Should the cell outputs of notebooks be cleared prior to publishing?

[norlandrhagen]

Wondering what the best practice is for notebooks within cookbooks. Is it generally preferred to clear the output of all cells or to push the notebooks with the rendered content included?

[brian-rose]

Yes, that is our standard practice, for two reasons:

• keeps the source repo lightweight and simplifies version control
• (probably more important) ensures that the published output is always rendered by our build system as a "guarantee" of reproducibility.

However we're trying to avoid being too dogmatic about any of this, and there are exceptions to the rule. Some cookbooks have heavyweight content that isn't well suited to executing on our CI system. See for example the CESM LENS cookbook. You can see in its config file that one notebook is flagged for exclusion from executing. This one notebook was run elsewhere and committed with its output.

[norlandrhagen]

Thanks for the detailed explanation! I'll be sure to clear the cell outputs.