Add support for UUIDv7

[mcuelenaere]

This implements prisma/prisma#24079. The change is pretty minimal, widening the support from @default(uuid()) to @default(uuid(7)).

It leverages the existing uuid crate instead of adding a new one (uuid7).

Right now this code is mostly untested (apart from the tests I added to psl), because I couldn't get the tests running locally. I'd appreciate some guidance in this area. CI runs fine

[CLAassistant]

All committers have signed the CLA.

[kibertoad]

uuid seems to be behind uuid7 with regards to how accurately it represents the spec. See discussion here: uuid-rs/uuid#717

[codspeed-hq]

CodSpeed Performance Report

Merging #4877 will not alter performance

_{Comparing mcuelenaere:feature/uuidv7 (81642e5) with main (34ace0e)}

Summary

✅ 11 untouched benchmarks

[mcuelenaere]

FYI: after this gets an approval, I'll rebase + squash the fixup commits

[kibertoad]

@Weakky Gentle ping. Is anything missing in this PR?

[Weakky]

Apologies for the delayed review and thanks for the quality work @mcuelenaere. Would you mind adding a query-engine test that uses uuid(7) and:

• create a row with it
• reads it from findMany
• reads it from findUnique

You could add this test in this file: query-engine/connector-test-kit-rs/query-engine-tests/tests/writes/ids/uuid_create_graphql.rs

[Weakky]

@mcuelenaere Thanks for the quick review fix. I have run the tests. Would you mind having a look at the failures?

[mcuelenaere]

@Weakky I am able to run the tests for sqlite locally, but they seem to fail in CI for the driver adapters. However, I can't seem to easily run these locally + the output of the CI does not give any indication as to what's going wrong (just that it's failing).

Any help would be appreciated.

[xiaolongkipsi]

follow

[mcuelenaere]

I've rebased on current main, as that seems stable and the tests were failing due to something else the last time

[mcuelenaere]

I've also updated uuid to v1.10.0 to address @kibertoad 's comment

[kibertoad]

@Weakky Is there anything missing?

[Weakky]

@mcuelenaere I have investigated why tests were failing. Turns out SystemTime::now is not implemented yet for Wasm. rust-lang/rust#48564

Fortunately, uuid has a "js" feature which works around this limitation.

https://github.com/uuid-rs/uuid/blob/main/src/timestamp.rs#L302-L327

Could you please add the "js" feature for the uuid crate?

Thanks

[mcuelenaere]

Aha, good catch! I've pushed a fixup commit for that.

I assume just straight-up adding the feature to uuid is fine and it does not need to be in some kind of [target.'cfg(target_arch = "wasm32")'.dependencies] context, right?

[Weakky]

Apologies again for the long delay to get this point. This looks good now. Great work, thanks a lot for the contribution. LGTM 🙏

[Weakky]

@mcuelenaere Yes, that's fine this way.