pytask icon indicating copy to clipboard operation
pytask copied to clipboard

ENH: Diffing commits in pytask project.

Open tobiasraabe opened this issue 3 years ago • 0 comments

Is your feature request related to a problem?

For finding bugs, detecting where changes come from, diffing commits in a pytask project can be extremely helpful.

One caveat is that pytask encourages a pattern where resource code in form of a package is developed in another repository and most of the time the error will happen in the interaction between this package and the research project which will obfuscate the source of the errors, but the implications should still be visible.

Describe the solution you'd like

Not solutions, but some starting points and ideas.

  • Use data diffing tools like https://github.com/dolthub/dolt
  • pytask diff which would run the project at two different commits and store the created content under two folders. Files with the same name could be diffed automatically for say numpy arrays, dataframes.

tobiasraabe avatar Jan 24 '22 10:01 tobiasraabe