Let's use a scripting language for a comfortable C ++ life C ++ implementation after verification with python

I've been prototyping in Python today to see if the algorithm works.

1. Write your ideas in Python.
2. Graph the results so that you can see what is going on in one figure. (Spyder integrated environment)
3. Make sure it works with multiple datasets.
4. Make sure that another dataset works.
5. Standardize the writing style with pylint.
6. Write a documentation comment on the function.
7. Be aware of the distinction between the basic part of the algorithm and auxiliary functions such as graphing.
8. Make the basic part of the algorithm a separate module.
9. Create a class if needed.
10. Complete design, implementation, documentation comments, and confirmation of execution results in python.
11. Clarify what needs to be implemented in C ++.
12. Start development by referring to the python implementation for the part to be implemented in C ++.
13. If there is C ++ of the module used in python, write it using it (example cv2.imread () → cv :: imread ()).
14. Make it work even if it is solid.
15. Implement and modularize (or classify) as it should be.
16. Implement it as a command line application and test it under multiple conditions.
17. Add const qualifier to possible arguments. Make data members private, use namespaces, etc.
18. Write doc comments for use with Doxygen.
19. Separate into the source code of the library to be incorporated as a module, the header file, and the source code of the main function for the executable file for stand-alone.
20. Confine functions that do not need to be exposed in an anonymous namespace.
21. Review the method name, function name, and variable name of the argument so that they are easy to understand as documents, and rename them if necessary.
22. Write unit tests suitable for your intended use.
23. For that unit test, improve the module until it passes the test.
24. Unit tests describe not only normal systems but also abnormal systems.
25. Incorporate the module after the test is completed into the target program.

In this way, you can verify the effectiveness earlier and then develop than developing in C ++ alone.

Note: If you want to speed up the process

--Manage script version control. --Perform a regression test each time you make a revision to confirm that the calculation results have not deteriorated. --Prepare and test a sufficient number of test samples to confirm that no deterioration has occurred. --Measure the processing time as a Python module. --Do not explicitly write a double loop in the calculation. Use the appropriate function. ――Write a comment that gives a hint as to why the process is being processed, and check the algorithm. --Find and simplify the parts that can be simplified by changing the order of operations. --Reduce the waste of calculation. --Confirm that the processing time is shortened when the waste of calculation is reduced. --Reread the script, keeping in mind the group of meanings that are easy to understand.

I am aware of these.