Given the news of Deno being acquired by Cloudflare and its already stated eventual sunset, someone has to take over deno.
Vercel certainly has the resources, I am looking for a job and would love to work on Vercel's deno maintenance (which I suggest should be renamed to "veno", or "vino")
It took nearly $30 million in VC to fund Deno's existence and it got dropped like a rock by Cloudflare so there's probably nothing financially-attractive about the software and no hope of self-sustainability.
Vercel taking it over would be a bandaid solution, back to square one as soon as they want to cut expenses or a PE firm makes an offer their investors can't refuse. It probably wouldn't survive this decade there.
The Apache Foundation is good for projects like this.
I used it for a while, it was a sane TypeScript. Most pieces came out of the box. Deno had its own formatter, test runner, standard library, server, could run TS without a transpilation step, KV DB, and god knows what else.
Switching between Deno and standard Node/TS had me thinking why can't all of Node be this nice. Over time Node adopted some of these ideas.
Deno also made some interesting choices around dependency management, its first iteration being literal links in the import statement. This bit and subsequent npm compatibility is guaranteed flamewar material.
sane and boring was my biggest takeaway too. Felt so nice to not have to do the version pinning ritual every few weeks because one of the half dozen dev interconnected dev components updated. Very sensible and complete environment.
They were great at standards too; if it wasn't in the Deno namespace it was standardised. Import maps and Temporal early were cool and i Learned a lot reading release blogbosts that linked to tc39 proposals.
It had some achilles heels with vite & meta frameworks, linter (there was a draft PR to use oxlint which would fix that) and the language server wasn't great but I'm sad to see it die over the last year, the writing was well and truly on the wall before officially closing.
From many accounts they also had fantastic webGPU support and they released Deno desktop not long before shutting down.
I liked it when it first came out and was a substantial DX improvement over node. Unfortunate then that it got undercut by Bun, which had a very similar 'faster, less legacy, and batteries included' value prop. That split the just-not-node audience, and ended up ensuring that neither one ever really got the scale needed to survive independently...
They need a non-profit foundation, not another VC-backed cloud vendor that could drop it just as easily.
Big if, but if they set up a foundation and it has enough steam to keep improving it could be a renaissance for Deno being no longer connected to a company that needs to focus on being a temu Vercel so it can be a stable, secure runtime that attract engineers that like typescript but don't want the node/npm quagmire that were, quite reasonably and rightfully in hindsight, turned off by Deno company.
Stable funding + a 3.0 with; oxlint and some rough edges with vite and LSP integration and I would be convinced to go back. (@std being bundled would be amazing too)
Vercel taking it over would be a bandaid solution, back to square one as soon as they want to cut expenses or a PE firm makes an offer their investors can't refuse. It probably wouldn't survive this decade there.
The Apache Foundation is good for projects like this.
What is the use case, allure, etc.?
I’d prefer to get your thoughts than ask GPT, since you seem to have experience with it.
Switching between Deno and standard Node/TS had me thinking why can't all of Node be this nice. Over time Node adopted some of these ideas.
Deno also made some interesting choices around dependency management, its first iteration being literal links in the import statement. This bit and subsequent npm compatibility is guaranteed flamewar material.
They were great at standards too; if it wasn't in the Deno namespace it was standardised. Import maps and Temporal early were cool and i Learned a lot reading release blogbosts that linked to tc39 proposals.
It had some achilles heels with vite & meta frameworks, linter (there was a draft PR to use oxlint which would fix that) and the language server wasn't great but I'm sad to see it die over the last year, the writing was well and truly on the wall before officially closing.
From many accounts they also had fantastic webGPU support and they released Deno desktop not long before shutting down.
You can build a server with a local database with zero external dependencies, it has JSX and SQLite built in.
Vercel will place in platform specific features or that work only on Vercel.
It makes no sense for Vercel.
Big if, but if they set up a foundation and it has enough steam to keep improving it could be a renaissance for Deno being no longer connected to a company that needs to focus on being a temu Vercel so it can be a stable, secure runtime that attract engineers that like typescript but don't want the node/npm quagmire that were, quite reasonably and rightfully in hindsight, turned off by Deno company.
Stable funding + a 3.0 with; oxlint and some rough edges with vite and LSP integration and I would be convinced to go back. (@std being bundled would be amazing too)
The funny part is I didn't even know Deno existed until I saw this post. Then I did some research on it.
If builders don't even know you exist what is the point???