Make co2 non-static #4260
Make co2 non-static #4260
Conversation
multitalentloes
force-pushed
the
make_co2_nonstatic
branch
from
10000c8
to
312559c
Compare
| @@ -0,0 +1,31 @@ | |||
|
|
|||
There was a problem hiding this comment.
Missing license and copyright header.
| #define OPM_CO2HELPER_STRUCTS_HPP | ||
|
|
||
| #include <vector> | ||
| struct TabulatedDensityTraits { |
There was a problem hiding this comment.
No namespace?
Also no whitespace (above the struct def.)...
| @@ -0,0 +1,31 @@ | |||
|
|
|||
| #ifndef OPM_CO2HELPER_STRUCTS_HPP | |||
| #define OPM_CO2HELPER_STRUCTS_HPP | |||
There was a problem hiding this comment.
No relation between:
- filename
- include guard macro name
- types defined in file
Please improve.
There was a problem hiding this comment.
updated headerguard to match filename, not sure what you want from the name of the structs
| { | ||
| public: | ||
| CO2Parameters(){ | ||
| co2Tables_ = CO2Tables(); |
There was a problem hiding this comment.
Nitpick: bad an d in con sistentformatting (2-8 spaces indent!), apply clang-format to new files as a rule of thumb.
More importantly: why a separate CO2Parameters class when CO2Tables could be used directly?
There was a problem hiding this comment.
All three new files have run clang-format now
There was a problem hiding this comment.
CO2Parameters now bypassed and using CO2Tables directly
| */ | ||
|
|
||
| #ifndef OPM_CO2PARAMETERS_HPP | ||
| #define OPM_CO2PARAMETERS_HPP |
There was a problem hiding this comment.
Macro and filename mismatch. I would also prefer a blank line after the macro define.
| module for the precise wording of the license and the list of | ||
| copyright holders. | ||
| */ | ||
| #include "CO2tables.inc" |
There was a problem hiding this comment.
I would prefer regular includes/paths, even for "private" ones that do not require redistribution for library use.
| /*! | ||
| * \brief The dynamic viscosity [Pa s] of CO2. | ||
| * | ||
| * This is only here so that the API is valid |
There was a problem hiding this comment.
Lots of these functions added, can we not just remove them? What would be the consequence?
There was a problem hiding this comment.
removing these functions causes compilation errors, seems related to the common interface described in "Components.hpp", althought I am not sure why it does not just inherit that implementation which just throws an error
There was a problem hiding this comment.
And having added new versions of these that also take in the parameter objects it is a bit ugly as inheriting the interface from components does not make it obvious that you are supposed to create a CO2Tables object and send that in instead, not sure how to clean this up nicely
| @@ -50,17 +50,19 @@ Co2GasPvt(const std::vector<Scalar>& salinity, | |||
| setActivityModelSalt(activityModel); | |||
| setThermalMixingModel(thermalMixingModel); | |||
|
|
|||
| co2Tables = Params(); | |||
There was a problem hiding this comment.
Will not this just do the entire table building over again, since it already would be default-initialized at this point?
Also in BrineCo2Pvt.cpp.
| * | ||
| * \return A const reference to the CO2 parameters. | ||
| */ | ||
| const Params& getCo2Tables() const |
There was a problem hiding this comment.
Is this just for debugging? Can it be removed? If not, perhaps also add it to the Co2GasPvt class.
| const float CO2<float>::brineSalinity; | ||
| template<> | ||
| const double CO2<double>::brineSalinity; | ||
| // template<> |
There was a problem hiding this comment.
I have just set the value inside the CO2 class, I dont think these lines are needed now
There was a problem hiding this comment.
I'll readd them iff the warnings are still there in jenkins
| @@ -305,7 +306,8 @@ class BrineCO2FluidSystem | |||
| } | |||
|
|
|||
| assert(phaseIdx == gasPhaseIdx); | |||
| LhsEval result = CO2::gasViscosity(temperature, pressure); | |||
| CO2Tables params; // TODO: avoid this initialization | |||
There was a problem hiding this comment.
I agree with the TODO...
We must find a solution to this (all of the similar TODO places here) before it can be merged.
There was a problem hiding this comment.
Resolved with a static singleton. This will not cause problems for the GPUification of the code as Flow does not use this fluid system regardless.
multitalentloes
force-pushed
the
make_co2_nonstatic
branch
from
e1cd51f
to
5880346
Compare
multitalentloes
force-pushed
the
make_co2_nonstatic
branch
from
dd10037
to
fbec2db
Compare
|
Works with new tables generated by OPM/opm-utilities#87 |
|
jenkins build this please |
Static classes, in particular static buffers are causing some troubles in the GPU porting of OPM. This PR seeks to avoid the usage of a static implementation of the CO2 class.
WIP