v1.4.2: technical accuracy of model documentation¶
Release date: 22 October 2026. Git tag: v1.4.2.
Changes¶
The model guide shows the exposure-time floors already used by circulation and distinguishes them from early radial wall cooling. Its drilling-rate equations describe the spallation, assisted and mechanical branches, including the penetration-rate cap and implemented face-heat terms. It distinguishes the standalone face-heat iteration from the reference evaluation's fixed 30 kW load.
The applicability table separates accepted inputs from property-grid coverage and describes clipping, the heat-capacity path's limited extrapolation and the friction-heating pressure floor. The API reference identifies absolute and gauge pressures and annotates pressure differences. Its introductory pressure section is generated from the pressure module docstring.
The validation guide identifies fitted parameters and shared inputs for each comparison. It describes Newberry as a consistency check, explains the wellbore-failure evidence behind Soultz's stress bounds and limits the FORGE conservatism claim to the observed tool-temperature comparison. Existing numerical values and tables are retained. Reporting property lookups outside the grid is recorded as deferred work; no instrumentation is added.
Calculation and data preservation¶
Calculations, numerical inputs and public signatures are unchanged. The changed calculation files contain docstring edits only. Source fingerprints include docstring bytes and therefore change; the fingerprint scheme and stored fingerprints are preserved.
The API generator's documentation insertion is the only executable Python change. Its remaining code is unchanged. The README, guides and summary PDF retain their v1.4.1 bytes, as do source data, benchmark records and saved results. No golden or tolerance is regenerated.
Verification¶
The final local fast suite passed 404 tests in 24.64 seconds, with 106 slow tests excluded. Local documentation links and absolute main-branch repository links passed across 47 maintained Markdown files. The API page exactly matches its generator. All nine documented scalar ranges, the property-grid endpoints and the pressure constants match the code.
Parsed code is identical to v1.4.1 after removing docstrings in all 101 tracked Python files outside the API generator. Removing the generator's pressure-section insertion restores its baseline parsed code exactly. All 387 checked source-data, benchmark, saved-result and golden files retain their v1.4.1 bytes.
The wheel and source archive built successfully and passed distribution validation. A fresh Python 3.14.4 environment installed the local wheel with the locked runtime dependencies and without Matplotlib. Installed public API, packaged site data, study imports, provenance, all three command-line help checks and dependency consistency passed.
The complete v1.4.0 suite passed 507 tests, including 106 slow tests, on candidate abdb56caec44e8d3a43905e869e87983f5fda38d in CI run 39660954178. This is inherited physical-regression evidence. A full physical regression run is not required for this Markdown and docstring patch: executable calculation code and numerical inputs are unchanged. The API generator's documentation addition is checked separately.
Installation and citation¶
python -m pip install borethm==1.4.2
The GitHub release and PyPI page identify this version. The all-versions software DOI and author ORCID iD are retained in CITATION.cff.