Repository navigation
Release/0.6.6 - #190
Merged
Merged
Release/0.6.6#190
Conversation
- Use benchmark scenario IDs (lab01_md1643, lab15_md10) for the Lab1 and Lab15 results - Point validation to the public disslucc-benchmark instead of test file paths - State the MAE metric (final year; mean over all years 0.0026) and iteration counts - Replace per-step timings with whole-run timings from `make timing` - Describe the correctCellChange guard neutrally (field name mismatch) - Fellowship paragraph in the present tense (four fellows since September 2026) - Add DisSLUCCBenchmark to paper.bib
and disslucc-benchmark with Zenodo DOIs
scenarios; cite released versions with DOIs
spelling, AI disclosure, trim to 1700 words, final DOIs
validation numbers to v0.3.0, step 11, Game of Life wording
- BR-MANGUE validation against TerraME (baseline and flooding scenarios), tied to brmangue-dissmodel v0.3.0 - Mesa-Geo in the state of the field; remaining labs explained - AI usage disclosure rewritten (scope, phases, validation references) - INPE seminar (7 May 2026) in the research impact - paper.bib: final Zenodo DOIs (disslucc, brmangue-dissmodel, disslucc-benchmark 0.2.2), Costa2009 URL, seminar entry
…lab01 and tolerance Findings from running the validation as a JOSS reviewer would: - BR-MANGUE goldens come from a TerraME 2.0 adaptation of the published model (brmangue-terrame), as brmangue-dissmodel's docs/model-fidelity.md states; say so, cite it, and point to where the flooding goldens are regenerated (they are not shipped with v0.3.0). - The 60×60 cross-substrate check uses a 5 cm tolerance; state it. - In disslucc-benchmark the scenario named lab01_md1643 is the one that matches (cell_correction=False); the MAE 0.0036 belongs to the default run. Reword so the text matches the benchmark table. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01FiTtjpqiqAutXkzYsVzepv
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01S4SjCckAQ8koVsqjYipUHr
…Record - New `--preserve-output-name` flag on `run`: write to --output exactly as given instead of injecting the short experiment ID (default unchanged). - Directory output used a hard-coded `.tif` even for vector executors; the extension now comes from the executor's optional `output_suffix` class attribute (default `.tif`, so existing behaviour is kept). - The record written by the CLI had fixed placeholders (model_name="local", model_commit="local-cli", code_version="dev"). It now records the executor's `name`, the installed dissmodel version, and the package that defines the executor (e.g. "brmangue-dissmodel==0.3.0"), also when the executor file is run as a script from inside an installed package. Loose scripts keep "local-cli". Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01FiTtjpqiqAutXkzYsVzepv
…or import - shapefile_to_raster.md (repo root, not in the mkdocs nav) documented `dissmodel.geo.raster.io`, which no longer exists: every import in it raised ImportError. Moved to docs/examples/vector_to_raster.md, rewritten for `dissmodel.io.convert.vector_to_raster_backend` and `dissmodel.io.raster.save_geotiff`, and added to the nav. Dropped the unmeasured "≈ 4,500× faster" claim in favour of a pointer to the benchmarks. All snippets were run against the synthetic 60×60 grid and `mkdocs build --strict` passes. - README step 3 used `ForestFireModel` without importing it. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01FiTtjpqiqAutXkzYsVzepv
`_fill_min_distance` measured every cell against every target geometry in Python (`apply` + `distance().min()`). It now asks the target's STRtree (`sindex.nearest`) for the nearest geometry of each cell and measures only that pair with `shapely.distance`. Same values as before (tests compare against the brute force for points, lines and polygons, bit for bit); missing/empty geometries and an empty target still give NaN; index and row order are preserved; a CRS mismatch still warns. On a 200×200 grid with 2,000 random lines: 66 s → ~7 s. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01FiTtjpqiqAutXkzYsVzepv
The two nested Python loops created one `shapely.geometry.box` per cell. `np.meshgrid` + `shapely.box` on arrays now builds all cells at once, in the same column-major order and with the same "row-col" ids, which neighbourhoods and raster conversion depend on. A new test keeps the old loop as reference and checks, for the dimension, bounds+resolution and bounds+dimension forms, that ids and every vertex are identical. 1000×1000 grid: ~20 s → ~4 s. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01FiTtjpqiqAutXkzYsVzepv
