Release/0.4.0

.Rbuildignore

@@ -1,14 +1,52 @@
 ^docs$
-^\\.Rproj\\.user$
 
 ^.*\.Rproj$
 ^\.Rproj\.user$
+^\.Renviron$
+^\.Rhistory$
+\.Rhistory$
 ^\.claude$
+^\.vscode$
+^\.idea$
+
+# Check / build artifacts produced in the source tree.
 ^splitGraph\.Rcheck$
+^.*\.Rcheck$
 ^splitGraph_.*\.tar\.gz$
+^revdep$
+^MD5$
+
+# Repository-only material: roadmaps, the JOSS paper, CI, coverage and
+# pkgdown configuration, and the developer scripts behind dev/release.md.
 ^ROADMAP.*\.md$
 ^\.github$
 ^paper\.md$
 ^paper\.bib$
+^paper-figures$
 ^codecov\.yml$
 ^CODE_OF_CONDUCT\.md$
+^cran-comments\.md$
+^_pkgdown\.yml$
+^_pkgdown\.yml\.bak$
+^\.lintr$
+^dev$
+^pkgdown$
+^pkgdown-site$
+
+# Python bytecode. The cache directory sits at
+# inst/python/splitspec/__pycache__, so match it at any depth rather than at a
+# fixed path: an anchored pattern silently let the .pyc files into the tarball.
+__pycache__
+\.pyc$
+
+# knitr leftovers from rendering a vignette in place. The vignettes shipped to
+# users are built into inst/doc by `R CMD build`, never from here.
+^vignettes/figure$
+^vignettes/.*_cache$
+^vignettes/.*_files$
+^vignettes/.*\.html$
+^tests/testthat/Rplots\.pdf$
+^Rplots\.pdf$
+
+^\.DS_Store$
+^.*/\.DS_Store$

.github/workflows/R-CMD-check.yaml

@@ -28,10 +28,21 @@ jobs:
     env:
       GITHUB_PAT: ${{ secrets.GITHUB_TOKEN }}
       R_KEEP_PKG_SOURCE: yes
+      # Neither check-r-package nor rcmdcheck sets NOT_CRAN, so without this
+      # every skip_on_cran() test (the Python conformance check, the
+      # performance budget) would be skipped in CI too.
+      NOT_CRAN: true
+      # The performance budget has its own Linux-only workflow; skip it here so
+      # slow shared Windows/macOS runners cannot fail the check with noise.
+      SPLITGRAPH_SKIP_PERF: true
 
     steps:
       - uses: actions/checkout@v4
 
+      - uses: actions/setup-python@v5
+        with:
+          python-version: "3.x"
+
       - uses: r-lib/actions/setup-pandoc@v2
 
       - uses: r-lib/actions/setup-r@v2

.github/workflows/perf-budget.yaml

@@ -0,0 +1,37 @@
+# Wall-clock budget guard for the core pipeline (tests/testthat/test-performance.R).
+# Runs on Linux only: the budgets are generous, but a shared Windows/macOS runner
+# can be slow enough to produce noise, and the point is to catch a return to
+# superlinear behaviour, not to benchmark.
+on:
+  push:
+    branches: [main, master]
+  pull_request:
+    branches: [main, master]
+
+name: perf-budget
+
+permissions: read-all
+
+jobs:
+  perf-budget:
+    runs-on: ubuntu-latest
+    env:
+      GITHUB_PAT: ${{ secrets.GITHUB_TOKEN }}
+      NOT_CRAN: true
+    steps:
+      - uses: actions/checkout@v4
+
+      - uses: r-lib/actions/setup-r@v2
+        with:
+          use-public-rspm: true
+
+      - uses: r-lib/actions/setup-r-dependencies@v2
+        with:
+          extra-packages: any::testthat, any::pkgload
+
+      - name: Run the performance budget tests
+        run: |
+          Rscript -e 'pkgload::load_all(".", quiet = TRUE); testthat::test_file("tests/testthat/test-performance.R", reporter = "progress", stop_on_failure = TRUE)'
+
+      - name: Print the benchmark table
+        run: Rscript inst/bench/pipeline.R 500 2000 5000

