Add proper deprecation warnings for deprecated module attributes

[Carreau]

Summary

This PR improves deprecation handling across the traitlets codebase by implementing proper DeprecationWarning emissions for deprecated APIs instead of silent aliases. This follows PEP 562 for module-level __getattr__ and ensures users are properly notified when using deprecated features.

Key Changes

• traitlets/traitlets.py: Replaced silent alias NoDefaultSpecified = Undefined with a module-level __getattr__ implementation that emits a DeprecationWarning when the deprecated name is accessed. Added __dir__() to keep deprecated attributes discoverable.

• traitlets/config/configurable.py: Added explicit DeprecationWarning to Configurable._get_log_handler() method with updated docstring indicating the deprecation version and recommended alternative.

• traitlets/config/loader.py: Converted Config.has_key from a simple alias to a proper method that emits a DeprecationWarning before delegating to __contains__().

• tests/test_traitlets.py: Added test_no_default_specified_deprecated() to verify the deprecation warning is emitted and that the deprecated name remains discoverable via dir().

• tests/config/test_loader.py: Added test_has_key_deprecated() to verify Config.has_key() emits appropriate deprecation warnings.

• tests/config/test_configurable.py: Added test_get_log_handler_deprecated() to verify _get_log_handler() emits a deprecation warning.

Implementation Details

The changes use Python's standard warnings.warn() with DeprecationWarning and stacklevel=2 to ensure warnings point to the caller's code. The module-level __getattr__ approach (PEP 562) allows deprecated names to be accessed with warnings while keeping the module namespace clean and maintaining discoverability through __dir__().

https://claude.ai/code/session_01AEeaWWBdmyp9x2DaCombXD

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 93.19%. Comparing base (1dae189) to head (7f791be).

Components	Coverage Δ
traitlets	85.60% <100.00%> (+0.10%)	⬆️
tests	99.07% <100.00%> (+<0.01%)	⬆️

☔ 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.

[Carreau]

The two red downstream checks (downstream / jupyter_server and the downstream_check gate that aggregates it) are not caused by this PR, and can't be fixed from this repo.

Single failure, in jupyter_server's own suite:

FAILED tests/test_utils.py::test_check_version - packaging.version.InvalidVersion: Invalid version: 1.0
1 failed, 1121 passed, 32 skipped

Root cause is a packaging behaviour change. test_check_version deliberately passes a float:

assert check_version(1.0, "1.0.1")  # type:ignore[arg-type]

and jupyter_server's check_version only catches TypeError:

try:
    return bool(Version(v) >= Version(check))
except TypeError:
    return True

Older packaging raised TypeError from the regex on a non-str; packaging >= 26 raises InvalidVersion instead, which isn't caught. Reproduced locally against unmodified traitlets:

packaging	Version(1.0)	test
24.0	TypeError (caught)	passes
26.3	InvalidVersion: Invalid version: 1.0	fails

Nothing in this PR touches version parsing — the diff only adds DeprecationWarnings to three already-deprecated APIs. The last downstream run on main was 2026-08-03, before packaging 26 landed, which is why main isn't showing this yet; it should go red on its next run too.

The fix belongs in jupyter_server (catch InvalidVersion as well, or stop passing a float). Every other check on this PR is green, including Test Lint.

---

Generated by Claude Code

[Carreau]

Failures will be fixed by jupyter-server/jupyter_server#1688