v1.4.0

Changes:

• Made aligned full-file CICA Atlas subset requests explicitly return original
  files without applying dataset fixes by default. Deployments can set
  project:c3s-cica-atlas.apply_fixes_to_full_files = true to force those
  requests through normal processing and the configured fix backend. Whole-globe
  area parameters, and areas containing a dataset's complete spatial extent,
  are treated as no-ops when checking file alignment. Spatially disjoint areas
  now fail before subset processing starts.
• Added a memory-bounded post-processing step that merges consecutive subset
  and CMIP6-decadal concat batches up to the 2 GB output limit, returning one
  file when the complete result fits within that limit.
• Added independently configurable subset and decadal-concat batching memory
  aims, defaulting to 4GB. The planners now use the stricter of their
  timestep and estimated process-memory targets; concat accounts for every
  ensemble member.
• Moved adaptive time batching into the reusable rook.batch package and
  applied it to CMIP6-decadal realization concat so each time batch is written
  before the next one is assembled.
• Removed the legacy spec-file.txt alias; deployments should use the
  canonical linux-64.spec file directly.

v1.3.0

Changes:

• Added the lightweight synchronous health WPS process for service health
  checks. Its raw output is ROOK_HEALTH_OK on success; failed checks return
  an OGC exception with an explanation and omit the success marker. Deeper
  operational checks can be added later without changing this contract. Named
  projects can be configured to verify read access to their
  base_dir/.health-check.txt sentinel files.

• Made Woodpecker the default dataset fix provider while retaining the legacy
  provider as a configuration-selected fallback.

• Replaced the Woodpecker Git dependencies with the released
  roocs-woodpecker, roocs-woodpecker-atlas-plugin, and
  roocs-woodpecker-cmip6-decadal-plugin 0.7.1 packages, installed with Rook
  on top of the Conda environment.

• Removed the temporary per-request fix-provider override. Smoke tests now use
  the single provider selected in roocs.ini instead of running both
  providers for every relevant case.

v1.2.3

Changes:

• Renamed the internal request-processing layer from rook.director to
  rook.pflow and aligned the code, tests, and documentation around request
  sources, request decisions, operation execution, and response adaptation.
• Removed remaining director-era compatibility code while keeping the public WPS
  process interface compatible.
• Added clearer dataset-processing flow documentation, including high-level and
  detailed Mermaid diagrams.
• Clarified concat operation fix stages so the current CMIP6 decadal behavior is
  easier to preserve before moving dataset fixes to Woodpecker.
• Disabled overdesigned template GitHub Actions workflows by making version
  bumping and Docker testing manual-only for now.

v1.2.2

Changes:

• Refactored the old director flow into smaller planning and execution modules
  with explicit request sources, dataset sources, and original-file response
  plans.
• Moved operation implementations into the rook.operations namespace,
  simplified the operation adapters, and removed obsolete compatibility shims
  and dead helper code.
• Added characterization coverage around director, workflow, and operation
  execution paths.
• Deprecated the internal use of the apply_fixes WPS input while keeping the
  public process interface compatible.
• Removed the old Director class and director compatibility module after the
  remaining processes were migrated to the request result wrapper.
• Fixed catalog-resolved dataset ID handling for operation inputs.
• Fixed prepared dataset-source handling in the concat operation so
  orchestrated c3s-cmip6-decadal workflows can consume catalog-resolved inputs.
• Reduced noisy workflow file-probe logging for inline workflow documents.
• Added Mermaid-enabled documentation for the dataset processing flow and the
  cleaner director decision model.
• Updated development and GitHub Actions maintenance dependencies.

v1.2.1

Changes:

• Added characterization coverage around director and operation execution paths.
• Deprecated the internal use of the apply_fixes WPS input while keeping the
  public process interface compatible.
• Moved operation implementations into the rook.operations namespace and
  removed obsolete compatibility shims and dead helper code.
• Split director behavior into explicit request planning and execution modules,
  including catalog-resolved dataset sources and original-file response plans.
• Removed the old Director class and director compatibility module after the
  remaining processes were migrated to the request result wrapper.
• Fixed catalog-resolved dataset ID handling for operation inputs.
• Added Mermaid-enabled documentation for the current dataset-processing and
  director decision tree as a blueprint for the next cleanup phase.
• Updated development and GitHub Actions maintenance dependencies.

v1.2.0

Changes:

• Added preliminary support for Kerchunk reference inputs, including URL-aware
  detection for JSON, Zstandard, and Parquet references.
• Added preliminary S3 support by declaring s3fs, allowing direct s3://
  NetCDF inputs, and preparing optional S3 base-path mapping for catalog-backed
  processing paths.
• Added preliminary support for opening local and S3-backed Zarr stores as
  operation inputs.
• Added a common dataset-opening API with immutable DatasetSource inputs,
  explicit NetCDF, Zarr, and Kerchunk dispatch, and independent transport
  detection and configuration.
• Centralized Rook configuration loading, project lookup, storage roots, and
  shared S3 options, with strict validation that avoids exposing configured
  values in errors.
• Simplified catalog consolidation, including independent handling of mixed
  direct and catalog inputs and per-project catalog caching.
• Replaced captured configuration globals and aligned the development setup
  with prek.
• Live S3 integration coverage remains deferred; current object-storage
  behavior is covered by unit and characterization tests.

v1.1.1

Changes:

• Fixed post-merge CI behavior by skipping the no-commit-to-branch pre-commit hook on main workflow runs while keeping it active for pull requests and local use.
• Gated PyPI/TestPyPI publish jobs behind explicit repository variables to avoid failing release/tag post-process workflows when publishing is not enabled.
• Added make conda-lock and make conda-spec targets to automate regeneration of conda-lock.yml and linux-64.spec (with spec-file.txt alias refresh) for release prep.
• Added a concise Patch Release Quickstart section in README.rst under Contributing with minimal patch-release steps.

v1.1.0

Changes:

• Fixed remote intake catalog loading for partner environments behind proxies by using proxy-aware HTTP filesystem handling and safer catalog storage option resets.
• Made fsspec an explicit runtime dependency for intake catalog loading.
• Mitigated the intake shell(...) catalog default expansion risk by disabling shell-based parameter expansion when loading intake catalogs (see https://github.com/roocs/rook/security/dependabot/3).
• Cleaned up docs and tests.

v1.0.0

Changes:

• First major release to make the transition explicit: daops is now integrated in rook and no longer shipped as a separate dependency.
• Fixed test configuration loading order/reload in tests/conftest.py so regular test paths resolve consistently.
• Added a conda transition workflow: environment.yml as source of truth, generated conda-lock.yml and linux-64.spec, and CI verification to detect artifact drift.

v0.19.0

Changes:

• Added support for atlas v2.5 data (#290).
• Updated smoke tests.