# iftypeset — Competitor / Positioning Matrix (v0.1)

This is a practical market map to keep us honest about what exists today and what `iftypeset` is trying to be.

## The category we’re building

Not “Markdown to PDF” (that’s solved), but:

**Deterministic publishing CI**: Markdown → styled output **plus enforceable QA gates** (widows/orphans/overflow/keeps/link wrap) with machine-readable reports and coverage tracking.

## High-level landscape

### A) Renderers / converters (output-first)

Great at converting formats, but usually do **not** ship “layout QA gates” as a product.

- Pandoc (+ LaTeX) / Quarto / RMarkdown
- Typst
- LaTeX toolchains
- Markdown site tools that export PDF (MkDocs, Docusaurus, GitBook, Notion exports)

### B) Paged-media engines (layout-first)

Excellent at pagination + print rules, but they don’t give you a Chicago/Bringhurst rule registry or a publishing QA runtime by default.

- PrinceXML
- Antenna House Formatter
- WeasyPrint
- Vivliostyle / Paged.js
- wkhtmltopdf (HTML → PDF, limited paged-media fidelity)

### C) SaaS PDF rendering APIs

Operational convenience; QA gates are typically “your responsibility”.

- DocRaptor (Prince-powered)
- Various HTML→PDF APIs (vendor-specific)

## Feature comparison (typical, not absolute)

Legend:
- **✓**: first-class / native
- **~**: possible but not the default product shape
- **—**: not typical / not supported

| Capability | iftypeset (goal) | Pandoc/Quarto | Typst | LaTeX | Prince/AH | WeasyPrint | Vivliostyle/Paged.js |
|---|---:|---:|---:|---:|---:|---:|---:|
| Markdown → HTML | ✓ | ✓ | ~ | ~ | — | — | ✓ |
| Markdown → PDF | ✓ (via adapters) | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ (via headless browser) |
| Deterministic artifacts (stable HTML/IDs) | ✓ | ~ | ~ | ~ | ~ | ~ | ~ |
| Profile tokens (web_pdf/print_pdf/etc.) | ✓ | ~ | ~ | ~ | ~ | ~ | ~ |
| Rule registry w/ pointers (no quotes) | ✓ | — | — | — | — | — | — |
| Lint + autofix (editorial hygiene) | ✓ | ~ | ~ | ~ | — | — | — |
| Post-render QA gates (widows/orphans/overflow) | ✓ | — | — | — | ~ | ~ | ~ |
| Coverage reporting for implemented rules | ✓ | — | — | — | — | — | — |
| Degraded-mode handling (garbage inputs) | ✓ | ~ | ~ | ~ | — | — | — |

## What’s actually different about iftypeset

1. **Quality is a build gate**  
   Not “this looks nice”, but “this fails CI if link-wrap/table overflow/stranded headings exceed thresholds”.

2. **Rules are a registry (not hardcoded CSS)**
   Chicago/Bringhurst become paraphrased, pointer-backed records you can audit, diff, and expand over time.

3. **Renderer-agnostic**  
   The PDF engine is pluggable. The “meaning” is in tokens + QA + reports, not the renderer choice.

4. **Traceability-compatible**
   Machine outputs (reports/coverage) can be hashed/signed and attached to IF.TRACE receipts.

## Who pays for this (practical)

- Teams who must ship PDFs that survive scrutiny:
  - **GRC / security** (SOC2/ISO evidence packs, policy docs)
  - **research/publishing** (tech reports, standards commentary)
  - **legal/professional services** (deliverables that must look “court-ready”)
  - **vendor marketing with constraints** (docs that must be consistent across versions)

## Messaging that is honest

- “Markdown in, publication-quality out — with QA gates and receipts.”
- “It fails the build if the PDF is sloppy.”
- “Rules are paraphrased + pointer-backed; no book text shipped.”