antonio-leblanc · Nov 11, 2026 · Nov 11, 2026 · Nov 11, 2026
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -80,6 +80,10 @@ history, so they summarise each release rather than list every change. The
 - `plot[...;filename=*.asc]` writes a valid ESRI ASCII grid: one row per y,
   north first, at full precision. It used to write one row per x and round
   values to six significant digits.
+- Documentation: removed `defaultFuelType`, which nothing reads, and the
+  `Kcurv`/`beta` columns from the WindDriven examples, which only use
+  `vv_coeff`. `minSpeed`, `relax` and `smoothing` now describe what the code
+  does, and Windows pip users are pointed to WSL.
 
 ## [v2.6.0] — 2026-10-26
 

diff --git a/CONTRIBUTING.md b/CONTRIBUTING.md
@@ -62,6 +62,10 @@ We welcome code contributions, from bug fixes to new features.
 5.  **Make Changes:**
 
     Implement your code changes, following existing coding style and conventions where possible.
+    *   Changes to propagation or flux models and their physics start as an issue: the maintainers decide those.
+    *   A change that moves simulation results goes in its own PR, and at the top of its CHANGELOG entry.
+    *   Never regenerate a `.ref` file to make a test pass.
+    *   Report security problems through [private vulnerability reporting](https://github.com/forefireAPI/forefire/security/advisories/new), never in a public issue.
 6.  **Add Tests (if applicable):**
 
     For new features or significant bug fixes, please add corresponding tests or update existing ones. See [TESTING.md](TESTING.md) for what each suite covers, how to run it, and which ones CI validates. Ensure all tests pass locally.
@@ -70,7 +74,10 @@ We welcome code contributions, from bug fixes to new features.
     cmake -S . -B build && cmake --build build -j
     ctest --test-dir build --output-on-failure
 
-    # The script-driven suites (runff, runANN)
+    # Regression check: compares a real case against the .ref files
+    cd tests/runff && bash ff-run.bash
+
+    # Every script-driven suite your environment allows
     cd tests && bash run.bash
     ```
 7.  **Record the Change:**
@@ -91,7 +98,7 @@ We welcome code contributions, from bug fixes to new features.
     ```
 10. **Open a Pull Request (PR):**
 
-    Go to the `forefireAPI/forefire` repository on GitHub and open a Pull Request from your branch to the `dev` branch (or `master` if that's the target).
+    Go to the `forefireAPI/forefire` repository on GitHub and open a Pull Request from your branch to the `dev` branch.
     *   Provide a clear description of the changes in the PR.
     *   Link any relevant issues (e.g., "Fixes #123").
     *   Ensure the CI checks pass on your PR. The maintainers will review your code, provide feedback, and merge it once approved.

diff --git a/README.md b/README.md
@@ -44,8 +44,7 @@ import pyforefire as forefire
 from pyforefire.helpers import printToPathe
 
 ff = forefire.ForeFire()
-ff["fuelsTable"] = "Index;vv_coeff;Kcurv;beta\n1;1.0;1.0;1.0"
-ff["defaultFuelType"] = 1
+ff["fuelsTable"] = "Index;vv_coeff\n1;1.0"
 ff["propagationModel"] = "WindDriven"
 ff.execute("FireDomain[sw=(0,0,0);ne=(4000,4000,0);t=0]")
 

diff --git a/TESTING.md b/TESTING.md
@@ -40,6 +40,10 @@ The primary physics regression test, validated in CI by `main.yml` and `macos.ym
 2.  Navigate to the test directory: `cd tests/runff`
 3.  Execute the test script: `bash ff-run.bash`
 
+The test data and `.ref` files are stored in Git LFS. Fetch them first with
+`git lfs pull --include tests/runff` (or `git lfs pull` for every case).
+Configure checks the LFS files; `-DFOREFIRE_CHECK_LFS=OFF` skips that check.
+
 **Test Logic:**
 
