Changelog as evidence for a version window

Changelog research starts with an explicit version window: which source version and target version define the changes being studied. Read official release notes, migration guides, or changelog sections for every relevant transition rather than relying on the name of the new version.

The goal is not to rewrite the entire changelog, but to extract changes that could affect the affected code, configuration, runtime, build, integrations, or operations.

Classifying discovered changes

Change typeWhat to check in the repositoryRecording result
Breaking APIimports, calls, signatures, adaptersaffected symbol and focused check
Deprecation/removalusage, configuration, replacement pathversion + official section + action
Behavior changeoutputs, errors, ordering, retriesrepresentative scenario and contract impact
Configuration/runtimeenv, defaults, startup, deployassumption, environment check, owner
Unknowninaccessible or ambiguous sourceUnknown + specific next check

Minimal changelog finding record

CHANGELOG_RESEARCH.md — documented example
## Finding: <version and change>
- Source: <official URL or repository file:section>
- Affected area: <module, API, config, runtime>
- Change type: <breaking | deprecated | behavior | operational>
- Expected impact: <what may change>
- Confidence: <confirmed | needs verification | Unknown>
- Next check: <focused command or scenario>

Changelog research sequence

  1. Record the source version, target version, and search boundaries.
  2. Collect official release and changelog entries within the version window.
  3. Filter breaking, deprecated, behavior, configuration, and operational changes.
  4. Map each change to a specific repository area or mark it as unaffected.
  5. Add an evidence anchor, confidence, and the next focused verification.
  6. Stop if the target or relevant release evidence remains unknown.

Canonical lesson source · JavaRush

Open JavaRush material