search

reichlab / covidHubUtils

Utility functions for the COVID-19 forecast hub
http://reichlab.io/covidHubUtils/
17 stars 16 forks source link

Cleanup installs happening in the github workflows #409

Closed bsweger closed 7 months ago

bsweger commented 8 months ago

resolves #404, resolves #405

The scope of these commits is making sure that all required dependencies are in the package's DESCRIPTION instead of being explicitly installed as a workflow step (for example, DT is required to build the vignettes).

Additionally, there's some cleanup to remove workflow-based installs if they aren't being used and/or reducing the footprint (e.g., explicitly use ggplot2 in the vignette rather than pull in the entire tidyverse).

Longer term, we have a goal to make these workflows better align to the newer ones used in the hubverse (#406 )

note: there are several smaller commits so it's easier to revert any breaking changes...happy to squash after approval to tidy up the commit history

nickreich commented 8 months ago

@bsweger I'm assuming you will ping me when this is ready for a full review, and am holding off on a review until then.

bsweger commented 8 months ago

@nickreich ready for review!

bsweger commented 8 months ago

A bigger point is that although the github action definition updates looked plausible to me, I don't know of a way to test that they work without actually merging the changes in and giving them a run... do you?

@elray1 r_cmd_check.yaml and pr_unittest.yaml run when a PR opens, so I made sure they worked before opening the PR for review (however, the tests started failing after I committed your suggested changes, so I'll look at that).

is pkgdown.yaml something we can run on a PR and then again on merge to main?

I recall that github actions have some capacity for ad-hoc runs, but I haven't looked into it yet.

bsweger commented 7 months ago

@elray1 I believe I found and fixed the issue that was causing the tests to fail (that said, I'm not sure why they were passing before merging in your requested changes (which seem unrelated..and I reverted them to just see if the tests would start passing again, and they did! :tableflip:)

Anyway, everything is green now, and I squashed the commits so the history is a bit cleaner. Can you re-review when you have a chance? Thanks!