fix(powermetrics): reject unsupported Macs and report 0 W when no samples

[davidberenstein1957]

Description

Two fixes to codecarbon/core/powermetrics.py. First, get_details() now guards against an empty match list before averaging, so it reports 0.0 W instead of NaN when powermetrics produced no CPU Power: / GPU Power: lines, and logs a warning pointing to the sudoers setup docs the first time there are no CPU samples; the previously-duplicated CPU and GPU branches are now one loop. Second, _setup_cli() now raises SystemError unless the CPU is Apple Silicon (the same is_mac_arm check used by resource_tracker), so an Intel Mac without Power Gadget falls back to TDP / cpu_load instead of selecting powermetrics, since Intel powermetrics output has no CPU Power: X mW line. The now-unreachable Intel-Mac powermetrics branch in resource_tracker._try_platform_cpu_backend is removed, and the command now uses self._cli and drops a stray "" argument. Master's Popen timeout and early-return in get_details are kept.

Related Issue

Fixes #1306

Motivation and Context

np.mean([]) returns nan and only emits a RuntimeWarning, so the surrounding except Exception never caught it. The NaN flowed into the tracker accumulators and, since NaN is absorbing under addition, made energy_consumed, emissions and emissions_rate NaN for the rest of the run, silently written to emissions.csv and POSTed to the API.

How Has This Been Tested?

New tests in tests/test_powermetrics.py cover empty logs (0 W, warning logged once), CPU-only logs, Intel / unknown CPU rejection, Apple Silicon setup, and the command built for Popen. test_resource_tracker.py now asserts that an Intel Mac does not use powermetrics. Full test-package passes.

Screenshots (if appropriate):

N/A

Types of changes

• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause existing functionality to change)

AI Usage Disclosure

• 🟥 AI-vibecoded: You cannot explain the logic. Car analogy : the car drive by itself, you are outside it and just tell it where to go.
• 🟠 AI-generated: Car analogy : the car drive by itself, you are inside and give instructions.
• ⭐ AI-assisted. Car analogy : you drive the car, AI help you find your way.
• ♻️ No AI used. Car analogy : you drive the car.

Checklist:

• My code follows the code style of this project.
• My change requires a change to the documentation.
• I have updated the documentation accordingly.
• I have read the docs/how-to/contributing.md document.
• I have added tests to cover my changes.
• All new and existing tests passed.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 91.71%. Comparing base (e5e46ab) to head (d537043).

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #1345      +/-   ##
==========================================
+ Coverage   91.70%   91.71%   +0.01%     
==========================================
  Files          49       49              
  Lines        5157     5156       -1     
==========================================
  Hits         4729     4729              
+ Misses        428      427       -1     

☔ 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.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[benoit-cty]

🤖 This review comment was written and posted by Claude Opus 5.5 (AI assistant), at the request of @benoit-cty. Findings were checked by reading the code and running tests locally (merged with current master where relevant), but please double-check before acting on them.

Verdict: 🔧 Request changes (rebase needed)

Both fixes are real:

• np.mean([]) gives NaN, which then poisons every total.
• On an Intel Mac without Power Gadget, master selects powermetrics. Intel output has no CPU Power: X mW line, so every sample is NaN.

Making _setup_cli raise SystemError for non-Apple CPUs correctly falls back to TDP/cpu_load. Removing the stray "" argument and using self._cli is good too.

Must fix:

1. Merge conflicts with master in codecarbon/core/powermetrics.py and tests/test_powermetrics.py.
   • Master commit 1ea5f93 rewrote _log_values to use Popen with a timeout, and added if not self._log_values(): return dict() in get_details.
   • The rebase must keep master's timeout and early return.
   • test_log_values_builds_clean_command mocks subprocess.call, which master no longer uses, so it needs rewriting against Popen.
2. No warning is logged, despite the description.
   • get_details falls back to 0.0 silently when there are no CPU samples.
   • In powermetrics mode there is no further fallback, so a missing sudoers rule or an empty log reports 0 W CPU for the whole run with nothing in the logs.
   • Please add a logger.warning, once, that points to the powermetrics/sudoers troubleshooting docs.

Nits:

• After this change, the Intel-Mac elif powermetrics.is_powermetrics_available() branch in core/resource_tracker.py (~L237) is dead code. Remove it or comment it.
• There are now two different Apple Silicon checks: startswith("Apple") here and is_mac_arm (\bM\d) in resource_tracker. Please use one helper.
• The title mentions rejecting unsupported Macs, but the description only covers the empty-sample fix. Please describe both.

[davidberenstein1957]

Made the changes in d537043: merged master (kept the Popen timeout and early return, rewrote the command test against Popen), added a one-time warning for empty CPU samples, switched to is_mac_arm, removed the dead Intel-Mac branch, and updated the description.