upstream the results code

[jordens]

Looks clean, generic, and mature. Is ARTIQ-specific but experiment-agnostic.
Any chance to get a PR for it?

(maybe without artiq_results_path())

[dnadlinger]

From my perspective, the main utility of this code is actually the artiq_results_path machinery, i.e. that you can just load results from a Jupyter notebook/script/… on any of our group's machines without having to worry about which paths to use.

But yes, just the directory listing/filtering aspect might also be useful. I'm not too sure about cleanliness (there are inconsistencies such as cls vs. class_name), but YMMV.

Where would you want to put this?

[jordens]

artiq_results_path seemed very specific with the environment vars and all.

Maybe protocols or a new user facing scripting module?

[dnadlinger]

artiq_results_path seemed very specific with the environment vars and all.

It is; I just wanted to point out that much utility of this in practice stems from not having to hardcode paths everywhere, rather than not typing out 2019-12-12/21/000….h5.

If you are interested in this, we should probably move the result tree walking code to whatever new data analysis scripting module, and reimplement oitg.results on top of that.

[jordens]

Agreed