Replication package: Implementation of Clean Architecture in the Development of a Course Scheduling Application

The paper compares the course scheduling application of the Faculty of Engineering, Universitas Riau, before (O1) and after (O2) it was restructured from a Laravel MVC layout into a modular monolith with Clean Architecture layers. This package holds the data behind the numbers in the paper and a script that recomputes them. Running python3 scripts/summarize.py (Python 3, standard library only) reads the raw tool outputs, recomputes 118 values and compares each with the value as printed. These include the paper's tables, the figures quoted in its text, and the figures of the submitted version that were later corrected. All 118 match. The output is written to results/paper-tables.md. Contents- source/O1, source/O2: the two measured source snapshots (O2 is commit 5d3cebc), without vendor/- config/: SonarQube project settings for both snapshots and the class list used for the sensitivity analysis- data/: PHPMetrics 2.9.1 JSON, SonarQube measures and full issue lists for O1 and O2, and the JUnit and Clover files of the posttest test run- history/: the PHPMetrics export of the intermediate iteration (It. 1), whose source state was not kept, together with earlier tool outputs and the SonarQube analysis history- scripts/: summarize.py and the scripts used to rerun PHPMetrics, SonarQube and PHPUnit- README.md, PROVENANCE.md, LICENSE-NOTICE.md, CITATION.cff, CHECKSUMS.sha256 ScopeBoth tools analysed the whole app/ directory at O1 and O2. A genetic scheduling component written by a collaborator in an earlier study is part of every main result (71 classes at O1, 122 at O2). Its O1 class has a cyclomatic complexity of 69, so results without it are also given (70 and 115 classes). CorrectionsWhile preparing this material we found that the submitted version of the paper compared measurements taken over different scopes. The pretest SonarQube analysis covered templates, configuration, routes and migrations in addition to app/. The posttest PHPMetrics report left out the genetic component and ten Livewire components. The posttest SonarQube analysis and test run came from an earlier commit than the PHPMetrics posttest, and two of its findings had been marked as false positives by hand. The camera-ready paper uses one scope and one posttest snapshot with no manual triage. PROVENANCE.md describes each difference and keeps the superseded values. Not includedvendor/ and node_modules/, .env files, SonarQube tokens, and the Excel import templates, which contain names and staff numbers of real lecturers. LicenceScripts and configuration are under MIT; data and documentation are under CC BY 4.0. The source snapshots are included only so that the measurements can be verified. Rights remain with their authors, and any other use needs their permission (see LICENSE-NOTICE.md). ToolsSonarQube Community Build 25.10.0.114319 (Sonar way profile), PHPMetrics 2.9.1, PHPUnit 10.5.48 with Xdebug 3.5.1 on PHP 8.2.4, Laravel 10.48.