eyeprocess 0.11.1

eyeprocess 0.11.1

CRAN portability hotfix following the automated incoming pretest of 0.11.0.

Fixes

• capability-aware Arrow compression fallback when zstd is unavailable
• default fallback order: zstd → snappy → uncompressed
• explicit unavailable compression requests still fail clearly
• absolute GitHub URLs replace repository-relative README links
• package spelling allowlist for legitimate technical terminology

Validation

• full test suite: PASS
• simulated Debian Arrow build without zstd: PASS
• R CMD check --as-cran: 0 errors / 0 warnings / 1 expected NOTE
• actionable CRAN findings: zero
• exact tarball MD5: beb8e4d2892354d1fafafc711fd173f0
• 1182 exports
• NAMESPACE unchanged
• lifecycle registry unchanged
• canonical M4 Stan source unchanged

eyeprocess 0.11.0

eyeprocess 0.11.0

This release introduces the M4 latent response-process state layer and its associated identifiability, recovery, diagnostic, sensitivity, ablation, and validation infrastructure.

Highlights

• M4 latent process-state specification, simulation, fitting, diagnostics, PPC, ablation, sensitivity, negative controls, and recovery workflows
• explicit RT adjacent-gap state anchoring for state-label identification
• explicit REVIEW gating for trait-conditioned Markov M4 specifications
• auditable state uncertainty and process-information diagnostics
• 1182 governed public exports

Scientific governance

Trait-conditioned Markov M4 remains gated. Structural adequacy alone does not establish posterior identification; satisfactory R-hat/ESS, stable state occupancy across chains, and separated state emissions are required before substantive interpretation.

Validation

• exact source tarball: eyeprocess_0.11.0.tar.gz
• MD5: 22a80e1c248a98842423fa20688bd5d2
• R CMD check --as-cran: 0 errors / 0 warnings / 1 expected CRAN incoming-feasibility NOTE
• actionable CRAN findings: zero
• canonical M4 Stan MD5: c5af3e5d25ff63db42c58573eb42124b
• lifecycle registry preserved
• source executables: zero

eyeprocess 0.8.0

eyeprocess 0.8.0

eyeprocess 0.8.0 formalizes the validated 0.8 development programme for vendor-neutral eye/process data harmonization, response-process measurement, psychometric integration, and deployment governance.

Release highlights

• Process-measurement and deployment-governance infrastructure.
• Biometric/process pre-flight audits and exclusion manifests.
• Deployment-drift and temporal process-window infrastructure.
• Advanced AOI trajectory and pupillometry workflows.
• Visual-context/testlet IRT and multiblock process representations.
• Process-profile mixtures, external-validity tooling, and streaming scoring.
• Validation bundles, item seeding, presentation-sensitivity analyses, and process-decision features.
• Advanced sensitivity and Bayesian/3PL diagnostic infrastructure.
• Frontier estimators remain explicitly gated when an exact validated implementation is unavailable; simpler estimators are not silently substituted.

Validation

• Complete eyeprocess source test suite: PASS.
• Three expected mirt EM-cycle warnings only.
• CITATION.cff schema validation: PASS.
• pkgdown validation and isolated release-mode build: PASS.
• Definitive eyeprocess_0.8.0 source artifact built successfully.
• Definitive artifact R CMD check: 0 errors, 0 warnings, 0 notes.
• Isolated installed package version: 0.8.0.
• Installed citation dynamically reports version 0.8.0.
• Installed ordinary export surface: 812 / 812.
• Post-merge GitHub CI: 6 / 6 checks successful.

Release provenance

• Release-candidate commit: d84e94f
• Formal release commit: 4672c12
• Git tag: v0.8.0
• Release date: 2026-08-10
• Zenodo 0.8.0 version DOI: pending archive processing.
• Zenodo concept DOI: 10.5281/zenodo.21844472.

Scientific governance

Process-derived evidence remains measurement evidence and is not automatically interpreted as a mental-state, clinical, ability, misconduct, or other person-level diagnosis.

eyeprocess 0.6.0

eyeprocess 0.6.0

eyeprocess 0.6.0 introduces the measurement-intelligence programme for
vendor-neutral eye-tracking, pupillometry, biometric, and psychometric
process-data workflows.

Measurement uncertainty and AOI evidence

• Probabilistic AOI assignment and uncertainty propagation
• Compositional AOI analysis
• Calibration-drift detection and offline recalibration
• Process uncertainty budgets
• Reliability and Generalizability Theory
• Cross-device linking and equivalence auditing

Temporal, spatial, and pupil-process science

• Pupil phase-amplitude registration
• Informative-missingness and MNAR sensitivity analysis
• Recurrence and cross-recurrence analysis
• Experimental fixation point-process models
• Representative scanpaths
• Cognitive-episode segmentation

Decision intelligence and provenance

• Multi-objective item-bank optimization
• Process-DIF and fairness-drift analysis
• Conditional process reference distributions
• Evidence and decision provenance graphs
• Shared plotting and diagnostic interfaces

Validation

• 563 tests passed
• 0 failures, 0 warnings, 0 skips
• R CMD check: 0 errors, 0 warnings, 0 notes
• pkgdown::check_pkgdown(): no problems
• Source package and vignettes built successfully
• GitHub checks: 8/8 passed

Reproducibility

Release tag: v0.6.0

Release commit:

94ccf7198e05e6484021329221395836401dc50b

Source archive SHA-256:

2b73e84bc6738789e4912baf818236f1ddd54197c6e735aff874a09a81200bbc