Changes in version 0.1.0.9000                      

NIfTI raw-map ingestion

  - Beta-only NIfTI sources now materialise as realised GDS objects by
    adding a synthetic var = 1 assay with a warning, matching
    gds_from_neurovols() behavior for raw maps without uncertainty
    images.
  - Documented beta-only gds_from_nifti_maps() workflows and the
    recommendation to use ols:voxelwise rather than fixed/random-effects
    meta-analysis when the variance assay is synthetic.
  - Tightened gds_from_nifti_maps() contrast relabeling for one-contrast
    raw-map layouts.
  - Added scalar-map workflow helpers: gds_from_scalar_maps() /
    as_scalar_map_gds(), group_ols(), one_sample(), two_sample(), and
    write_nifti_assays() with a per-file manifest.

CLI overhaul

  - Expanded the bundled fmrigds CLI into a plan-oriented interface with
    probe, plan, preview, run, and list commands.
  - Added saved-plan workflows via --save-plan / --load-plan, plus
    repeatable passthrough flags for adapter, reducer, post-hoc, and
    writer options.
  - Added CLI support for attaching subject/sample/contrast metadata
    tables, mask policies, preview tables, and registry inspection.
  - Added a new cli-workflows vignette covering the main command
    patterns and advanced control flags.
  - Added focused CLI tests covering plan construction, JSON output, raw
    previews, and reducer registry listing.

Sprint 10 / Repeated-measures LMM polish

  - Added restricted repeated-measures Gaussian LMM reducers: lmm:ri and
    lmm:ri_slope1.
  - Added theta_mode = "pooled" and theta_mode = "voxelwise" for shared-
    vs sample-specific variance parameters.
  - Added long-table repeated-measures ingestion via contrast_data_cols,
    so contrast metadata can be extracted directly from tabular sources.
  - Hardened the low-rank LMM kernels against numerical asymmetry in
    intermediate SPD matrices.
  - Expanded mixed-model tests to cover lme4 parity, pooled/voxelwise
    equivalence for identical responses, and sample-specific theta
    recovery.
  - Added a dedicated repeated-measures vignette and reducer-specific
    documentation for the supported LMM contract.
  - Added CI coverage for lme4-backed restricted LMM parity tests plus a
    scheduled benchmark workflow.
  - Warm-started voxelwise theta optimization from pooled and
    previous-sample fits to reduce repeated optimizer cold starts.

Release-track cleanup

  - Aligned package metadata, docs, and provenance fields around the
    fmrigds name.
  - Replaced scaffold-era package documentation with a user-facing
    package overview.
  - Added console print methods for realised GDS objects and plans.
  - Tightened README examples and installation guidance to reflect the
    current package surface.

Sprint 9 / GDS-first integration

  - Added in-memory tabular support: gds(<data.frame>) now works via the
    tabular adapter (no temporary files needed).
  - Added roi_col alias in tabular ingestion to treat ROI/parcel labels
    as the sample axis.
  - NIfTI source normalization: adapters now accept list(beta=...,
    se=...), directories, or vectors; classification by filename
    patterns for beta vs se.
  - Introduced an in-memory adapter and as_plan.gds() so verbs like
    reduce()/posthoc() accept realized GDS directly.
  - Added ergonomic alias plan(x) for as_plan(x).
  - Relaxed as_gds() to accept 2-D inputs and standardized all assays
    to 3-D [sample × subject × contrast] shapes.
  - Auto-derivation of var from se (and vice versa when required) in
    reducer/posthoc preflight.
  - Implemented spatial FDR (fdr:spatial) posthoc with group-wise Simes
    + weighted BH; documented usage and options.
  - Promoted OLS covariance triangles to assays (cov:<term_i>:<term_j>)
    while retaining attachments for metadata.
  - Added S3 accessors for plans: subjects.gds_plan() and
    contrasts.gds_plan().
  - New helper nifti_source(beta=NULL, se=NULL) for explicit NIfTI list
    sources.

Guidance for integrators (fmrireg):

  - Prefer gds(df, format="tabular", roi_col="roi") or as_gds(df,
    mapping=list(roi="roi", ...)) over temporary CSV writes.
  - Use plan(gds_obj) or as_plan(gds_obj) to chain reduce()/posthoc() on
    realized results.
  - Expect consistent 3-D assay shapes; 2-D inputs are upcast
    automatically.

Sprint 6

  - Added an HDF5 storage adapter (gds-h5) with probe and block-read
    support so plans can lazily stream assays from disk.
  - Documented the HDF5 layout under inst/schemas/hdf5.md and wired
    helpers for writing simple gds stores.
  - Expanded tests to cover the new adapter and serialization pipeline.

Sprint 7

  - compute() now supports sink = "h5" and write_out() queues, emitting
    provenance-aware gds-h5 stores (with tests).
  - Map families registered on plans/GDS are serialised under
    /gds/alignments, rehydrated at load time, and available via
    align(plan, "<name>").