feat: make it possible to run tests in any directory, not only the current one

[maelle]

Seems useful to have before a CRAN release 😸

[maelle]

Now I understand why you made it so @krlmlr, it's somehow very hard to change without breaking the tests?! they still pass from devtools::test() but not in R CMD check.

[maelle]

in the end, the test setup was the culprit 🙃

[maelle]

From looking in testthat tests quickly, it seems that instead of callr I could use a silent reporter but it somehow is less clear for me? Or did I get too invested in the current "weird" setup?!

[codecov]

Codecov Report

Merging #24 (9920bcf) into main (34b8ba7) will increase coverage by 0.65%.
The diff coverage is 100.00%.

❗ Current head 9920bcf differs from pull request most recent head 6169e5d. Consider uploading reports for the commit 6169e5d to get more accurate results

@@            Coverage Diff             @@
##             main      #24      +/-   ##
==========================================
+ Coverage   77.50%   78.15%   +0.65%     
==========================================
  Files           3        3              
  Lines         120      119       -1     
==========================================
  Hits           93       93              
+ Misses         27       26       -1     

Impacted Files	Coverage Δ
R/lazytest_local.R	98.21% <100.00%> (+1.72%)	⬆️

[krlmlr]

No strong opinion, we can also leave the current behavior for the CRAN release.

[maelle]

But should this PR be merged?

[krlmlr]

The Windows tests keep failing.

I'm seeing similar failures with dev pkgload, and other failures with dev usethis: https://github.com/cynkra/lazytest/actions/runs/5019480976/jobs/9000043037 . These should be fixed before release, with or without the path feature. Can you please take a look?

[maelle]

I can't reproduce the pkgload issue, might have to explore next week.

[maelle]

I get a failure locally 🎉

[maelle]

still no luck exploring the pkgload failure.