GPUI 2 is now in production – Zed | Hacker News Reader

GPUI 2 is now in production – Zed

zed.dev

50 pointsDAlperin18 comments

I'd love to try this, though the large number of projects that only support MacOS perplexes me. This excludes at least 80% of people with computers that would like to use your software and effectively limits it to a class of folk with money, creating a sort of nasty exclusivity to the software that does this when there doesn't seem to be a platform-limiting factor.

Their Windows Support Issue was opened on Jun 29, 2022 and the Linux one on the same day, so they're just under two years old. It seems that this isn't source-available right now either, though they mention they'll open source it at some point, but it prevents contributions toward this end too.

The websites this (https://github.com/zed-industries/community/issues/2197) list as the reason for blocking the other OSes though it's quite sparse and doesn't say much.

Genuine and non-argumentative question; why? Easier to focus on one platform for the purposes of achieving good stability before supporting more than one OS, or trying to get a userbase that is typically more willing to pay for your software? Investor pressure?

Edit: Another comment mentioned the FAQ. So it is about validating the project toward the end of making money it seems. I guess they chose Mac because Apple people have a history of being more willing to pay for stuff.

PowerPoint, Photoshop, Excel, Illustrator, Lightroom, Premiere, Atom, and After Effects were all released first on the Mac. In fact the only applications I'd call landmarks that weren't initially released on the Mac are Word and VS Code, which weren't released first on the Mac for obvious reasons.

Your argument is that products that start on Mac and later get acquired and ported by megacorps do better?

This is a text editor similar-ish to VS Code, though Code is made by MS so yeah. Sublime’s first release was on Windows.

Atom supported all three above at release.

Other landmark software not launched exclusively on Mac first; Steam, Google Chrome(though I guess based on WebKit, itself based on KDE-built KHTML, so I guess WebKit can be added to the list).

btw, I got a Quest 3 recently. I saw that you work in that area, so thanks for the awesome software. Quite enjoying it so far.

I assume it’s a few reasons:

* Project’s developers use Macs - you are going to code to what you use. If the source was available then I’d say the onus is on developers for the other platforms to port it if they want it

* Simpler target - both in hardware and software Mac as a platform is easier to target IMHO. Like some of the new AI stuff that gets ported to Mac or built for Mac, even if the hardware isn’t the best available (GPUs mainly but also CPU) it’s a known quantity and set configuration. Every M3 is like every other M3. Much easier than “Hey I have 2 ABC123 video cards, how do I use them with your library?”

* Smaller problem space - instead of having to deal with the lowest common denominator or edge cases in other OSes you can focus on just 1. That greatly simplifies things.

Windows may be 80% but the configurations are wildly different and Macs are far more homogenous (with less baggage due to less focus on backwards compatibility). You might be hard pressed to find a subset of the windows population who all have a similar homogenous makeup that rivals the Mac size. Similarly an iPhone is an iPhone is an iPhone, the same is not true at all for Android even though its market share is much higher (aside from the US).

I've been using Zed for the last few months as my secondary driver. Still miss some features that I really find helpful in VSC, such as showing the Tailwind class on hover in Elixir LiveView code, but apart from that, Zed is really good and smooth, specially when it comes to collaboration. Collaboration just works, including a clear audio stream.

One thing I do use Zed heavily for, though, is their assistant panel to interact with GPT-4-Turbo. That panel is really well done (its essentially a text editor), and using GPT-4 on a pay-per-request basis instead of $20/month for ChatGPT is just much more cost-effective for me. The panel stores all previous conversations in a folder on your system too, so it's easy to grep through it if required. You can even edit the model's reply, just like on the OpenAI Playground.

If Zed plan to open source the editor, what is the monetisation strategy? Paid features like "multiplayer"? How does that feature work for companies that don't allow code to leave their networks?

> GPUI rasterizes the entire window on the GPU

Other programs I have seen do this on Linux and Windows tend to have blurry fonts. I hope that is not also the case here.

P2P networking or self-hosting?

I really love Zed, was my primary driver.

Then I went to update to a newer version and Zed stopped working and I can’t revert back to an older version. I see someone else also has the same bug filed. Hope they can figure out the fix. I miss using Zed.

Zed is great, have been using it to do the Rustlings exercises and learn Rust:

https://github.com/rust-lang/rustlings

If you've been looking for an excuse to learn Rust, check it out.

I see the measurements on the homepage lists zed and sublime as basically the same. Anyone know how they measure - and how helix and neovim compare?

I'm not bothered at all with helix startup time...

Nothing about whether they might support other OSs :/

They mention it in their Issues on GitHub though it appears no work has been done toward that.

From their FAQs:

  We are a small team, so it's critical for us to be laser-focused. As a startup, one of our key priorities at this early phase is learning, and right now, we're focused on the following questions:

    What are the key features we need to get traction on any platform?
    Are our assumptions about our eventual business model valid?

  While we'd love to support users on Linux and Windows, adding those platforms doesn't really help us answer those questions.

https://zed.dev/faq

Thanks. I went off of what they had posted on GitHub and the linked issues and reasons.

I wouldn't say nothing:

"We've also laid down a solid foundation for upcoming improvements such as multi-platform support and animation."

This is terribly off topic, but I immediately thought this was Zed Shaw and became genuinely excited that he was jumping into AI (and out of, iirc, painting and music). While I don’t consider myself a fanboy (fanman?) of anyone, I definitely admired him/his code some of which I ran in production. It has been a while since I’ve seen him post here.

So instead of making an app people can actually use (outside of the 10 people with Macs) they're rewriting large parts of it instead?

Makes sense..

Read literally the first paragraph and you'd already see that's what this is a precursor to.

> We've also laid down a solid foundation for upcoming improvements such as multi-platform support

The old UI may have been terribly written to support anything other than macOS. A rewrite is the right play if you identify that.

Reply on news.ycombinator.com