search

ImperialCollegeLondon / pyrealm

Development of the pyrealm package, providing an integrated toolbox for modelling plant productivity, growth and demography using Python.
https://pyrealm.readthedocs.io/
MIT License
23 stars 9 forks source link

Remove R code from repo or integrate its execution (with documentation). #60

Closed davidorme closed 1 year ago

davidorme commented 1 year ago

There are several "golden output" datasets currently embedded in the tests directory. These most logically belong in the pyrealm_build_data sdist package and indeed a number of bits of test data are already stored in there.

The code, inputs and outputs should probably moved into sensible directory structures in pyrealm_build_data along with better documentation and clear instructions on how to regenerate the outputs.

One issue here - I do not really want to make R a build dependency, so I'd be tempted to just simply document that developers will need to install R and the rpmodel package for that dataset.

davidorme commented 1 year ago

This is closely related to #24, which was aimed at using pytest-datadir to better organise data within the tests directory. However, given all the code involved with the golden output generation, I think it is probably better to use pyrealm_build_data and keep the tests directory clear of data generation code.