[somethingwithproof]

This change avoids repeated RRDtool capability checks during one graph or update operation. It reuses each result within that call, including a negative result, and starts fresh on the next call. This removes repeated preparation work while preserving generated commands and avoiding a persistent cache.

Implementation details

RRDtool graph construction compares the same version four times, and update batches repeat the skip-past-updates capability check for every timestamp. Reuse each result within its function call, including a false result, and reset it on the next call. Empty update batches perform no version lookup.

The regression executes the production bodies with instrumented I/O, covers two RRD files, six timestamps, versions 1.3/1.4/1.5/1.8, and a return to 1.3. Update checks fall from six to one; graph checks fall from four to one. Commands and reference mutations remain identical.

PHP 8.5.11 through mise: 2,765 Unit tests pass with four existing skips; PHPStan, formatter, sink/hotspot guards and quick/empty poller regressions pass. Five alternating diagnostic rounds show lower median graph-option and batch-update costs, with unchanged allocated memory. Timing ranges overlap and include scheduling outliers; these measurements do not establish production throughput.

The same optimization is carried in #8012 and a minimal 1.2.x companion. Release notes remain in the final changelog PR #8320.

Maintenance companion: #8341. The rewrite #8012 includes this PR as an ancestor. Merge #8340 before #8012; a merge commit preserves that ancestry. If #8340 is squashed, refresh #8012 against the merged develop head before merging it.

Revision and checks

Description scope: target develop, 4 changed files, head 773d61d41. Checked 2026-10-23 03:06:04 UTC.

GitHub check snapshot: 2 passed. Local test figures and earlier CI runs above are recorded evidence for their stated revisions and environments; they do not assert that every check passes at this head.

[bmfmancini]

Re-reviewed the incremental commit ('test: measure coverage at the exact PR revision'): a single-line change pinning the changed-coverage checkout to ${{ github.event.pull_request.head.sha }} so coverage evidence always measures the exact reviewed PR revision even when later pushes race the run. All checks pass on the updated head (PHP lanes, changed-coverage, POST guard, docker probe; only Semgrep OSS neutral + unrelated skips). Production code unchanged from my prior review. Approving again.

[sonarqubecloud]

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
1.9% Duplication on New Code

See analysis details on SonarQube Cloud