 The `ff-run.bash` script:

diff --git a/app/forefire/commands.md b/app/forefire/commands.md
@@ -48,12 +48,15 @@ std::string commandHelp = R"(
 
 ## startFire
     startFire[loc=(x,y,z)|lonlat=(lon,lat,z);t=seconds|date=YYYY-MM-DDTHH:MM:SSZ]
-    Creates the smallest possible triangular fire front at the specified location and time
+    startFire[geojson=<file or inline GeoJSON>]
+    Creates the smallest possible triangular fire front at the specified location and time, or a front along a GeoJSON polygon
     Example: startFire[loc=(0.0,0.0,0.0);t=0.]
+    Example: startFire[geojson=ignition.geojson]
     Arguments:
      - 'loc': Starting location of the fire (coordinates x,y,z)
      - 'lonlat': WGS coordinates of the ignition point (lon,lat,z); all three values are required
-     - 't': Time when the fire is started (in seconds or as an absolute ISO GMT date);
+     - 't': Time when the fire is started (in seconds or as an absolute ISO GMT date)
+     - 'geojson': GeoJSON file or inline text (Polygon/MultiPolygon in lon,lat,z); a top-level "valid_at" date sets the time;
 
 ## step
     step[dt=seconds]

diff --git a/bindings/python/README.md b/bindings/python/README.md
@@ -44,9 +44,10 @@ If you need either, build from source (below).
 
 ### Building from source
 
-Any platform without a published wheel — Windows, musl-based Linux, or an
-unusual architecture — falls back to compiling the sdist, which needs a C++
-compiler, CMake ≥ 3.15, and the NetCDF C and legacy C++4 libraries:
+On Linux or macOS without a published wheel (musl-based Linux or an unusual
+architecture), pip compiles the sdist, which needs a C++ compiler, CMake ≥ 3.18,
+and the NetCDF C and legacy C++4 libraries. Native Windows is not supported;
+use WSL.
 
 ```bash
 # Debian/Ubuntu

diff --git a/bindings/wasm/README.md b/bindings/wasm/README.md
@@ -8,8 +8,8 @@ This is an experimental build, not a supported one yet. It answers the three
 questions the issue left open: does it link, does it run, how big is it.
 
 `.github/workflows/wasm.yml` builds the module and runs `smoke.mjs` on every
-pull request, and `.github/workflows/wasm-pages.yml` publishes the browser
-demo to GitHub Pages on every push to `dev`. The `.wasm` is built in CI rather
+pull request, and the same workflow publishes the browser demo to GitHub Pages
+on every push to `dev`. The `.wasm` is built in CI rather
 than committed: it is a megabyte of generated output, and a stale binary
 sitting next to live sources is worse than a two minute build.
 

diff --git a/docs/source/reference/commands.rst b/docs/source/reference/commands.rst
@@ -123,21 +123,24 @@ startFire
 .. code-block:: none
 
    startFire[loc=(x,y,z)|lonlat=(lon,lat,z);t=seconds|date=YYYY-MM-DDTHH:MM:SSZ]
+   startFire[geojson=<file or inline GeoJSON>]
 
-Ignites a minimal clockwise triangular fire front (3 ``FireNode`` markers) at the given location and time.
+Ignites a minimal clockwise triangular fire front (3 ``FireNode`` markers) at the given location and time, or a front along a GeoJSON polygon.
 
 **Arguments:**
 
 * ``loc=(x,y,z)``: Ignition point in Cartesian domain coordinates (meters).
 * ``lonlat=(lon,lat,z)``: Ignition point in WGS84 coordinates (longitude, latitude, elevation). All three components are required (``lonlat=(lon,lat)`` is ignored).
 * ``t=seconds`` or ``date=YYYY-MM-DDTHH:MM:SSZ``: Ignition time in seconds or ISO 8601 UTC format.
+* ``geojson=``: Path to a GeoJSON file, or the GeoJSON text inline between triple quotes. The ``Polygon`` or ``MultiPolygon`` coordinates are ``[lon, lat, z]``. A top-level ``"valid_at"`` date, if present, sets the ignition time. See ``tests/runff/rungeojson.ff``.
 
 **Example:**
 
 .. code-block:: none
 
    startFire[loc=(5000,5000,0);t=0]
    startFire[lonlat=(9.0,42.0,0);date=2024-07-15T10:00:00Z]
+   startFire[geojson=ignition.geojson]
 
 
 .. _cmd-step:

diff --git a/docs/source/reference/parameters.rst b/docs/source/reference/parameters.rst
@@ -32,7 +32,7 @@ propagationSpeedAdjustmentFactor
 
 minSpeed
 ~~~~~~~~
-* **Description:** Minimum propagation speed floor (in m/s) enforced when advancing a ``FireNode``.
+* **Description:** A ``FireNode`` whose speed (in m/s) is at or below this value stops advancing. It is a stopping threshold, not a floor on the speed.
 * **Default:** ``0.005``
 
