[SebKrantz]

No description provided.

[SebKrantz]

@claude please debug this and fix with a PR to development.

[claude]

Claude finished @SebKrantz's task in 38s —— View job

---

Diagnosis done, but I couldn't push the fix

• Found the failing CI check (test-coverage)
• Found the cause
• Fixed it locally
• Push and open a PR to development: blocked

Cause: the one failure is in the new test, not in the C fix for #883.

Failure ('test-fmedian.R:702:3'): fmedian with weights is invariant to weight scaling (#883)
Expected `fmedian(x, w = w * 1000, g = rep(1L, 6))` to equal 35.
names for target but not for current

A grouped call returns a named result by default (use.g.names = TRUE). The test compared it to an unnamed 35. The numeric values match, and the other 22,877 tests pass.

Fix: add use.g.names = FALSE to that call at tests/testthat/test-fmedian.R:702. The earlier test at lines 690–691 already does this.

expect_equal(fmedian(x, w = w * 1000, g = rep(1L, 6), use.g.names = FALSE), 35)

Why it isn't pushed:

• Pushing to development was rejected because the branch is protected.
• My attempt to create a new branch was blocked by a permission prompt, so there is no branch or PR yet.
• The fix is committed locally in the runner only.
• I haven't run the tests. The fix is based on the CI log.

You can apply the one-line change above directly. Or allow git checkout -b in my allowed tools and re-trigger me, and I'll push a branch and open the PR.