Release v0.10.0

🚀 TopMark 0.10.0 — Major Pipeline & Machine Output Redesign

This release delivers a major internal and external evolution of TopMark, featuring a complete redesign of the machine-output schema, a reworked pipeline architecture, and substantial CLI behavior changes.

⚠️ This is a breaking release. Downstream consumers of JSON/NDJSON output or internal pipeline APIs must update.

---

⚠️ BREAKING CHANGES

Machine Output (JSON / NDJSON)

• Completely redesigned schema
  • Every record now includes a kind discriminator (config, config_diagnostics, result, summary)
  • All records carry a top-level meta block (version, intent, timestamps)
  • File-level results use a stable envelope (file, statuses, hints, diagnostic_counts, outcome)
  • NDJSON is now strictly one record per line, with consistent keys
• ❌ All pre-0.10.0 machine formats are removed
• 🔧 Downstream tools must update parsers

CLI Output & Presentation

• CLI rendering fully rewritten:
  • New bucketing semantics aligned with pipeline axes and policy
  • Summary footer replaced by a consistent, structured report
  • Changed / unchanged / would-change classification now driven by the comparison axis
  • Hints grouped, severity-ranked, and verbosity-aware
• Legacy formatting and summary behavior removed

Pipeline Architecture

• ProcessingContext split into clear responsibilities:
  • pipeline.context.model — orchestration & state
  • pipeline.context.policy — pure policy and feasibility
  • pipeline.context.status — axis status tracking
• Legacy pipeline modules removed
• Internal imports of old pipeline paths must be updated

CLI Commands

• Deprecated and legacy commands removed:
  • Old topmark header … compatibility paths
  • Legacy updater / stripper debug modes
• Removed commands now fail fast with explicit errors

---

✨ Highlights

• Clean separation of pipeline concerns (context / policy / status)
• Unified, stable machine-readable output for long-term integrations
• Significantly clearer CLI output with accurate bucket and hint logic
• Simplified typing and import model (collections.abc, Ruff UP / TC)
• Large set of correctness fixes across scanning, rendering, and writing

---

➕ Added

• New pipeline/context/ package:
  • model.py, policy.py, status.py
• Centralized machine-output builder (cli_shared.machine_output)
• Structured summary renderer (topmark.api.view.format_summary)
• Linting policy documentation in CONTRIBUTING.md
• GitHub-Flavored Markdown table support via mdformat-gfm

---

🔄 Changed

• ProcessingContext
  • Policy logic removed from the core context
  • Stronger invariants and explicit public interfaces
• Pipeline Steps
  • Rewritten to rely on pure policy helpers and explicit axes
• Rendering
  • Fully redesigned summary, bucketing, and writer output
• Machine Output
  • Predictable envelopes, deterministic NDJSON ordering
  • Config dumps and diagnostics included in machine mode
• Tooling
  • Dependency ranges tightened
  • Requirements regenerated
  • mdformat-tables replaced by mdformat-gfm

---

➖ Removed

• ReasonHint
• Legacy updater header paths
• Deprecated CLI commands and pre-0.9 compatibility code
• Legacy summary and bucketing pipeline

---

🐛 Fixed

• BOM / shebang / newline interactions
• Indentation handling for Markdown, HTML, and XML
• Header bound edge cases (multi-header, malformed, block comments)
• Writer stability in dry-run vs apply
• Accurate “would change” vs “changed” classification under policy constraints
• Improved file type detection for HTML / Markdown block comments

---

ℹ️ Notes

topmark.api remains public and stable, but all machine-readable output formats and internal pipeline interfaces have changed.

➡️ Integrators consuming JSON or NDJSON must migrate to the new schema and envelopes.

---

Full Changelog: v0.9.1...v0.10.0