Revamp Fortran interface

[robertodr]

New year, new API. I started this PR thinking to rewrite the Fortran interface using iso_c_binding but it spiraled out of control and I rewrote the C API to the library too. There are breaking changes, but I plan to write a migration guide. With these changes there is no longer a dependency on the C compiler. I don't plan to document the API in this PR. I have #106 in progress for that purpose and I don't want the two branches to conflict too much.

This will close #61 and #60

To potential reviewers Please let's review and get in ~#109~ #111 first.

Todos

• [X] BREAKING Rewrite Fortran interface using iso_c_binding
• [X] Added API functions xcfun_which_vars and xcfun_which_mode.
• [X] BREAKING All API functions are uniformly namespaced with the xcfun_ prefix.
• [X] BREAKING API functions xc_serialize, xc_deserialize, xc_set_fromstring, and xc_derivative_index.
• [X] Example of C host code linking to XCFun provided.
• [X] Example of Fortran host code linking to XCFun provided.
• [x] Write migration guide to use new API

Types of changes

• [ ] Bug fix (non-breaking change which fixes an issue)
• [ ] New feature (non-breaking change which adds functionality)
• [x] Breaking change (fix or feature that would cause existing functionality to change)

Status

• [x] Ready to go

[codecov-io]

Codecov Report

Merging #98 into master will increase coverage by 0.44%. The diff coverage is 68.75%.

@@            Coverage Diff             @@
##           master      #98      +/-   ##
==========================================
+ Coverage   52.54%   52.99%   +0.44%     
==========================================
  Files          78       78              
  Lines        1922     1970      +48     
==========================================
+ Hits         1010     1044      +34     
- Misses        912      926      +14

Impacted Files	Coverage Δ
src/specmath.hpp	100% <ø> (ø)	:arrow_up:
src/config.hpp	0% <0%> (ø)
test/testall.cpp	91.97% <100%> (ø)
src/densvars.hpp	48.88% <60%> (ø)	:arrow_up:
src/xcint.cpp	90.24% <60%> (+9.8%)	:arrow_up:
src/XCFunctional.cpp	65.65% <65.65%> (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 3cc5e13...6e9b5da. Read the comment docs.

[robertodr]

This is ready for review. Please squash-merge The history is messy, not all intermediate commits build cleanly, and the changes in each commit are not atomic anyways.

[chjacob-tubs]

I will try to test and review this tomorrow!

[robertodr]

I will try to test and review this tomorrow!

Thank you! The C++ and Python interfaces look quite silly and contrived at the moment. I have decoupled the C++ implementation from the C API, which allows to do more sophisticated stuff C++-side. However, the binding still goes through a flat C API.

[bast]

Much praise and respect for this. I will test this within LSDalton/OpenRSP/XCint. Probably not quite tomorrow (travel) but this week.

[chjacob-tubs]

I will try to test and review this tomorrow!

Thank you! The C++ and Python interfaces look quite silly and contrived at the moment. I have decoupled the C++ implementation from the C API, which allows to do more sophisticated stuff C++-side. However, the binding still goes through a flat C API.

Maybe open an issue to also simplify the Python interface? I would be happy to work on that, but will wait for the examples first

[bast]

Much praise and respect for this. I will test this within LSDalton/OpenRSP/XCint. Probably not quite tomorrow (travel) but this week.

I am testing this tonight ...

[bast]

Great migration doc! I am getting segfaults when called from XCint but most probably my fault. I need to look at it with fresh eyes.

This confused me a bit (outdated example): https://github.com/robertodr/xcfun/blob/re-fortran/examples/CXX_host/example.cpp

[robertodr]

I will try to test and review this tomorrow!

Thank you! The C++ and Python interfaces look quite silly and contrived at the moment. I have decoupled the C++ implementation from the C API, which allows to do more sophisticated stuff C++-side. However, the binding still goes through a flat C API.

Maybe open an issue to also simplify the Python interface? I would be happy to work on that, but will wait for the examples first

I have opened #112 with a longer explanation.

[robertodr]

Great migration doc! I am getting segfaults when called from XCint but most probably my fault. I need to look at it with fresh eyes.

This confused me a bit (outdated example): https://github.com/robertodr/xcfun/blob/re-fortran/examples/CXX_host/example.cpp

I have fixed the example (was a bit sloppy with the rebase yesterday) Let me know about those segfaults :crossed_fingers:

[bast]

Thanks I will restart testing. I am 100% sure the segfaults will turn out to be my fault. I will report very soon.