 maxSpeed
@@ -111,12 +111,12 @@ curvatureScheme
 
 smoothing
 ~~~~~~~~~
-* **Description:** Smoothing weight applied during normal vector computation.
+* **Description:** Read but not used by the current node update.
 * **Default:** ``1``
 
 relax
 ~~~~~
-* **Description:** Relaxation factor (:math:`0\text{–}1`) used when updating node normal vectors and front depth.
+* **Description:** Relaxation factor (:math:`0\text{–}1`) that blends a node's previous speed and front depth with the newly computed ones.
 * **Default:** ``0.5``
 
 frontDepthComputation

diff --git a/docs/source/user_guide/ensembles.rst b/docs/source/user_guide/ensembles.rst
@@ -30,8 +30,7 @@ against a ``pip``-installed ForeFire.
    L = 4000.0
 
    def setup(ff):
-       ff["fuelsTable"] = "Index;vv_coeff;Kcurv;beta\n1;1.0;1.0;1.0"
-       ff["defaultFuelType"] = 1.
+       ff["fuelsTable"] = "Index;vv_coeff\n1;1.0"
        ff["windReductionFactor"] = 1.0
        ff["propagationModel"] = "WindDriven"
        ff.execute(f"FireDomain[sw=(0,0,0);ne=({L},{L},0);t=0]")

diff --git a/docs/source/user_guide/forefire_script.rst b/docs/source/user_guide/forefire_script.rst
@@ -31,9 +31,9 @@ Grammar & Syntax Rules
 
    .. code-block:: none
 
-      setParameter[fuelsTable="""Index;vv_coeff;Kcurv;beta
-      1;1.0;1.0;1.0
-      2;0.3;1.0;1.0"""]
+      setParameter[fuelsTable="""Index;vv_coeff
+      1;1.0
+      2;0.3"""]
 
 5. **Parameter Substitution** (``*param*``) **and Protected Values** (``!``):
 

diff --git a/docs/source/user_guide/fuels_and_models.rst b/docs/source/user_guide/fuels_and_models.rst
@@ -37,7 +37,7 @@ Model Families and Their Fuel Columns
    * - ``Iso``
      - **None.** Spreads at the constant speed given by ``setParameter[Iso.speed=0.5]`` (m/s) in all directions.
    * - ``WindDriven``
-     - ``Index;vv_coeff;Kcurv;beta`` (idealized wind-driven model used in :doc:`python_arrays` and :doc:`ensembles`).
+     - ``Index;vv_coeff``: the rate of spread is ``vv_coeff × normal wind`` (idealized wind-driven model used in :doc:`python_arrays` and :doc:`ensembles`).
    * - ``Rothermel``, ``BalbiNov2011``, ``Balbi2020``, ``Balbi2015``
      - The standard 23-column Rothermel/Balbi fuel table detailed in :ref:`rothermel-fuel-file` below.
    * - ``RothermelAndrews2018``

diff --git a/docs/source/user_guide/mesonh.rst b/docs/source/user_guide/mesonh.rst
@@ -80,6 +80,7 @@ While it runs, open ``http://localhost:8080/mnh3d.html``. The page displays
 the terrain, fuel, fire perimeter, plume top and surface wind. The browser
 needs internet access to load its JavaScript dependencies.
 
