Changes in version 0.9.0 (2026-09-09)                  

  - Added formal software authorship metadata for Raúl Aguilar-Elena and
    Ana Delgado-García, including verified ORCID identifiers.

  - Corrected the ERBio method reference to DOI 10.5281/zenodo.22107465.

  - Distinguished the ERBio method reference from the open
    core-questionnaire and validation dataset
    DOI 10.5281/zenodo.22069658.

  - Consolidated bilingual Shiny, cross-language Excel import, and
    macOS/Windows header-normalization fixes.

  - Recovered and integrated the v0.8 independent EU regulatory audit
    layer.

  - Added the approved 1,008-item Spanish expert preventive-action
    registry.

  - Added the approved 1,008-item Spanish-English scientific translation
    registry.

  - Added language API (erbio_set_language(), erbio_get_language()) and
    bilingual questionnaire rendering.

  - Enriched failed controls with expert preventive actions while
    preserving deterministic scoring.

  - Added bilingual scientific invariance checks.

  - Technical/linguistic English approval is explicitly separated from
    psychometric validation.

                     Changes in version 0.7.0.9000                      

  - Adds a 509-row scientific source audit of the Spanish
    biological-agent registry against the official Annex II replacement
    in Order TES/1287/2021 (BOE-A-2021-19371).
  - Makes the 2021 official amending act the primary Spanish legal row
    source and retains consolidated RD 664/1997 as a secondary
    informative reference.
  - Corrects three SARS-CoV/SARS-CoV-2/MERS-CoV spacing artifacts
    inherited from consolidated HTML rendering using the official BOE
    PDF.
  - Rejects one false-positive BASEBiO linkage from Picornaviridae to
    Coronaviridae and reviews six other non-exact technical links with
    explicit scope/nomenclature states.
  - Keeps all independent EUR-Lex rowwise-verification flags false until
    the dedicated 509-row EU audit is completed.
  - Adds rowwise audit, source manifest and audit-summary assets.
  - Does not modify the frozen v0.4 deterministic ERBio calculation
    logic.

                     Changes in version 0.6.0.9000                      

  - Adds formal package and method citations, including the ERBio Zenodo
    DOI 10.5281/zenodo.22107465.
  - Adds a source manifest separating ERBio method provenance, Spanish
    legal classification, EU concordance and BASEBiO technical
    enrichment.
  - Adds a source-audit summary that explicitly records pending
    row-by-row EUR-Lex verification rather than overstating concordance.
  - Expands user documentation and executable examples by API layer.
  - Adds an explicit testthat expectation to the frozen v0.4 regression
    test so the suite is no longer reported as an empty/skipped test.
  - Updates the package development version to 0.6.0.9000.
  - Does not change the frozen v0.4 ERBio calculation logic.

                     Changes in version 0.5.0.9000                      

  - Converts frozen v0.4 into a formal R package layout.
  - Adds installed-package extdata discovery so users do not need to set
    erbio.data_dir.
  - Preserves the complete v0.4
    deterministic/questionnaire/agent/workplace engine.
  - Adds erbio_version().
  - Adapts the 38-block v0.4 regression suite to testthat.
  - Passes R CMD check with 0 errors, 0 warnings and 0 notes on
    R 4.6.1/macOS aarch64 under --as-cran (2026-08-25).

                         Changes in version 0.4                         

  - Workplace-level assessment.
  - Core + sector questionnaire validation.
  - Multi-agent handling without silent aggregation.
  - Preventive-plan generation and reproducibility bundle export.
  - 38/38 regression blocks passed on 2026-08-25.

Pre-CRAN candidate patch 2

  - Fixed a syntax-only parse error in
    tests/testthat/test-precran-metadata.R caused by an accidental
    standalone backslash.
  - No computational or methodological ERBioR code was changed.

Cross-language Shiny import candidate 3

  - Spanish and English Excel workbooks are accepted independently of
    the current UI language.
  - Workbook language is detected from metadata or stable
    item-id/question-text matching.
  - Responses are normalized to the same internal Spanish-core response
    codes before scientific scoring.
  - The selected UI language controls questionnaire presentation and
    approved preventive-action output.
  - Cross-language imports display an explicit input/output language
    notice.