stgl/scarplet

Improve scarplet examples; add missing data files

mcflugen opened this issue · 4 comments

mcflugen commented

This issue is in regards to my JOSS review at openjournals/joss-reviews/issues/1066.

In running through the scarplet examples (here and here) I have a couple suggestions.

  1. From the instructions, it's not clear that the examples need to be run from inside the scarplet/tests folder.
  2. Not all of the data files are available in the repository (e.g. data/grandcanyon.tif from the Extracting Channels example)
  3. Provide the examples as notebooks users can download and run
  4. It would be nice if users could run the examples without having to grab the source code

In regards to the last point, I can think of two solutions

  1. simply indicate at the start of each example where to get the data files (a URL to files within the scarplet repository would be fine)
  2. include the data files with the package and have some way for users to load example data from within the installed package
fclubb commented

I agree with this: it wasn't immediately clear to me that the example data was included within the GitHub repository, as I had just done a pip install. It would be great to include a clearer statement of how users can get the example data (either state that they have to clone the GitHub repo, or provide the example data in a different way).

rmsare commented

Ah, this is unclear, I agree. I've updated the Readme and notebooks headers and made clear that examples and data live in the docs/source/examples/ directory over in #66

rmsare commented

It's still not great to download stuff from the repo. I'm adding a datasets submodule so users can load sample data included with the package a la sklearn.datasets.load_iris.

rmsare commented

Okay, the examples have been refactored and documented with more clarity.

I am going to close this and re-open if either reviewer does not have success with installation and testing using conda.