Vercel Python 函數套件大小限制提升至 500MB

🔑 增強重點摘要

• •Vercel's Python runtime lacks automatic tree-shaking, bundling all reachable project files by default, which often leads to oversized deployments requiring manual exclusions via configuration.[3]
• •The 250MB limit stemmed from AWS Lambda constraints minus Vercel overhead, prompting users to tree-shake dependencies or switch to custom Lambda images for larger needs.[1]
• •Community reports showed Python apps with libraries like Plotly and pandas frequently exceeding 250MB (e.g., 360MB), causing deployment failures on Hobby plans.[6]

🛠️ 技術深入

• •Python Functions bundle all project files reachable at build time without Node.js-style tree-shaking; developers must limit pyproject.toml or requirements.txt to runtime essentials and exclude files like tests or static assets.[3]
• •Unzipped bundle must not exceed the limit post-build (previously ~50MB compressed equating to 250MB uncompressed); dynamic imports or large deps like Puppeteer historically caused issues across runtimes.[4][5]
• •Runtime memory defaults to 1024MB on Hobby but is separate from bundle size; total env vars limited to 64KB combined for Python deployments.[1][2]

🔮 前景展望AI analysis grounded in cited sources

⏳ 時間線