Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 83.64%. Comparing base (fdcaafc) to head (e134a61).
⚠️ Report is 565 commits behind head on develop.

@@             Coverage Diff             @@
##           develop     #303      +/-   ##
===========================================
+ Coverage    76.48%   83.64%   +7.15%     
===========================================
  Files           22       23       +1     
  Lines         5576     6274     +698     
===========================================
+ Hits          4265     5248     +983     
+ Misses        1311     1026     -285     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

Is the goal of this fully replace pytest, or to have at least a set of tests for every API functions?

Yes, the goal is to replace them for the main C interface. We can run more complete checks with specific compiler flags, while the other api tests should test only that, api specific stuff, including memory allocation/deallocation to the language native objects.

If so, it will not be a small work.

Well, no. These are all functional (regression tests), so the only missing part is getting input data and compared data which are already present in the python database. That is not so difficult, I just need to actually sit down and do it.

However, these are not appropriate tests to rely on for the long-run, since it is hard to pin down where things are breaking and they are vulnerable to numerical fluctuations. The difficult work is to do proper unit (+ integration) testing, which requires going into each component and writing minimal tests. The framework for that is already in place it's just time and work needed to go through each one. I am not tackling this issue for now, but if I have some time to do #301 (#262) that will be unit tested as it evolves.