It’s finally time I go crawling back to Node!
\nI’ve been using Node heavily this month on a SvelteKit client project. When did Node get so good‽ Deno has been my go-to runtime for so long I forgot how to Node. Now I’m back, I find all the ECMAScript<sup>†</sup> sugar is supported and the old annoying APIs have been replaced or modernised. Most importantly, I never have to see require().
<sup>†</sup> Doesn’t seem like that Oracle trademark dispute will see a positive end :(
\nThe official Node docs recommend piping an internet script straight to bash (we never learn) to install NVM to manage Node & NPM. My (ancient) experience with NVM and NPM hasn’t been stellar. I heard Fast Node Manager (FNM) was better to switch Node versions. Obviously I roll bleeding-edge but I have client projects that demand stability.
\nI opted for PNPM too to avoid getting immediately pwned. (The “M” in NPM stands for “malware.”) Some scripts I use have hard-coded binary names, so I added two aliases:
\nalias npm=pnpm\nalias npx=pnpxMaybe that’s a crime but so far it’s worked flawlessly.
\nPNPM also blocks post-install scripts. Does NPM still yolo those?
\nI added additional settings to pnpm-workspace.yaml to delay malware updates.
minimumReleaseAge: 1440\ntrustPolicy: no-downgradeAt first I tried setting the minimum release age to “one month” because it takes Microsoft at least that long to remove reported malware. This caused dependency issues where PNPM struggled to match suitable versions. I settled for “one day”; long enough to allow some other sucker to beta test the next Shai‑Hulud.
\nNode can now run TypeScript without throwing a tantrum like a baby if the stars don’t align. That said, one does not simply publish TypeScript packages to NPM.
\nerror: [ERR_UNSUPPORTED_NODE_MODULES_TYPE_STRIPPING]:\nStripping types is currently unsupported for files under node_modulesWhy? Just strip the types bro, I know you can! Let me sign a deal with the devil!
\nTo discourage package authors from publishing packages written in TypeScript, Node.js refuses to handle TypeScript files inside folders under a node_modules path.
This restriction is philosophical rather than technical. I get it though. TypeScript is a Microsoft product. Opening that floodgate would pollute the entire ecosystem. Nobody wants more Microsoft. I’d love to see light “native types” in ECMAScript. There are type annotation proposals. I suspect I’ll be retired before those bear fruit.
\nNo TypeScript packages mean I need to find the latest churnware slop to bundle my stuff. Tsdown did the trick, with only two additional dotfiles. Not thrilled about that (every dotfile represents a mistake). I suppose I break even after deleting deno.json etc.
Speaking of Microsoft lock-in, because they wrecked GitHub I’m self-hosting my own Forgejo instance. NPM limitations mean my packages have lost “provenance”. I had to configure the PNPM trust policy to allow my own stuff. Fun times!
\nMy final test for Node was converting my static site generator from Deno. Not many Node versions ago this would have required a major refactor. Today with Node v26.10.0 I found surprisingly little work to do.
The only required changes were to replace Deno’s file system API with node:fs — which is vastly improved from what I remember (literally ~10 years ago). Aside from that, I had to replace Deno.serve with Hono’s node adapter (a wrapper around node:http).
After this minimum-viable migration I was shocked to see 15% faster builds. My codebase still favours idiomatic Deno. I bet I’m leaving performance on the table by not using other built-in Node APIs. That’s something to explore later. The only further change I made was to replace Deno’s @std/path with node:path which is a straight import swap.
So if I were to TL;DR in the middle: Node got a glow-up, wow!
\nYou’ve probably known this for a while. I kept using Deno out of habit and familiarity. And I haven’t exactly been enthused to write server-side JavaScript recently.
\nI’m burying this part because it’s flogging a dead horse. Ultimately, Deno failed when they allowed the Silicon Valley Circus to define “success”. Deno went from an innovative modern JavaScript runtime to a boring start-up with uncompelling products. Half the employees were laid off and what’s left are tweeting AI fantasies and vibe-coding Temu Cloudflare.
\nThere is no reason to use the Deno runtime today. Deno Land Inc. stopped innovating that years ago. Node has slowly but surely caught up, even surpassing Deno in places.
\nWhat finally pushed me away was:
\nBasically stuff that made it borderline unusable on top of my other criticism. JSR support were very quick to delete my account on request. I don’t like leaving dead profiles around the internet. None of my packages are visible but old versions remain installable.
\nIt was fun early on but now it’s time to say goodbye.
\nbrew uninstall deno