11thDwarf πŸ¦€1d ago
rate limiting is one of those things every API needs but half implement wrong. fixed window leaks at the boundary. sliding window is better. token bucket handles bursts best. return 429 with Retry-After β€” that one header saves your consumers so much pain. #apidesign #developer #programming
sayzard4d ago

You can't whisper at an AI agent

Stripeμ—μ„œ AI μ—μ΄μ „νŠΈκ°€ APIλ₯Ό μ˜¬λ°”λ₯΄κ²Œ μ‚¬μš©ν•˜λ„λ‘ μœ λ„ν•˜λŠ” μ‹€ν—˜μ„ μ§„ν–‰ν•œ κ²°κ³Ό, μ—μ΄μ „νŠΈλŠ” κ²½κ³ λ‚˜ λΆ€λ“œλŸ¬μš΄ μ‹ ν˜Έλ₯Ό λ¬΄μ‹œν•˜κ³  λͺ…ν™•ν•œ 였λ₯˜λ‚˜ 직접적인 μ§€μ‹œκ°€ μžˆμ„ λ•Œλ§Œ 행동을 μˆ˜μ •ν•˜λŠ” κ²½ν–₯이 있음이 ν™•μΈλ˜μ—ˆλ‹€. λͺ¨λ“ˆν™”λœ μŠ€ν‚¬ 파일과 CLI 둜그인 μ‹œ μŠ€ν‚¬ μ„€μΉ˜ μœ λ„λŠ” νš¨κ³Όμ μ΄μ—ˆμœΌλ©°, μ—μ΄μ „νŠΈ λŒ€μƒ 개발자 κ²½ν—˜μ€ μ½˜ν…μΈ λ³΄λ‹€ 배포가 ν•΅μ‹¬μž„μ„ μ‹œμ‚¬ν•œλ‹€. λ˜ν•œ, μ—μ΄μ „νŠΈλŠ” νƒμƒ‰ν•˜μ§€ μ•Šκ³  주어진 μ»¨ν…μŠ€νŠΈ λ‚΄μ—μ„œλ§Œ μž‘μ—…ν•˜λ―€λ‘œ, μ—μ΄μ „νŠΈ μΉœν™”μ  μ„€κ³„λŠ” ν•˜λ“œ μŠ€ν‹°μ–΄λ§(였λ₯˜, λͺ…μ‹œμ  μ§€μ‹œ)에 집쀑해야 ν•œλ‹€λŠ” 결둠을 λ‚΄λ Έλ‹€.

https://stripe.dev/blog/ai-steering-experiments

#aiagent #developerexperience #apidesign #cli #agentsteering

You can't whisper at an AI agent

What a dozen steering experiments taught us about how LLM-based agents actually behave in the wild.

Miriam4d ago

GraphQL vs RESTful Fitment Architecture Costly Lie Exposed

A single GraphQL query can replace three legacy calls, cutting integration headaches for parts catalogs. See how the switch solves mismatches and speeds time‑to‑marketβ€”no more data silos.

https://partsfit.online/graphql-vs-restful-fitment-architecture-costly-lie/

#graphql #fitmentdata #crossplatformcompatibility #automotivepartsintegration #apidesign

GraphQL vs RESTful Fitment Architecture Costly Lie Exposed

Discover why the REST myth about fitment data is wrong and how GraphQL delivers faster, safer, cross‑platform integration for automotive parts catalogs.

Parts Fit
sayzardMay 11

I keep tripping over "true, false, true"

μ½”λ“œμ—μ„œ μ—¬λŸ¬ 개의 λΆˆλ¦¬μ–Έ 인자λ₯Ό μˆœμ„œλŒ€λ‘œ μ „λ‹¬ν•˜λŠ” 방식은 가독성을 크게 λ–¨μ–΄λœ¨λ €, ν•¨μˆ˜ 호좜 μ‹œ 각 인자의 의미λ₯Ό κΈ°μ–΅ν•˜κ±°λ‚˜ ν•¨μˆ˜ μ •μ˜λ₯Ό λ‹€μ‹œ 확인해야 ν•˜λŠ” λΆˆνŽΈν•¨μ„ μ΄ˆλž˜ν•œλ‹€. μ €μžλŠ” 이런 'flag arguments' 문제λ₯Ό ν•΄κ²°ν•˜κΈ° μœ„ν•΄ μ˜΅μ…˜ 객체λ₯Ό μ‚¬μš©ν•˜λŠ” 방식을 μΆ”μ²œν•˜λ©°, λ•Œλ‘œλŠ” λΆˆλ¦¬μ–Έ ν”Œλž˜κ·Έκ°€ 숨기고 μžˆλŠ” λ‹€λ₯Έ 행동을 λͺ…ν™•νžˆ κ΅¬λΆ„ν•˜λŠ” 별도 ν•¨μˆ˜λ‘œ λΆ„λ¦¬ν•˜λŠ” 것이 더 λ‚«λ‹€κ³  μ„€λͺ…ν•œλ‹€. TypeScript의 νƒ€μž… μ •λ³΄λ§ŒμœΌλ‘œλŠ” 이 문제λ₯Ό ν•΄κ²°ν•  수 μ—†μœΌλ©°, λͺ…ν™•ν•œ API 섀계가 μ€‘μš”ν•˜λ‹€κ³  κ°•μ‘°ν•œλ‹€.

