<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:content="http://purl.org/rss/1.0/modules/content/"><channel><title>构建 on Lsekit's Blog</title><link>https://blog.001k.site/categories/%E6%9E%84%E5%BB%BA/</link><description>Recent content in 构建 on Lsekit's Blog</description><generator>Hugo</generator><language>zh-CN</language><lastBuildDate>Sat, 11 Jul 2026 15:52:00 +0800</lastBuildDate><atom:link href="https://blog.001k.site/categories/%E6%9E%84%E5%BB%BA/index.xml" rel="self" type="application/rss+xml"/><item><title>npm 的 prepare 钩子如何让构建跑了两遍</title><link>https://blog.001k.site/npm-prepare-double-build/</link><pubDate>Sat, 11 Jul 2026 15:52:00 +0800</pubDate><guid>https://blog.001k.site/npm-prepare-double-build/</guid><description>CI 里构建时间翻倍，日志出现两段相同的编译输出。用 npm 11.19.1 做最小复现，数清 prepare 在各命令下的触发次数，再给三种修法。</description></item><item><title>Go 编译时的 GOAMD64 到底影响什么</title><link>https://blog.001k.site/go-goamd64-explained/</link><pubDate>Sat, 29 Nov 2025 10:18:00 +0800</pubDate><guid>https://blog.001k.site/go-goamd64-explained/</guid><description>结论先摆：GOAMD64=v3 不会向量化循环，它换来的是 TZCNT/LZCNT。用 go1.27.3 编译两遍，反汇编逐类数指令，把差异定位到具体文件。</description></item></channel></rss>