CLI output name and provenance fixes, vector-to-raster docs, faster MIN_DISTANCE fill and vector_grid, plus the raster georeference changes already in Unreleased. Stays in 0.6.x so packages pinned to `dissmodel>=0.6.4,<0.7.0` (disslucc 0.5.0) pick it up unchanged. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01FiTtjpqiqAutXkzYsVzepv
…ismatch Running the README executor as a reviewer would, with `--output results/`: - the CLI did not create `results/`, so the save failed with FileNotFoundError; `run` now creates the parent of a local output path. - with the directory created, the generated `simulacao_<id>.tif` made save_dataset hand a GeoDataFrame to the GeoTIFF writer, which failed with `AttributeError: 'GeoDataFrame' object has no attribute 'arrays'`. save_dataset now checks the data type against the URI's format first and raises a TypeError that says what to do; the README executor declares `output_suffix = ".gpkg"`, so `--output results/` writes a GeoPackage. Both problems predate 0.6.6 (the `.tif` was hard-coded before). Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01FiTtjpqiqAutXkzYsVzepv
For a model with no annotation of its own, obj.__annotations__ resolves to Model's annotations, so the sidebar rendered env/start_time/end_time/_step and the env text box replaced the environment with a string. Collect the annotations of the user's classes only (base classes included) and skip private names. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01FiTtjpqiqAutXkzYsVzepv
Replace the ASCII box diagram (misaligned, and naming executors that no longer exist: CoastalRasterExecutor, LUCCVectorExecutor) with a Mermaid flowchart using current names (BrmangueRasterExecutor, LuccContinuousExecutor) and showing which core module each layer uses.
- README ecosystem table: disslucc is on PyPI (`pip install disslucc`); add disscube (`pip install "disscube[dissmodel]"`), which the paper cites but the table did not list - executor schemas example: coastal-dynamics was renamed to brmangue-dissmodel; use its `brmangue_raster` executor and the v0.5.0 tag
Replace the ASCII box diagram (misaligned, and naming executors that no longer exist: CoastalRasterExecutor, LUCCVectorExecutor) with a Mermaid flowchart using current names (BrmangueRasterExecutor, LuccContinuousExecutor) and showing which core module each layer uses.
0.6.6, coverage badge - paper: list AI uses in general terms (coding, bug fixing, refactoring, code translation, tests, validation scripts, documentation, text) and state that reference outputs are produced by TerraME itself, with AI-assisted changes to model files checked to leave outputs unchanged; replaces the claim that only the BR-MANGUE headless driver was AI-written - paper: cite luccme-goldens v1.1.0 (10.5281/zenodo.23161342), the reference outputs the disslucc benchmark compares against - README: citation block cited 0.6.5; coverage badge said 79% (CI: 83%)
In headless mode Chart and RasterMap write PNG frames to the working directory, so running pytest from the repository root left chart_frames/ and raster_map_frames/ behind. An autouse fixture moves each test of those two modules into tmp_path.
… 'adapts' TerraME paradigm
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Description
Release 0.6.6: bug fixes in the CLI, I/O and widgets, two vector performance improvements, and the
JOSS paper updated with Zenodo DOIs for every software dependency and benchmarks re-run on the
current stack.
Related Issue
N/A (release branch)
Changes Made
output_suffix), parent directory of--outputis created,ExperimentRecordrecords the executor name, dissmodel version anddefining package
save_dataset: clearTypeErrorwhen data and URI format do not matchdisplay_inputs: renders only user-declared annotations; for models without their ownannotations it exposed framework attributes and replaced
envwith a stringvector_gridcells built in one vectorized call;MIN_DISTANCEfill via spatial indexchart_frames/andraster_map_frames/in the repository root)dissmodel-ca 0.1.0, dissmodel-sysdyn 0.1.1 and luccme-goldens 1.1.0; benchmarks re-run
(dissmodel 0.6.6, NumPy 2.5.3, GeoPandas 1.2.0); statement of need, research impact and AI
disclosure revised;
SyncSpatialModel/SyncRasterModeldescribed0.6.6, coverage badge 83%
docs/examples/vector_to_raster.mdContributor Checklist
pytest): 489 passed, 2 skipped; doctests 19 passedruff check dissmodel/, as in CI);mypy dissmodelclean