Move some initialization/finalization functions and eigenvalue-related functions to C++ (!1105) · Merge Requests · Shikhar Kumar / openmc

Move some initialization/finalization functions and eigenvalue-related functions to C++

Created by: paulromano

This PR moves many initialization/finalization functions to C++, including openmc_init, openmc_finalize, openmc_simulation_init, and openmc_simulation_finalize as well as some functions related to calculating k-effective. At this point, eigenvalue.F90 is almost completely converted. openmc_reset and openmc_hard_reset are also on the C++ side too. There's not too much else to say about this -- it was a pretty straightforward translation of code. There's a net reduction of about 1k lines of Fortran code and we're now down to 32.7k (version 0.10.0 was at about 50k).

I've also fixed two bugs in this PR:

In my conversion of synchronize_bank to C++, the temporary bank (temp_sites) was allocated on the stack. Because it can potentially be quite large, it really needs to be on the heap in order to avoid a stack overflow. I've now changed it to a std::vector so that it appears on the heap.
One version of read_attribute made an invalid use of strlen that is now fixed.

*Created by: paulromano*

This PR moves many initialization/finalization functions to C++, including `openmc_init`, `openmc_finalize`, `openmc_simulation_init`, and `openmc_simulation_finalize` as well as some functions related to calculating k-effective. At this point, `eigenvalue.F90` is almost completely converted. `openmc_reset` and `openmc_hard_reset` are also on the C++ side too. There's not too much else to say about this -- it was a pretty straightforward translation of code. There's a net reduction of about 1k lines of Fortran code and we're now down to 32.7k (version 0.10.0 was at about 50k).

I've also fixed two bugs in this PR:

1. In my conversion of `synchronize_bank` to C++, the temporary bank (`temp_sites`) was allocated on the stack. Because it can potentially be quite large, it really needs to be on the heap in order to avoid a stack overflow. I've now changed it to a `std::vector` so that it appears on the heap.
2. One version of `read_attribute` made an invalid use of `strlen` that is now fixed.