Antti
3.2K posts

Antti
@Hiljaa
DNA infused sentient collection of atoms. Optimizing synergies around the world. Word person.
Berlin Katılım Mart 2009
857 Takip Edilen284 Takipçiler

We try to make things "just work" as much as possible for our users. We ship with Node because a ton of language servers and now MCP servers are JS-based, including the one we use for our JSON settings.
However, we understand auto-downloading is not ideal, especially given the recent supply chain attacks. It's something we're aware of, and we want to solve this in a way that we aren't introducing undue friction for our users who appreciate having this functionality built in.
As always we'd love help: github.com/zed-industries…
English

"But Linux package maintainers are so needed" and here we see Arch packagers hard-require a supply chain provider when you install Zed editor, even if you never plan on using it for web projects.
Of course, Zed is a little bit to blame here - by default, Zed ships with Prettier enabled for JSON and markdown files, so what it'll do even if you just download it from official site - it'll detect that prettier needs to be installed, it'll download bundled nodejs and npm from nodejs site, and then will install prettier from npm (see screenshots). Completely behind your back!
Now, if Linux package maintainers actually upholded their self-assigned Importance Mission, here's what they would have done:
1) Patched Zed to disable prettier by default
2) Provide npm and node as _optional dependencies_
3) Patch out managed node/npm download from zed
But instead they just decided to cope out (they will say excuses like "unpaid labor" and stuff) and subject ANYONE who wants to use Zed even for Rust/C/C++ programming to security nightmare that is npm.
So yeah, Zed does some shady stuff for sure - in my humble opinion, a text editor SHOULD NOT DOWNLOAD STUFF FROM THE INTERNET BY DEFAULT BEHIND USER'S BACK.
On the other hand, huge L for Arch Linux.




English

@valigo @ryanflorence Have you ever baked cookies at home? They've got plenty of butter :)
English

@ryanflorence all these junk food places (and honestly even mid-tier restaurants) use ungodly amount of oil that is not very good for you. And I guess it's even worse in America.
English

@buildwithdanm Oxc is faster and has more rules, and js-plugins. But it's only faster if you don't use js-plugins, which you probably will use.
English

@zeddotdev Please make TypeScript lsp go brr too :) The new native lsp is in beta.
English

@devongovett @theorangenotice oxlint is faster than biome, but when you start using oxlint your lint time will most likely be slower, because you'll start using js-plugins :)
English

@theorangenotice does it support running eslint plugins? we have a lot of custom rules that we need to enforce. oxlint worked with them without any changes which was nice.
English

@devongovett Fun fact, eslint-plugin-react-x is doubly faster than eslint-plugin-react-hooks and has almost all the same rules.
English

@iamsahaj_xyz I'm not the only one thinking this! Been thinking of adding a dedicated cursor parking spot.
English

I was planning to use UUIDs to represent the databases we have on the @tursodatabase Cloud. I got a bit worried that we would perhaps run out of UUIDs. I just double-checked and I think we'll be fine for the next year or so.
Will use UUIDs for now, and if needed, rearchitect later.

English

@ryanflorence Oh yes, full of juicy corner cases. If you throw in retries and retry error messages. How to store them, where's the state, state updates for multiple processes, update favicon based on state, what if you deploy while task is running, authing and rate limiting the socket...
English

@zeddotdev Might be asking too much but make toggling JSX comments sane please :)
English

@jdxcode @OneManSaas Got `fetch` and `deploy` commands too, nice :) So far those two have been pnpm's moat basically.
English

Just yesterday for the millionth time I opened the notification panel on accident in @zeddotdev while trying to open the agent panel and stopped to think what does this thing even do? Guess nothing :)

English

@DavidKPiano history, context, canvas
I think some variation of it is the final UI
English

@beeman_nl @zkochan @pnpmjs @NxDevTools @turborepo Nice. I've been using proto version manager from @tothemoonrepo. Very handy.
English

@Hiljaa @zkochan @pnpmjs @NxDevTools @turborepo A global install and per project.
Codex found it and fixed it for me though, check here
beeman 📱@beeman_nl
Codex fixed my issue! 🥳 It tracked my mystery pnpm v11 symptoms to: - a hidden pnpm@11.0.0-rc.0 inside fnm’s Node install - Corepack pointing to 11 And my shell still reported pnpm 10 and wrote a v11 store 🧐 Thanks to @zkochan and @anthonysheww for their support! 🙏
English

It looks like @pnpmjs has been changing it's lockfile format (in a minor release).
I've been getting warnings from both @NxDevTools and @turborepo 👀
Very inconvenient and pretty annoying - seems it should have been a breaking change, not ship in a minor version bump.

English

@mitchellh Simple landing pages are great, concentrate shipping the feature first, then polish the landing page :)
English

One of the most requested GitHub features in years and the website looks like it was designed by someone 9 years into a 2 year community college program. github.github.com/gh-stack/
English

This screenshot is from my root directory that doesn't have a package.json
I do see this (screenie 1) appear in one of my projects which seems to hint at v11.
If I look into the ~/Library/pnpm/pnpm (screenie 2) it seems to be linking to 10.33.0.
In `~/Library/pnpm/store/*` I see v11 did appear.
I'm fairly sure I didn't update to v11 as I don't like running beta versions on my daily driver, I'm confused how it seems to have sneaked in here.
Also both projects that have this symptom have a packageManager set in package.json which should make it use that version, right?
Any tips that I can check where it is and get rid of it? Really appreciate your response 🙏




English








