Closed roigcarlo closed 8 years ago
Is that really necessary?
I currently think that the names of the modelling engines do not belong to the CUBA keywords ...
Or did you mean that there should be CUBA.MODELLING_ENGINE which can store the name?
Or did you mean that there should be CUBA.MODELLING_ENGINE which can store the name
Yes, exactly. We need CUBA keys to represent engines, visualizers and their corresponding parameters in order to have an independent and inter-operable CUDS. The goal is to abstract away wrappers from users. They would be instantiated by SimPhoNy based on the metadata information present inside the CUDS.
@kemattil However, this needs to be discussed more. In order to instantiate a wrapper SimPhoNy would need to know the wrappers name, i.e. the user provided name should match exactly one wrapper's name.
Exactly.
So the user should be able to query the names of all available wrappers (+ short description) in order to assign, e.g., the CUBA.MODELLING_ENGINE variable with a correct string.
But in general the idea is ok.
In any case, wrappers should register themselves to SimPhoNy in a consistent way. Each wrapper which is available at simphony.engine
entry point, should either implement something like a variable filled with its wrappers or a method like get_wrappers
which does the same thing when called. This way, we can make sure that SimPhoNy knows how to find the corresponding wrappers.
Knowing and documenting the exact engine used to produce data from one simulation is an important point. The engines name and the version of the engine (release data etc), should be part of the metadata. One can then either define a simple key for the engine, e.g.: CUBA.LAMMPS_$VERSIONSTRING or define a CUBA.ENGINE as a child of CUDS component (i.e., it will have uuid, name, description, and additional attributes as CUBA.VERSION).
I just made a PR which adds new keywords for different software tools, please review and comment.
@ahashibon @ahashibon @kemattil
Please review the latest changes to the https://github.com/simphony/simphony-metadata/pull/14.
apart from some comments, all seems good to me.
Done as part of #14.
We need new CUBA keywords to represent the different modeling Engines (OpenFoam, nCad, ...)