Improvements

[ambmax00]

Putting up a list for things that would need to be changed and/or added.

I started to write this software when I barely had two years of experience in C++. For this reason, there are some poor design decisions that need to addressed.

Before adding new modules, or new functionalities, at least some of the issues should be solved.

Important:

• Add extensive testing for each module. As an inexperienced developer, I did not know the power of test-driven development...
• (Done) Fix the SAD guess in Docker container
• Add testing for important classes.
• Make the J/K/Z kernels and ADC routines external, i.e. they can be used as external routines in other quantum chemistry software.
• Work on the QR-DF method to make it faster.

Less important:

• Write a better MPI logger which uses a printf-type style.
• Do not make everything a shared pointer...
• Get rid of the macro magic for the builder classes. It was fun to write, but will be difficult to maintain
• Use constexpr instead of macros
• Write a better interface for dbcsr, especially for batch mode: Handling complicated series of tensor contractions nearly drove me insane.
• The "disk" mode in tensor batching is practically unexplored. It is working, but I have no idea how fast it is, as I never got around to test it.
• Get rid of exceptions and use error codes (Yes, I am that kind of person now)
• Make i/o less cumbersome. We have a json input, a hdf5 output, and the program also outputs to console...
• Make initialization less complicated. You have to take care of a million things before you can use any module in your program.

Not so important:

• Change the coding style. Probably wont be using clang-format in the future, and just write down the rules somewhere
• Fix ugly code, rename variables, and do not use any abbreviations except obvious ones (HF, MP2, ADC...)

More stuff might be added