This draft PEP proposes adding optional timestamps to Python exception
objects that can be displayed in tracebacks. Based on CPython PR #129337
by Gregory P. Smith.
The feature adds a __timestamp_ns__ attribute to BaseException and allows
configuration via environment variables or command-line options. This is
particularly useful for debugging async applications with exception groups
and distributed systems.
🤖 Generated with Claude Code (https://claude.ai/code)
Co-Authored-By: Claude <noreply@anthropic.com>

Added two important open issues to the PEP:
1. Whether to always collect timestamps unconditionally (performance
 testing shows it's cheap) and the implications for pickle size and
 cross-version compatibility testing
2. Configuration of control flow exception exclusions (StopIteration,
 AsyncStopIteration) and whether subclasses should be included, noting
 the performance constraints of the hot path
🤖 Generated with Claude Code (https://claude.ai/code)
Co-Authored-By: Claude <noreply@anthropic.com>

Added two new sections to address questions raised at core team sprint:
1. "Why Exception Groups Need Timestamps" - Explains that while exception
 groups are conceptually unrelated, in practice they have important
 temporal relationships for debugging causality, performance analysis,
 and correlation with external observability tools.
2. "Why Not Use .add_note() When Catching?" - Details six key drawbacks
 of using add_note() instead: not all exceptions are caught, timing
 accuracy issues, inconsistent application, performance overhead,
 complexity burden, and loss of original timing information.
Key insight: When an exception occurs is intrinsic, immutable information
that should be captured at the source, not added later by consumers.
🤖 Generated with Claude Code (https://claude.ai/code)
Co-Authored-By: Claude <noreply@anthropic.com>

... need
Based on feedback from @picnixz about Sphinx's Skip exception and other
projects using exceptions for control flow:
1. Renamed "Special Cases" to "Control Flow Exceptions" for clarity
2. Added acknowledgment that other projects need this configurability
3. Deferred specific API design to open issues
4. Updated Open Issue python#5 with key challenges:
 - API design options (env var vs Python API)
 - Performance constraints in the hot path
 - Subclass handling complexity
 - Whether to expand the default exclusion list
The PEP now acknowledges the need for configurability while leaving the
specific implementation approach as an open question requiring careful
API design and performance testing.
🤖 Generated with Claude Code (https://claude.ai/code)
Co-Authored-By: Claude <noreply@anthropic.com>