Raise Infection from no-op to enforced signal and cover newly exposed mutation gaps

[Copilot]

The Infection job was effectively non-blocking and under-configured: CI accepted any MSI, and PHPStan integration was memory-constrained enough to blur real signal. This PR tightens the mutation gate to the current measured baseline and adds targeted regression coverage for escaped mutants surfaced by the stricter run.

• Infection policy

  • move MSI enforcement into /home/runner/work/Stringy/Stringy/infection.json.dist
  • set:
    • minMsi: 87
    • minCoveredMsi: 87
  • add staticAnalysisToolOptions: "--memory-limit=512M" so PHPStan-backed mutation analysis remains usable under Infection

• CI behavior

  • stop overriding Infection with --min-msi=0 --min-covered-msi=0
  • keep CI aligned with the checked-in Infection config instead of a weaker workflow-specific invocation

• Regression coverage for escaped mutants

  • StaticStringyTest
    • assert invalid method message content
    • force refresh of the cached methodArgs map before validating __callStatic() argument counting
  • CollectionStringyTest
    • verify CollectionStringy::create() still enforces element type validation by default
  • StringyTest
    • cover full concatenation behavior in appendStringy()
    • cover default empty-value preservation in words()
    • cover default chunk size behavior in chunk() and chunkCollection()

• Net effect

  • Infection now acts as an actual gate
  • the stricter run exposed real missing assertions rather than configuration noise
  • the added tests lock in the currently recovered behavior while leaving remaining escaped mutants visible for follow-up tightening

{
  "staticAnalysisTool": "phpstan",
  "staticAnalysisToolOptions": "--memory-limit=512M",
  "minMsi": 87,
  "minCoveredMsi": 87
}

---

This change is 

[coveralls]

coverage: 97.891%. remained the same — copilot/integrate-infection-into-ci-pipeline into master

[voku]

/gemini review

[gemini-code-assist]

Code Review

This pull request introduces mutation testing configuration via infection.json.dist and expands the test suite to cover default behaviors for the create, appendStringy, words, and chunk methods. It also refines existing tests by adding exception message checks and ensuring cache resets during static analysis testing. Feedback was provided regarding the static analysis suppression in CollectionStringy.php, suggesting the addition of an explanatory marker for better clarity and consistency with the rest of the codebase.

[gemini-code-assist]

The use of a broad /** @phpstan-ignore-next-line */ without an explanatory comment is discouraged as it can hide unrelated static analysis issues on the same line. For consistency with other ignores in this file (e.g., lines 95 and 110), consider adding a brief explanation or a marker like | FP? to indicate why this suppression is necessary.

        /** @phpstan-ignore-next-line | FP? */

[voku]

@copilot PS: upfront my two cent, since version 2 of phpstan we can and should use suppress errors by identifier and not per line anymore! Now your task, check if the next llm generated feedback is valid or just AI slop: Code Review
This pull request introduces mutation testing configuration via infection.json.dist and expands the test suite to cover default behaviors for the create, appendStringy, words, and chunk methods. It also refines existing tests by adding exception message checks and ensuring cache resets during static analysis testing. Feedback was provided regarding the static analysis suppression in CollectionStringy.php, suggesting the addition of an explanatory marker for better clarity and consistency with the rest of the codebase.

src/CollectionStringy.php
string $iteratorClass = \Arrayy\ArrayyIterator::class,
bool $checkPropertiesInConstructor = true
) {
/** @phpstan-ignore-next-line */
@gemini-code-assist
gemini-code-assist Bot
19 minutes ago
medium

The use of a broad /** @phpstan-ignore-next-line */ without an explanatory comment is discouraged as it can hide unrelated static analysis issues on the same line. For consistency with other ignores in this file (e.g., lines 95 and 110), consider adding a brief explanation or a marker like | FP? to indicate why this suppression is necessary.

    /** @phpstan-ignore-next-line | FP? */

[Copilot]

@copilot PS: upfront my two cent, since version 2 of phpstan we can and should use suppress errors by identifier and not per line anymore! Now your task, check if the next llm generated feedba...

The feedback was valid. I reproduced the suppressed PHPStan error, confirmed the identifier is return.type, and replaced the broad suppression with an identifier-based one in 3b374b2.