[jackielii]

Follow-ups to #34, which was squash-merged before these landed. Found by the eos agent and an adversarial review.

Fixes

• Dead branch removed (fa8ca2f). Registrations are keyed by their dynamic type, so the exact-interface-key lookup in getArg could never hit. The doc line promising "exact registration wins" for interfaces is corrected.
• Regression fixed: a registered pointer to an interface fills it again.
  • Before feat: fill interface parameters from the one registered implementation #34, var s Svc = &A{}; WithArgs(&s) filled Props(s Svc) through the pointer-to-value rule.
  • feat: fill interface parameters from the one registered implementation #34 broke this: *Svc doesn't implement Svc, so it returned "not found". Worse, if an implementation was also registered, the parameter silently got that implementation instead.
  • getArg now looks for a registered *Iface first. That also gives a way to choose one implementation when several are registered, without touching the handler.
  • New test: TestInjection_pointerToInterfaceIsUsedFirst.

Docs (review findings)

• The Mount ambiguity check covers Props and extended ServeHTTP only. Init/Middlewares fail Mount anyway, and Page/component parameters fail at request time.
• The error names every implementation, not "both".
• Known edge, documented rather than worked around: the check knows the writer only as http.ResponseWriter.
  • Example: Props(sw io.StringWriter) with two StringWriters registered fails Mount, although the real writer would fill it.
  • Workaround: declare the writer as http.ResponseWriter or io.Writer.

Left as is (intended or contrived)

• An any parameter beyond the request values now takes a lone registration, and with two registrations it fails Mount. It used to be "not found".
• A custom selector returning a nil RenderTarget isn't modelled. A broken page can pass Mount; a working one is never rejected.

go test ./..., golangci-lint and gopls are clean.

https://claude.ai/code/session_01Dp3Jmu2ePyCME2yG9yd672

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 95.21%. Comparing base (f2bc9b2) to head (9ebfc1d).

Additional details and impacted files

@@           Coverage Diff           @@
##             main      #35   +/-   ##
=======================================
  Coverage   95.21%   95.21%           
=======================================
  Files          16       16           
  Lines        1547     1547           
=======================================
  Hits         1473     1473           
  Misses         44       44           
  Partials       30       30           

☔ 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.