-This is a live view of the calculation: it does not provide a replay or
-fast-forward control. Use one viewer tab at a time. For comparisons of
+The page follows the run live. Once it has steps to show, a Timeline slider
+lets you scrub or play them back. For this bundled case, opening the page
+after the run replays the steps saved in ``ForeFire/Outputs`` and ``MODEL2``. Use one viewer tab at a time. For comparisons of
 simulation outputs, leave the viewer closed, as explained in the guide.
diff --git a/docs/source/user_guide/python_arrays.rst b/docs/source/user_guide/python_arrays.rst
@@ -45,8 +45,7 @@ against a ``pip``-installed ForeFire.
    # Fuel table, inline — no CSV file needed. The columns depend on the
    # propagation model: the WindDriven model used here only needs vv_coeff
    # (see "Choosing a fuel table" below).
-   ff["fuelsTable"] = "Index;vv_coeff;Kcurv;beta\n1;1.0;1.0;1.0\n2;0.3;1.0;1.0"
-   ff["defaultFuelType"] = 1.
+   ff["fuelsTable"] = "Index;vv_coeff\n1;1.0\n2;0.3"
 
    # Solver parameters.
    for key, value in {
@@ -122,7 +121,7 @@ The columns of ``fuelsTable`` are read by the propagation model, so they change
 with the model you select:
 
 * ``WindDriven`` — a minimal table with ``vv_coeff`` (the rate of spread is
-  ``vv_coeff × normal wind``), plus ``Kcurv`` and ``beta``. Convenient for
+  ``vv_coeff × normal wind``). Convenient for
   idealized and teaching cases.
 * ``Rothermel``, ``BalbiNov2011``, ``Balbi2015`` and the other physical models
   — the full fuel description in ``fuels.csv``. See

diff --git a/tests/README.md b/tests/README.md
@@ -102,7 +102,8 @@ are demonstrations of the API rather than tests of the physics.
 compares ForeFire against a FARSITE case, and needs `flatland.lcp`, which is
 not in the repository — `python/README.md` has the download URL.
 
-Three test scripts sit here too, and `run.bash` runs none of them:
+Eight `test_*.py` scripts sit here too, and `run.bash` runs only
+`test_validate.py`. `python/README.md` covers them all; the main ones:
 
 | File | What it checks | Where it runs |
 | --- | --- | --- |

diff --git a/tests/python/README.md b/tests/python/README.md
@@ -3,7 +3,8 @@
 Everything here needs the `pyforefire` module: either a build configured with
 `-DFOREFIRE_BUILD_PYTHON=ON`, or `pip install forefire`.
 
-`../run.bash` runs only the first two, through `$PYTHONEXE`.
+`../run.bash` runs only the first two examples and `test_validate.py`, through
+`$PYTHONEXE`.
 
 ## `percolation.py`
 
@@ -35,8 +36,8 @@ The weather it uses, `flatland_3mph0deg7hr.raws`, *is* in this directory.
 
 ## The test scripts
 
-These four are tests rather than examples, and `../run.bash` runs none of
-them. They are plain scripts, not pytest modules. `TESTING.md` covers each in
+The `test_*.py` scripts are tests rather than examples, and `../run.bash`
+runs only `test_validate.py`. They are plain scripts, not pytest modules. `TESTING.md` covers each in
 detail, including how to build the module they import.
 
 ### `test_moisture_invariants.py`