Steps to build a CSDMS Workbench component #18

Here are the steps to build a component for the CSDMS Workbench from a library containing a model or a tool.

To help explain these steps, consider a library foo packaged in a Python module foo that exposes a class Foo.

1. Wrap the library with a Basic Model Interface (BMI). For the case of the foo library, create a class BmiFoo and include in the foo module or in a separate bmi_foo module.
   • Learn more about BMI from its documentation
   • See examples of wrapping a simple model with a BMI in C, C++, Fortran, and Python
   • See examples of wrapping data sources with soilgrids, nwis, and gridmet
2. [Optional, but recommended] Create a conda-forge package for the BMI library. For the foo library, the conda-forge package could be foo or bmi-foo.
   • The conda-forge documentation on contributing packages is essential reading
   • Examples of conda-forge packages that wrap a model or data source with a BMI include hydrotrend, permamodel, gipl, and gridmet_bmi
3. Run the babelizer over the BMI library to create a component that can be imported into pymt. Babelizing the BmiFoo class would produce the pymt_foo module exposing the Foo class.
   • Learn more about the babelizer from its documentation, and about pymt from its documentation
   • Examples of components include pymt_hydrotrend, pymt_permamodel, pymt_gipl, and pymt_gridmet
4. [Optional, but recommended] Create a conda-forge package for the pymt component. For example, pymt_foo.
   • As mentioned above, the conda-forge documentation on contributing packages is essential reading
   • Examples of conda-forge packages of CSDMS Workbench components include pymt_hydrotrend, pymt_permamodel, and pymt_gipl

@mcflugen @gantian127 @gregtucker: Please feel free edit these steps as you see fit, and add any discussion as comments below!