stackoverflow.com/questions/36174499/why-add-header-files-into-add-library-add-executable-command-in-cmake

Some people prefer that style because then you get the headers in the auto-generated IDE solutions.

[mm318]

I don't think that's the intention here, otherwise this is a consistency issue, because why would this be the only header file under add_library()?

[thomas-moulard]

I don't understand why the partial hit is still there - this is why the coverage tests is failing. Merging as I don't know how long it will take to understand the root cause here.