feat: re-add simplify= to print.compare.loo (#366)

[LeonidasZhak]

Summary

This follow-up keeps the simplify restoration on print.compare.loo() and adds the missing focused coverage for the full-output path.

Thanks to maintainers

Thanks for the review and for pointing out the missing tests, scope cleanup, and AI disclosure requirement.

Issue or motivation

This PR follows up on #366 by restoring simplify = FALSE for print.compare.loo(), while keeping the default simplified output unchanged.

Root cause

The earlier branch restored the argument and print-path behavior, but it did not include focused test coverage for the full-output branch and it also carried an unrelated generated-file change.

Change

• restore simplify in print.compare.loo() with default TRUE
• document the restored full-output mode
• add focused tests for print(comp2, simplify = FALSE) and the full comparison columns
• keep the PR diff scoped by excluding the unrelated loo-package.Rd drift

Tests

• Rscript -e 'devtools::test(filter = "compare", stop_on_failure = TRUE)'
• R CMD build --no-build-vignettes .
• _R_CHECK_FORCE_SUGGESTS_=false R CMD check --no-manual --ignore-vignettes --no-build-vignettes loo_2.9.0.9000.tar.gz

I used AI assistance to help check for issues, refine the code, and run tests.

Scope

This follow-up does not change the default printed output, does not alter model-comparison calculations, and does not expand the feature beyond the existing print.compare.loo() path discussed here.

[codecov-commenter]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 92.72%. Comparing base (57aab3e) to head (806dcf7).

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #369      +/-   ##
==========================================
+ Coverage   92.70%   92.72%   +0.01%     
==========================================
  Files          31       31              
  Lines        3029     3037       +8     
==========================================
+ Hits         2808     2816       +8     
  Misses        221      221              

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[github-actions]

This is how benchmark results would change (along with a 95% confidence interval in relative change) if 806dcf7 is merged into master:

• ✔️loo_function: 1.95s -> 1.96s [-0.47%, +1.12%]
• 🚀loo_matrix: 1.83s -> 1.8s [-2.12%, -0.9%]
  Further explanation regarding interpretation and methodology can be found in the documentation.

[florence-bockting]

Thank you @LeonidasZhak for opening this PR. We had some discussion about whether to include/exclude this argument from loo_compare in this PR.
Therefore, we (@avehtari, @jgabry) need probably to discuss first how we want to handle this aspect and then we will come back to your PR.

[florence-bockting]

Thanks again @LeonidasZhak for the PR.
Please read our AI contribution policy and disclose any use of AI.
Additionally, please add tests for any changes you make to functions; stating "manual verification" is not sufficient.
Finally, please avoid making changes to files unrelated to this PR, as was done in loo-package.Rd.

[LeonidasZhak]

Thanks for the guidance. I pushed a follow-up commit that adds focused test coverage for print(comp2, simplify = FALSE), keeps the assertions narrow, and removes the unrelated loo-package.Rd drift from the PR diff.

I used AI assistance to help check for issues, refine the code, and run tests.

Local checks I ran before pushing:

• Rscript -e 'devtools::test(filter = "compare", stop_on_failure = TRUE)'\n- R CMD build --no-manual --no-build-vignettes .\n- _R_CHECK_FORCE_SUGGESTS_=false R CMD check --ignore-vignettes --no-manual loo_2.9.0.9000.tar.gz\n- git diff --check

[florence-bockting]

I will close this PR as it is not possible for me to differentiate this PR from other fully automated PRs, which are not welcome according to our AI contribution policy. Specifically, as this PR has been created in a batch of > 100 other PRs.

In future, I recommend that you write PR description and also comments in the PR with your own words. Furthermore, when proposing new features, it is helpful to propose your suggested solution in the respective Issue before creating a PR. This allows us to guide and support you better. Thank you very much.