.github/workflows/pkgdown.yaml

@@ -0,0 +1,53 @@
+# Build the pkgdown site and publish it to GitHub Pages. The DESCRIPTION `URL`
+# field advertises https://selcukorkmaz.github.io/splitGraph/, so this workflow
+# has to run at least once (and Pages has to be enabled for the repository)
+# before a CRAN submission, or `R CMD check --as-cran` reports that URL as a
+# 404 NOTE.
+on:
+  push:
+    branches: [main, master]
+  pull_request:
+    branches: [main, master]
+  release:
+    types: [published]
+  workflow_dispatch:
+
+name: pkgdown
+
+permissions: read-all
+
+jobs:
+  pkgdown:
+    runs-on: ubuntu-latest
+    # Only restrict concurrency for non-PR jobs
+    concurrency:
+      group: pkgdown-${{ github.event_name != 'pull_request' || github.run_id }}
+    env:
+      GITHUB_PAT: ${{ secrets.GITHUB_TOKEN }}
+    permissions:
+      contents: write
+    steps:
+      - uses: actions/checkout@v4
+
+      - uses: r-lib/actions/setup-pandoc@v2
+
+      - uses: r-lib/actions/setup-r@v2
+        with:
+          use-public-rspm: true
+
+      - uses: r-lib/actions/setup-r-dependencies@v2
+        with:
+          extra-packages: any::pkgdown, local::.
+          needs: website
+
+      - name: Build site
+        run: pkgdown::build_site_github_pages(new_process = FALSE, install = FALSE)
+        shell: Rscript {0}
+
+      - name: Deploy to GitHub pages
+        if: github.event_name != 'pull_request'
+        uses: JamesIves/github-pages-deploy-action@v4.5.0
+        with:
+          clean: false
+          branch: gh-pages
+          folder: pkgdown-site

.github/workflows/test-coverage.yaml

@@ -14,10 +14,18 @@ jobs:
     runs-on: ubuntu-latest
     env:
       GITHUB_PAT: ${{ secrets.GITHUB_TOKEN }}
+      # covr does not set NOT_CRAN either; without it the Python conformance
+      # test is skipped and its lines are never counted.
+      NOT_CRAN: true
+      SPLITGRAPH_SKIP_PERF: true
 
     steps:
       - uses: actions/checkout@v4
 
+      - uses: actions/setup-python@v5
+        with:
+          python-version: "3.x"
+
       - uses: r-lib/actions/setup-r@v2
         with:
           use-public-rspm: true

.gitignore

@@ -1,4 +1,36 @@
+# R session state
 .Rproj.user
 .Rhistory
 .RData
 .Ruserdata
+.Renviron
+
+# Build and check artifacts produced in the source tree by `R CMD build` /
+# `R CMD check` / revdepcheck. None of these belong in the repository.
+*.tar.gz
+*.Rcheck/
+revdep/
+
+# Generated by `R CMD build --md5` (as CRAN does) inside the tarball; never part of the source tree.
+MD5
+
+# Rendered pkgdown site (the workflow deploys it to the gh-pages branch).
+pkgdown-site/
+
+# knitr leftovers from rendering a vignette in place; the shipped copies are
+# built into inst/doc by `R CMD build`.
+vignettes/figure/
+vignettes/*_cache/
+vignettes/*_files/
+vignettes/*.html
+Rplots.pdf
+
+# Python bytecode from running the shipped splitspec reader.
+__pycache__/
+*.pyc
+
+# Claude Code's machine-local permission allowlist. A shared `.claude/settings.json`
+# would still be committable; this one is per-machine.
+.claude/settings.local.json
+
+.DS_Store

.lintr

@@ -0,0 +1,14 @@
+linters: linters_with_defaults(
+    line_length_linter(160),          # signature-heavy R; 8 offenders were rewrapped, this is now a real gate
+    object_name_linter = NULL,        # `.depgraph_*` internal prefix and S3 method names
+    object_length_linter = NULL,      # the same prefix pushes internal names past 30 characters by design
+    commented_code_linter = NULL,
+    indentation_linter = NULL,        # roxygen-heavy files trip this on continuation lines
+    object_usage_linter = NULL        # false positives on data.frame column names in `[`
+  )
+exclusions: list(
+    "inst/bench",
+    "docs",
+    "tests/testthat/_snaps"
+  )
+encoding: "UTF-8"

DESCRIPTION

@@ -1,6 +1,6 @@
 Package: splitGraph
 Title: Dataset Dependency Graphs for Leakage-Aware Evaluation
-Version: 0.3.0
+Version: 0.4.0
 Authors@R: c(
     person("Selcuk", "Korkmaz", role = c("aut","cre"), email = "selcukorkmaz@gmail.com",
            comment = c(ORCID = "0000-0003-4632-6850"))
@@ -17,8 +17,10 @@ BugReports: https://github.com/selcukorkmaz/splitGraph/issues
 Encoding: UTF-8
 Depends: R (>= 4.1.0)
 Imports: graphics, igraph, stats, utils
-Suggests: bioLeak, jsonlite, knitr, pkgload, rmarkdown, testthat (>= 3.0.0)
+Suggests: bioLeak, jsonlite, knitr, pkgload, rmarkdown, rsample,
+    SummarizedExperiment, testthat (>= 3.0.0)
 VignetteBuilder: knitr
 Config/testthat/edition: 3
+Config/Needs/website: selcukorkmaz/leakdown
 NeedsCompilation: no
 RoxygenNote: 7.3.3