Added creating solver interface #35
Conversation
musica-fortran/src/micm/MICM_mod.F90
Outdated
| subroutine delete_micm_polymorph(this) | ||
| implicit none | ||
| class(micm) :: this | ||
| call delete_micm_c(this%ptr) | ||
| end subroutine |
There was a problem hiding this comment.
My understanding is that yes, we need this. I also had the same question and without it, it doesn't work. It needs to declare micm as class instead of type to be able to have access to its members. I found this answer useful to understand.
There was a problem hiding this comment.
I updated it to class and removed type version.
boulderdaze
changed the title
boulderdaze
force-pushed
the
adding_solver
branch
2 times, most recently
from
f159a5e to
80f0858
Compare
musica-fortran/src/micm/micm_mod.F90
Outdated
| private | ||
| type(c_ptr) :: ptr | ||
| contains | ||
| procedure :: delete => delete_micm |
There was a problem hiding this comment.
@mattldawson is this a time when finalize could/should be used?
There was a problem hiding this comment.
yeah, I think so - if this function is always called right before the object is deleted, it could be:
| procedure :: delete => delete_micm | |
| final :: delete_micm |
There was a problem hiding this comment.
the function itself will have to be modified to take a type(micm_t) instead of a class(micm_t) with intent(inout), but other than that should be mostly the same (I think). This final function gets called automatically when a micm_t object goes out of scope
There was a problem hiding this comment.
Would it be okay to leave the way it is then? It needs to be class(micm_t) for micm_t to have access to its members.
There was a problem hiding this comment.
I think you should still have access to the data members, you just would need to change it to:
subroutine delete_micm(this)
type(micm_t), intent(inout) :: this
call delete_micm_c(this%ptr)
end subroutine
I think the only difference between class(foo_t) and type(foo_t) is that class(foo_t) will accept types that extend foo_t, but type(foo_t) won't, and that type-bound procedures have to take the polymorphic class argument, but final routines need to take the concrete type argument.
There was a problem hiding this comment.
Ah, thank you for correcting me. I fixed it
musica-fortran/src/micm/micm_mod.F90
Outdated
|
|
||
| contains | ||
| function create_micm(config_path) | ||
| implicit none |
There was a problem hiding this comment.
Please place implicit none at the top of this on the line before private and remove implicit none in all of these functions
There was a problem hiding this comment.
Yes, it's done now.
musica/src/micm/micm.cpp
Outdated
| return success; | ||
| } | ||
|
|
||
| void MICM::delete_solver() const |
There was a problem hiding this comment.
Please make this a destructor so that in delete_micm you can just call delete micm and not have to care about calling this function
There was a problem hiding this comment.
Yeah, I updated the destructor.
| void solve(double temperature, double pressure, double time_step, double*& concentrations, size_t num_concentrations); | ||
|
|
||
| private: | ||
| static constexpr size_t NUM_GRID_CELLS = 1; // TODO(jiwon) |
There was a problem hiding this comment.
we can set up a separate issue for supporting multiple grid cells
Completed