Finish the implementation of @default(nanoid())

[tomhoule]

See prisma/prisma-engines#3556 for context.

The missing pieces we identified:

• Langage-tools completion feat(ls): completion for nanoid() language-tools#1372
• Migrate support (engines-side) for diffing & special handling of prisma-level defaults re: warnings.
• A TS-side tests in prisma/prisma repo
  • add it to

    prisma/packages/migrate/src/__tests__/fixtures/nativeTypes-postgresql/prisma/schema.prisma

    Line 97 in c70940d

    model with_all_field_defaults {

  • add it to

    prisma/packages/migrate/src/__tests__/fixtures/nativeTypes-cockroachdb/prisma/schema.prisma

    Line 91 in c70940d

    model with_all_field_defaults {

  • And a Client test, see https://github.com/prisma/prisma/blob/main/TESTING.md#functional-tests-for-the-client

Once these tasks are done, we need to:

• update / merge docs from docs: nanoid() schema function docs#4367
• merge language-tools feat(ls): completion for nanoid() language-tools#1372
• mention nanoid() in our next release notes

Note: an improvement has been identified but is not planned for now -> #17294

[Jolg42]

Note: there is a docs issue at prisma/docs#4363
And PR prisma/docs#4367

[Jolg42]

For Language-tools 2 files need to be modified, basically copy what we do for cuid and rename to nanoid
https://github.com/search?q=repo%3Aprisma%2Flanguage-tools%20cuid&type=code

[rintaun]

@tomhoule For clarification, when you say

special handling of prisma-level defaults re: warnings

are you talking about when you run prisma db pull it emits warnings saying These models were enriched with ... information taken from the previous Prisma schema-type messages?

If so, I'm not sure such warnings are actually emitted for cuid() and uuid() defaults right now, though I may be wrong. That's not to say I don't think it's a good idea, though.

[Jolg42]

Note: @tomhoule thinks he can point you to the functionality in the sql-destructive-change-checker you need to make work for this.
But he is quite busy today and would need to dive into the implementation again for the details and write a bunch of tests for diffing schemas with these defaults. They will be broken, and they will have to be fixed.

[rintaun]

I think I see where you're talking about (warnings.rs and scalar_field.rs seem relevant, among others in that call chain), but I don't see any warnings related to cuid() or uuid() except regarding Prisma 1.

[tomhoule]

For the warnings: it's about migration warnings in rather than introspection warnings, I'd follow the code around https://github.com/rintaun/prisma-engines/blob/c71fd52daf0c7a10c154d02e2a6ef791e4a792c3/migration-engine/connectors/sql-migration-connector/src/database_schema.rs#L8

For the implementation: adding a simple test with a schema with a nanoid default in https://github.com/rintaun/prisma-engines/tree/c71fd52daf0c7a10c154d02e2a6ef791e4a792c3/migration-engine/migration-engine-tests/tests/single_migration_tests should make the problem with migration generation visible.

[Jolg42]

@rintaun I updated the TODO list at the top.

Did you have time to check for Migrate support and what @tomhoule mentioned? Do you need help?

Once the TODO tasks are complete in the description, we will officially add it as supported by Prisma in the next release.