https://allthingssmitty.com/2026/05/11/i-keep-tripping-over-true-false-true/

#softwareengineering #javascript #typescript #apidesign #codereadability

I keep tripping over

createUser(user, true, false) works. It's also surprisingly hard to read. Here's why I've stopped writing APIs like this in JavaScript.

sayzardMay 7

Idempotency Is Easy Until the Second Request Is Different

이 글은 API μ„€κ³„μ—μ„œ μ€‘μš”ν•œ κ°œλ…μΈ λ©±λ“±μ„±(idempotency)의 λ³΅μž‘μ„±μ„ 닀룬닀. λ‹¨μˆœνžˆ ν‚€-κ°’ 쑰회둜 멱등성을 κ΅¬ν˜„ν•˜λŠ” 것은 μ‰¬μš°λ‚˜, 동일 킀에 λŒ€ν•΄ λ‹€λ₯Έ μš”μ²­μ΄ λ“€μ–΄μ˜¬ λ•Œ λ°œμƒν•˜λŠ” λ¬Έμ œμ™€ 이λ₯Ό μ²˜λ¦¬ν•˜λŠ” 방법에 λŒ€ν•΄ μƒμ„Ένžˆ μ„€λͺ…ν•œλ‹€. 특히, μš”μ²­μ˜ 동등성 νŒλ‹¨, μž¬μ‹œλ„ μ •μ±…, 응닡 μž¬μƒ, 그리고 λ©±λ“±μ„± ν‚€μ˜ λ²”μœ„ μ„€μ •κ³Ό ν•΄μ‹œ 계산 방법 등이 핡심 주제둜 닀뀄진닀. λ©±λ“±μ„± κ΅¬ν˜„ μ‹œ λ°œμƒν•  수 μžˆλŠ” λ‹€μ–‘ν•œ μ˜ˆμ™Έ 상황과 이λ₯Ό ν•΄κ²°ν•˜κΈ° μœ„ν•œ λ°μ΄ν„°λ² μ΄μŠ€ 섀계 및 API 응닡 μ „λž΅λ„ μ œμ‹œν•œλ‹€. μ΄λŠ” μ‹ λ’°μ„± 높은 API 섀계에 ν•„μˆ˜μ μΈ λ‚΄μš©μ΄λ‹€.

https://blog.dochia.dev/blog/idempotency/

#idempotency #apidesign #reliability #http #softwareengineering

Idempotency Is Easy Until the Second Request Is Different | Dochia CLI Blog

Idempotency is not just an HTTP header or a key lookup. This article covers the failure cases that bite real APIs: different requests with the same key, concurrent retries, partial success, downstream uncertainty, response replay, expiry, and duplicate message handling.

JAVAPROMay 6

Ever shipped an API and regretted your generic signatures later? Wildcards everywhere. Confusing bounds. Mental overhead. Michel Charpentier breaks down why variance mattersβ€”and why #Java still feels heavy here.

Read + apply: https://javapro.io/2026/01/27/what-i-still-miss-my-most-wanted-java-features/

#APIDesign #Generics #JVM

Aryan JaswalMay 4

Most devs think backend = APIs.
It’s not.
It’s:
β€’ Efficient request handling
β€’ Clean architecture
β€’ Smart DB design
β€’ Caching strategies
β€’ Security
β€’ Reliability under load
Great backend β‰  just code
It’s systems that don’t break in the real world.
Tools change. Principles don’t.

https://jaswalaryan.space/article/backend-development-beyond-apis-complete-guide

#BackendDevelopment #WebDevelopment #APIDesign #SoftwareEngineering #SystemArchitecture #DatabaseDesign #Caching #Security #PerformanceOptimization #DevOps #Scalability #CodeQuality #Programming

Show threadNicola AsuniApr 16
If this resonates, share your service template patterns or open an issue with gaps you hit in production.
If gogen helps, a star and field feedback help prioritize what to improve next.
#OpenSource #GoLang #BackendEngineering #Observability #APIDesign #CloudNative #DistributedSystems
TechinoApr 12
Why did Anthropic quietly reduce the Claude Code prompt cache TTL from 1 hour to 5 minutes? This caused a 20-32% spike in costs for users and wasn't documented. Makes me wonder about API transparencyβ€”do we need better standards for communicating infrastructure changes that affect pricing? Curious what others think about silent changes vs. proactive updates #AI #APIDesign #DevTools #OpenSource #Anthropic
Phil NortonApr 9

Currently integrating with an API that uses pagination.

To get the next page of results you just provide a page=x parameter.

The number of pages info is in the results, so you can use that to build the pagination engine.

What happens if you go over the number of pages available? You get the last page of results again, with a 200 response code.

Fair play I guess. A simple trap to catch the idiot developer who can't count :D

#api #apiDesign