Skip to content

Latest commit

 

History

History
64 lines (48 loc) · 4.92 KB

File metadata and controls

64 lines (48 loc) · 4.92 KB

OmniGlyph-க்கு பங்களிப்பது

உங்கள் ஆர்வத்திற்கு நன்றி! இந்த திட்டத்திற்கு இரண்டு பேரம்பேசவியலாத culture விதிகள் உள்ளன — அவைதான் README-இல் உள்ள ஒவ்வொரு எண்ணையும் நம்பக்கூடியதாக மாற்றும் காரணம்.

விதி 1 — Strict TDD

அனைத்து production code-உம் முதலில் தோல்வியடைந்த ஒரு testலிருந்து பிறக்கிறது:

  1. testஐ எழுதி அது சரியான காரணத்திற்காக தோல்வியடைவதைப் பாருங்கள்.
  2. அதைப் pass செய்ய தேவையான குறைந்தபட்சத்தை எழுதுங்கள்.
  3. green ஆக இருக்கும்போதே refactor செய்யுங்கள்.

முழு bar: pnpm run typecheck && pnpm test && pnpm run build — எப்போதும் மூன்றும் (docs link-lint மற்றும் rebrand guard pnpm testக்குள் இயங்குகின்றன tests/docs-integrity.test.ts மூலமாக).

விதி 2 — கூற்றுகளுக்கு முன் அளவீடு

Geometry, atlas, billing formula, அல்லது model scopeக்கான எந்த மாற்றமும் அளவிடப்பட்ட ஒரு எண் இல்லாமல் இறங்காது. இந்த repository இந்த கட்டுப்பாட்டைச் சுற்றி கட்டமைக்கப்பட்டுள்ளது:

  • Billing செலவு → benchmarks/billing-sweep/-உடன் நிரூபியுங்கள் (count_tokens இலவசம்; எதிர்பார்க்கப்படும் residual: பூஜ்ஜியம்).
  • Legibility → benchmarks/density-frontier/-உடன் நிரூபியுங்கள் (n≥30 ஒரு armக்கு, deterministic scoring, benchmarks/*/results/-இல் commit செய்யப்பட்ட JSONL ஆதாரங்கள்).
  • ஒரு production defaultஐ மாற்றுவதற்கான acceptance bar: gist == text baseline மற்றும் பூஜ்ஜிய silent exact-string errors மற்றும் நேர்மறை savings.

எண்கள் இல்லாத hypothesesகள் docs/ROADMAP.md-க்கு hypothesesஆகச் செல்கின்றன — ஒருபோதும் README-க்கு உண்மைகளாக அல்ல. இரண்டு "தெளிவான" யோசனைகள் ஏற்கனவே data மூலம் மறுக்கப்பட்டுள்ளன (high-res பக்கம், anti-aliased atlas); இந்த செயல்முறை வேலை செய்கிறது.

அமைப்பு

pnpm install
pnpm test              # full suite, ~40–90s
pnpm run dev:node      # watch modeஇல் local proxy

Node ≥18 (CI 20/22/24 testஆகிறது), pnpm 10 (package.jsonஇல் packageManager மூலம் pinned).

கட்டமைப்பு

folder விதி
src/core/ runtime-agnostic (Web APIs மட்டும் — Node மற்றும் Workersஇல் இயங்குகிறது)
src/node.ts / src/worker.ts host plumbing மட்டும்
benchmarks/ மறுஇயக்கக்கூடிய harnesses; JSONL results ஆதாரங்கள், committed
docs/ benchmarks/ (எண்கள்), architecture/ (map), ROADMAP (hypotheses), ops/ (OmniRoute)

Commits மற்றும் PRs

  • Conventional commits (feat:, fix:, perf:, docs:, refactor:, test:, chore:), relevant எண்களுடன் ஏன் என்பதை விளக்கும் body.
  • சிறிய, focused PRs; behavior மாற்றங்கள் அவற்றை pin செய்யும் testஉடனும், பொருந்தும்போது அவற்றை நியாயப்படுத்தும் benchmarkஉடனும் வருகின்றன.
  • Client-இன் cache_control blocksஐ மறுஎழுதாதீர்கள், விவாதம் இல்லாமல் runtime dependencies சேர்க்காதீர்கள் (core வேண்டுமென்றே dependency-light), render pathsஇல் Math.random/timestamps பயன்படுத்தாதீர்கள் (determinism ஒரு கடினமான invariant, byte-identity மூலம் tested).