Progress in refactoring code and implementing unit tests

[kousuke-nakano]

What I am doing right now are:

• Adding intent(in, out, inout) statements,
• Adding use only statements, and
• Implementing unit tests for the updated codes if possible.

In this pull request, I have changed:

• Updated test_zsktri.f90
• Updated test_zsktrs.f90
• Updated upwinv.f90
• Updated upwinvp.f90
• Introduced new tests
  • test_zsktri.f90
  • test_zsktrs.f90
  • test_upwinvp.f90
  • test_upwinvp_complex.f90

#12

[addman2]

There is a question if zsktr[si] should go to pfapack part. Then they should be rewritten to fixed from fortran.

[kousuke-nakano]

Thanks @addman2!

zsktr[si] should go to pfapack part. Then they should be rewritten to fixed from fortran.

Yes, they are subroutines related to Pfaffian. So we can move them. Did you find corresponding subroutines in pfapack?

[addman2]

Dear @kousuke-nakano,

well the original pfapack does not contain z double-complex skew-symmetric solver and invertor. But according to the license we can modify pfapack. So there is no problem to add zsktr* from m_common to m_pfapack where it by nature belongs.

However by our linting rule it should be rewritten to fixed-form fortran. But let's do one step at a time. I would merge this PR, create an issue about it and leave it to next PR. That is my suggestion.

[kousuke-nakano]

well the original pfapack does not contain z double-complex skew-symmetric solver and invertor. But according to the license we can modify pfapack. So there is no problem to add zsktr* from m_common to m_pfapack where it by nature belongs.

However by our linting rule it should be rewritten to fixed-form fortran. But let's do one step at a time. I would merge this PR, create an issue about it and leave it to next PR. That is my suggestion.

Sure! Let's fix this problem later.