Show HN: Onlyrecipe 2.0 – I added all features HN requested – 4 years later

Original Article ↗ Hacker News Discussion ↗

Core value and reception

  • Many commenters are enthusiastic: the tool solves a widely-felt pain of cluttered, ad-heavy recipe sites and AI-churned content.
  • People like the clean layout, grocery-list integration, scaling, timers, cross-device sync, and that it works in a regular browser (not just mobile apps).
  • Several say this replaces homegrown attempts they’d been building; Paprika is cited as the main incumbent.

Parsing coverage and reliability

  • Works well for many sites but fails on some that lack standard recipe schema (e.g., certain blogs).
  • Author confirms primary use of JSON-LD/schema data; is adding an LLM-based fallback for “schema-less” pages.
  • Some specific misparses: yields (“2 dozen” → “2 servings”), ingredient units (“12 ounces” → “12 ounce”), instructions that don’t scale, and non-amount number words getting converted inside directions.
  • Importers from other apps (Paprika, PlanToEat) and CSV are requested and partially implemented.

UX, accessibility, and performance

  • Web app criticized for:
    • Hijacking the browser back button and creating circular history / infinite loops.
    • Laggy animations and heavy transitions that make navigation feel slow.
    • Poor keyboard navigation (arrow/PgUp/PgDn), oversized images, and issues with desktop layout (line breaks, multi-page prints).
  • Inability to select text for copy/paste raises accessibility and screen-reader concerns.
  • Signup and email confirmation initially broken; password manager autofill hints are missing.
  • Requested enhancements:
    • Print-only core recipe (no photos/metadata), Windows app, ingredient highlighting in steps, random recipe / recipe-of-the-day, clearer pricing before signup, better scaling (including <1x) and syncing of quantities within instructions.

Unit conversion and measurement debate

  • Conversion engine currently over-precise (e.g., 391.32 g) and sometimes wrong (Greek yogurt density) or blindly converts where it shouldn’t (“cut in 0.50 lengthwise”).
  • Long subthread debates:
    • Rounding vs precision; most agree 2–3 significant figures is enough.
    • Cups vs grams, especially for flour and bread: some say volume is fine; several bakers argue only weight yields consistency.
    • Metric vs imperial defaults; different cup/tablespoon standards by country; spoons and cups still common in Europe.
  • Suggestions: round values, keep tbsp/tsp, show both volume and mass, prefer weight where possible.

Implementation and infrastructure challenges

  • Author notes major unexpected complexity in subscription handling (trials, upgrades, cross-device sync), ultimately offloaded to RevenueCat.
  • Self-hosting Supabase and setting up reliable email (AWS SES, spam/reputation issues) described as painful.
  • Scraping around Cloudflare uses rotating datacenter IPs and a residential fallback (Raspberry Pi at home); commercial residential proxy services are mentioned.