The design decision behind the new main branch was among other things to differentiate between the interface of a model and its implementation such that you don't need to have a model included in the repo to actually use it. Therefore, we didn't want to have a large collection of models in the src directory. This is great for development, since a developer of a new model does not need to open a PR in order to have a model that they can use with a solver that has been written for fenics-constitutive.
However, at some point it would be useful to have a collection of finished models that can just be installed and imported in python. Currently there are some models in the examples directory which are all tested, but you can't from models import MisesPlasticity.
I still think it is a good idea to have the strong seperation between the interfaces and the actual models, so should we just open a new repo for a model collection or try to have two packages in the same repository (Monorepo example)?
The design decision behind the new main branch was among other things to differentiate between the interface of a model and its implementation such that you don't need to have a model included in the repo to actually use it. Therefore, we didn't want to have a large collection of models in the
srcdirectory. This is great for development, since a developer of a new model does not need to open a PR in order to have a model that they can use with a solver that has been written forfenics-constitutive.However, at some point it would be useful to have a collection of finished models that can just be installed and imported in python. Currently there are some models in the examples directory which are all tested, but you can't
from models import MisesPlasticity.I still think it is a good idea to have the strong seperation between the interfaces and the actual models, so should we just open a new repo for a model collection or try to have two packages in the same repository (Monorepo example)?