Backend engineer in Taipei. Go & Postgres by day; writing at night.https://terrychengtw.com/zh-Hanthttps://terrychengtw.com/posts/why-i-started-writing/https://terrychengtw.com/posts/why-i-started-writing/去年我花兩週解掉的 deadlock，我完全忘了。翻回當時的筆記，看到的自己像個陌生人。這篇是我給未來自己的備忘。Thu, 02 Apr 2026 00:00:00 GMTmetaessayhttps://terrychengtw.com/posts/go-concurrency-shapes/https://terrychengtw.com/posts/go-concurrency-shapes/大部分 Go 並發的 bug 都是「用錯姿勢」，不是工具本身有問題。這篇整理我看過的三種最常見模式。Fri, 20 Mar 2026 00:00:00 GMTgodeephttps://terrychengtw.com/posts/ddia-ch7-transactions/https://terrychengtw.com/posts/ddia-ch7-transactions/邊讀邊寫的筆記，不是完整文章。主要記我原本以為懂、讀完才發現不懂的那些。Wed, 11 Feb 2026 00:00:00 GMTreadingnotehttps://terrychengtw.com/posts/neovim-2026/https://terrychengtw.com/posts/neovim-2026/每兩年重寫一次 config。這次砍掉 40% 的 plugin，反而比較順手。Mon, 05 Jan 2026 00:00:00 GMTtoolsessayhttps://terrychengtw.com/posts/postgres-locking-surprises/https://terrychengtw.com/posts/postgres-locking-surprises/SELECT FOR UPDATE 的鎖範圍不是你想的那樣。一篇關於 row-level lock、advisory lock、與我踩過的三個坑。Sun, 14 Dec 2025 00:00:00 GMTpostgresdeephttps://terrychengtw.com/posts/engineer-writing-curve/https://terrychengtw.com/posts/engineer-writing-curve/把寫作當肌肉練。從 commit message 到 postmortem 到文章，中間的卡點在哪。Sun, 02 Nov 2025 00:00:00 GMTmetaessayhttps://terrychengtw.com/posts/note-on-timeouts/https://terrychengtw.com/posts/note-on-timeouts/最近才想清楚：client timeout、server timeout、context timeout 三者該怎麼配合。Sat, 18 Oct 2025 00:00:00 GMTsystemsnotehttps://terrychengtw.com/posts/slow-query-weekend/https://terrychengtw.com/posts/slow-query-weekend/explain analyze 從 300 行讀到 0 行，中間換了三次假設。這是慢 query 修復的完整過程。Fri, 05 Sep 2025 00:00:00 GMTpostgresdeephttps://terrychengtw.com/posts/stopping-reaching-for-orm/https://terrychengtw.com/posts/stopping-reaching-for-orm/不是 ORM 不好，是 default 選擇太便宜，常常遮掉了資料庫真正在做的事。Sun, 10 Aug 2025 00:00:00 GMTgoessayhttps://terrychengtw.com/posts/moving-to-hetzner/https://terrychengtw.com/posts/moving-to-hetzner/一個下午的遷移紀錄，包含一個我忘記備份 DNS 的尷尬時刻。Tue, 22 Jul 2025 00:00:00 GMTopslog