I can see that. Pydantic is great but relatively slow (which matters on edge devices) and can be bloated.

The fact that all your projects use Pydantic makes it an easy starting point and created standardisation - of course.

Nevertheless, I can definitely see some use-cases for lightweight JSON-serialisation without bringing in Pydantic. Dataclasses are great, but lack proper json handling.

Cool!

I wonder if this isn't toooo much magic. How does this scale beyond simple use-cases? I fear one might hit a wall at 80%, where the first part is super easy indeed but not customizable enough to get to 100% of all use-cases.

Thanks! It's a great point to make. Early on I worked on a "migration" ability for the resources, so that every change you make automatically updates the underlying database. But the longer I worked on that, the more I realized I could create/delete as many APIs as I wanted working out what's necessary. Then scale it up for longevity. And the time it took to create/delete APIs was nearly as fast as some of the build times I've experienced for front-end applications.

That's ultimately what I think the value is: API development as hyper-iterative as front-end apps.

I'm still clearly working out how to describe this value. I'm developing more examples to help define what I think the best use-cases are.

RIP Popcorn Time. It was fucking amazing.

Is it actually gone? I found official looking sites on a quick Google.

could we not bring back popcorn time if it goes over to decentralized github : https://radicle.xyz/

its strange how we can't just build software that then works for ever. The emulation boys got something going but netowkred software unlikley

Love this! Finally something happening in this space. dbeaver was the best, but it's so so ugly and old. I love Datagrip, but no free version.

Please have good formatting of SQL queries - like Datagrip. Very little open source SQL formatters out there, so might require something custom.

And very much waiting for Snowflake support (incl. OAuth)

What tech are you running under the hood for this? Is this similar to entity detection using BERT?

just regex filters

I like this!

Overall love the DX with Claude Code in my vscode (kinda hate all the CLI only approaches the rest have), but would be great to still utilize other models.

Impressive speed performance! And love the magic, like wiring services and repos.

Not super sure about the magical SQL queries OOTB, but that might be a preference thing.

Lots of docs too, great job!

(This also triggered a question in me: should Astral also support/write a web framework?)

Thanks for the kind words!

The speed mostly comes from building it on top of Starlette and Granian, while keeping overhead low, so I can't claim much credit on that part. In the end, business logic will be the bottleneck anyway :)

Yeah, I've seen mixed responses on SQL magic. Spring and Ruby devs I talked to seemed to like it (with Ruby active records having a similar feature), but JS, Python and other devs I talked to found it odd.

I guess it depends on the ecosystem people get used to?

Would be exciting to see Astral come out with a server! Though, with the current landscape, it feels like there isn't too much to be done without massive efforts, so I don't imagine they could justify spending the time given how well they're doing in their niche. Could be wrong, though.

Yeah, I think the Spring folk typically like running ORMs and don't want to get their hands dirty on SQL.

Whereas the Python folk are often used to running SQL manually, so like less of the magic?

How do you see this compare to things like Amazon Bedrock, where it runs OSS in my own infra?

Bedrock has strong contractual guarantees, but it's still only a legal contract and runs on AWS infra. This is certainly okay for many use cases, we're trying to build for users who want verifiable privacy guarantees beyond legal contracts.

We're also doing more than pure inference, and trying to work with other companies who want to provide their users additional verifiability and confidentiality guarantees by running their entire private data processing pipeline on our platform.

Exactly! The huge benefit is that you're just writing before/after Java code, so not regex or other complex AST code.

For an avid Excel/GSheets user and programmer, this is amazing! When is a GSheets version coming ;-)

For gsheets, depends on one's language tolerance. "using python in google sheets" & just use system call for above examples. ;)

For excel:

Lean proof of concept https://matt-rickard.com/spreadsheets-and-lisp

Extended proof of concept for thinking outside the [cell box] : https://www.gnu.org/software/emacs/manual/html_node/emacs/En...