Prisma is removing Rust?

Over the years I've moved away from Prisma towards Drizzle. There are tons of things I miss, but one thing I do not miss is the performance, which is why I'm so hyped they're removing Rust. I promise it makes sense.

SOURCE

S/O Ph4seon3 for the awesome edit 🙏

[PraiseYeezus]

To be honest...after all these years, I still can't think of a single reason to ever use Prisma over a traditional query builder like Kysely or an ORM like Drizzle. These guys feel like they're selling us on a problem that doesn't exist. I don't have a problem with SQL. I just want type safety for it, and intellisense is nice. That's it.

[vipzip]

0:00 Prisma's Evolution and Future Roadmap
1:00 SingleStore Database Recommendation
2:35 Prisma's Manifesto: Vision for the Future
3:20 Prisma's History and Pivot from Graphcool
5:20 Prisma's Changes: Core Logic Migration and Focus on Supported Databases
12:20 Improved Issue Management and Community Engagement
13:20 TypeScript Migration for Increased Extensibility
18:00 Maintaining a Strong and Transparent Community Connection

Generated by Snorvia AI chapter generator

[MatthiasFeist-de]

Unfortunately Drizzle is also pretty bad at responding to issues. Even the ones labelled as priority aren't fixed. They are very focussed on onboarding new partners. So a lot of parallels to Prisma

[aaronv]

If anyone is interested, the reason for performance bottlenecks was the serialization taking place to pass data across the FFI boundary to rust, not rust itself.

[sergioengineer]

I believe that there's a big difference between prisma and turbopack. For prisma, rust was a hindrance in more than just one way and, besides, there are other aspects which impacted the low contribution numbers. If you wanted to change even the smallest thing, You had to go out of your way to discuss with two teams about the api design because of the architecture. I was going to send a PR to them a couple of years ago, but then I realized that there were thousands of unanswered PRs. Mine would be just one more on the stack. Hours of work wasted. Months after I decided to give up, they marked the issue as "not planned" through one correlated issue.

For prisma rust also didn't offer any benefits. Querying the database is a very I/O bound operation. It is not about JS' JIT taking too long to compile all the different versions of a function because it received undefined | null | number | string | object. Rust(a queryClient) was just a weird requirement which people would find out about once they tried to use it on a serverless context.
That's not the case for turbopack. The performance gains are extremely important for DX. They are a real benefit and not just a random quirk waiting to surprise people.

[LawJolla]

Fun fact: Prisma 1 was actually written in Scala (ripped from GraphCool) and required the user to maintain their own JVM server.

[boredstudent]

Reminder that the issue count on flutter doesn't say anything, they Spam themselves from CI. At least 60% of them just say that some test failed on some feature branch.
That desicion it self is kinda weird and maybe questionable, but people don't actually have *that* many issues with flutter.

[mornwind]

I’ve removed Prisma from any possibile future project so that checks out.

[rakinar]

Finally they did it. They made the ORM much simpler now. Previously the Rust part of the ORM was pretty much just an isolated part, which was hard to integrate and extend. But now it's all written in the same language. I never saw any specific benefit of using Rust in a JS/TS ORM, its just adding complexity unless there's a good reason to do so.

[SilentEdits]

The issue was the architecture, not Rust.

[mlwElrohir]

I'll call you out on this one Theo, you are comparing Flutter issues while evangelizing for Drizzle which has 1.1k+ issues opened and is barely usable in the production environment (contributed to drizzle and migrated away to Kysely) is rather short-sighted.

[januslast]

The great thing about this channel is that we can check in every 5 years or so, and look back on all the tech that was hyped up and then dropped. So then we can stay away from it. Javascript is still hot. Typescript, not so much. We'll see where Rust ends up... after a few years. It's so great that so many companies are spending their $ on this kind of research. This is the true "open source" community.

[ggascoigne]

patch-package has been a godsend, when you need it, you really need it. It's an amazing escape hatch.

[hokkos]

all bundler are moving to rust core and typescript front/CLI/adapter, there is no going back, the perf difference is too huge. Prisma is the outlier with it's own story here.

[ealmansouri]

How is it free to self host? The screen you showed says "Contact us for pricing"

[hussienaitbella]

Rust is slow when your algorithm is slow

[collinoly]

I’ve had a ticket open for 2 years with prisma. If the typescript file generated is larger than two gigs it will fail.

[boredstudent]

Rust could still be useful if performance is an issue but then stuff like FFI or Node Native calls not an entirely separate GraphQL server. And to be fair if performance is that important, don't use JS in the Backend so Prisma will be fine.

[kay-xr]

I don't think this was Rust's fault...

[Duskdown]

Unless they join tables in the DB and not in-memory, Prisma remains a tool for pet projects to me