Revert expect_mapequal() changes

[hadley]

Reverts #2150 since it does actually break quite a few packages on CRAN. This change reverts the original behaviour (with more tests) making use of more new helpers.

Fixes #2245

[claude]

Claude finished @hadley's task —— View job

---

Code Review Analysis

Reviewing PR #2255 that reverts expect_mapequal() changes from #2150 due to CRAN package breakage.

Todo List:

• Read and analyze changed files
• Review the reversion logic in R/expect-setequal.R
• Check test coverage and snapshots
• Identify any potential bugs or issues
• Provide final review feedback

Review Findings

The reversion looks correct and addresses the CRAN breakage issue appropriately:

Positive aspects:

• expect_mapequal() now properly handles edition differences (lines 110-117 in R/expect-setequal.R)
• Edition 2 behavior is preserved using expect_equal() for backward compatibility
• Edition 3+ uses the newer expect_waldo_equal_() approach
• Comprehensive test coverage including edge cases (empty vectors, validation, etc.)
• Good input validation with check_map_names() helper function
• Proper error messages with clear descriptions

Minor observations:

• The warning message for empty vectors (object and expected are empty vectors.) uses backticks inconsistently with other error messages that use proper formatting
• The code correctly preserves the original behavior that caused CRAN issues while maintaining the improvements

Technical correctness:

• Edition-based behavior switching is implemented correctly
• Name validation logic is sound
• Test snapshots cover all expected failure cases
• The reversion maintains API compatibility

LGTM - No significant issues found. The reversion appropriately addresses the CRAN breakage while maintaining test coverage and proper edition handling.

---