A (set of) testing source(s)?

[jonkeane]

When working on the {arrowbench} integration, I found myself wanting to use either a repo with (small) local files already pointed to in datalogistik, or to be able to provide a repo file + where those files are locally in {arrowbench}. Then use those for testing.

The way we did this in benchmarks and {arrowbench} before was basically that: having a set of testing files that came with the packages to use.

What I've done for {arrowbench} for now is to basically hand-craft the cache entires, and then copy that over to a temp cache for testing:

https://github.com/voltrondata-labs/arrowbench/blob/54ace647121349d489a50533580880849b351868/tests/testthat/setup.R#L5-L18

This isn't ideal (though not awful...). Should datalogistik come with that temp | testing cache? Or have a way to go from repo file + the small data files on disk to a cache?