[theontho]

Summary

Prepare CLI 0.6.3, following v0.6.2's release PR #172. This changes only the synchronized versions in pyproject.toml, pixi.toml, and src/wgsextract_cli/__init__.py, using scripts/bump_version.py.

There is no tracked CLI CHANGELOG. The draft release notes and root-only operator checklist are local, ignored artifacts in the rollout worktree:

• scratch/release/v0.6.3-notes.md
• scratch/release/v0.6.3-checklist.md

Merged since v0.6.2

• Fail early for Windows-unsupported commands (D5) #174: fail early and clearly for Windows-unsupported commands; retain table-backed PhyloP.
• Parallelize whole-genome SNP and indel calling by contig (O1) #175 / Make per-contig small-variant calling opt-in for exactness #199: optional per-contig SNP/indel workers. Serial is the default: independent bcftools subsampling streams change some deep-site likelihoods/calls; full-genome output was not byte-identical (O1 parallel SNP calling differs from serial at 9 high-depth low-MQ sites on full genome #197).
• Make BWA and bwa-mem2 batches independent of thread count (O2) #176: fixed BWA/BWA-MEM2 -K 100000000 batches for thread-count reproducibility.
• Guard BWA/MEM2 pipelines with available-memory component estimates (D3) #179: measured, native available-memory guard; explicit MEM2 refusal, automatic BWA fallback, BWA warning only.
• Fix hs38DH indel calling at IUPAC reference bases #190: hs38DH IUPAC indel normalization and compatible consensus generation.
• Scale alignment memory and sort spill estimates by input read count #194: input-read-scaled duplicate-marking RAM and sort-spill estimates using bounded FASTQ sampling / indexed alignment counts.
• fix(vcf): honor trio --input and normalize parent-only contigs #195: trio --input precedence/BAM/CRAM support and parent-only random/unplaced/ALT contig aliases.
• Add CLI preflight memory and disk resource estimates #198: advisory memory/disk preflight, guarded estimation failures, retained-BAM/unknown-reference budgeting, and strict reusable GUI API.
• Fix native Windows GATK drive-path handling #201: native Windows GATK/Java drive-path handling; real hosted Windows GATK smoke.
• Fix reference refresh downloads resuming onto completed files #203: atomic verified reference refresh; only genuine .partial files resume.

Release checklist — root executes after gates pass

• Fill final native-Windows full-genome validation/pins/exactness/timing/waiver placeholders; no full-human MEM2 claim.
• Decide Allow align staging and sort spills on a chosen scratch volume (O4) #177 (align --temp-dir) / Cap native MEM2 sort memory at 256 MiB per thread (O5) #178 (MEM2 sort cap) after their independent 5M-pair gates; both remain draft/pending, not included.
• Refresh this release branch if additional work merges; require final exact-head local checks, CI, and reviews before marking ready or merging.
• Root tags only the approved release merge SHA. Require Release verification and its generated draft release; review notes before root publishes.
• GUI pin: follow gui-for-cli#168 / the hourly CLI-release watcher; open the pin PR as a person, update the GUI CHANGELOG, require GUI CI, merge, then release the chosen GUI tag through Distribution.
• Artifacts: verify published DMGs, Windows installer, Linux packages and updater signatures/digests, not just local builds. CLI v0.6.2 had no uploaded release assets; GitHub source archives and separate Windows-tool pins remain the model.
• Install Lab: dispatch install_lab.yml with app_release=<APP_TAG> or await the app-release watcher. Require Linux/macOS/Windows pinned acceptance, negative controls and windows-judge, not only tree/collection jobs.
• Preserve Install Lab install-lab-* evidence artifacts (14-day retention), release URLs/SHAs/digests, signing checks, oracle verdicts and uninstall restoration.

Validation

• Version synchronization / bump-script tests: 3 passed.
• Mandatory Ruff check/fix and format: passed.
• Source mypy: passed, 129 files.
• Repository lint/site/actionlint/Actions checks: passed.
• Built and installed wgsextract_cli-0.6.3-py3-none-any.whl; package smoke passed.
• Evidence: scratch/p2-code/scratch/release-0.6.3/out/phase2/code/release-prep/verification.log.

All local work used at most one core / 2 GiB. Full ordinary/Windows CI remains a final release gate, not a claim that the pending full-genome validation passed. No VM handoff, VM/lab dispatch, merge, tag, publication, GUI pin update, or release artifact publication is authorized by this PR.

Keep this PR draft and unmerged until root explicitly approves the final gates.

Compare: v0.6.2...v0.6.3

[coderabbitai]

Important

Draft PR not reviewed

Draft PRs are not automatically reviewed by default.

• Trigger a manual review

To automatically review draft PRs, update your CodeRabbit configuration:

reviews:
  auto_review:
    drafts: true

• Autofix · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}