Dealing with large schemas and static type checking

[scumbkt19]

Hello,

We have a reasonably large schema.prisma file, currently at 1859 lines, and we've just crossed the threshold where the prisma-client-py generated types.py file is 51MB. This is causing our pyright static type checking to fail because pyright has a file size limit of 50MB . Specifically, the checking we do in CI to verify whether a commit is safe to merge is now failing with many reportUnknownArgumentType/reportUnknownMemberType/etc errors that were previously valid when the types.py file was small enough to analyze.

I'm surprised I wasn't able to find any existing issues referencing this sort of thing before, does the community have suggestions for how to configure our schema or the generator to generate smaller files?

[RobertCraigie]

Sorry about this, have you enabled recursive types? https://prisma-client-py.readthedocs.io/en/stable/reference/config/#recursive

[scumbkt19]

sorry, I should have included my generator block. We have

generator clientpy {
  provider             = "prisma-client-py"
  interface            = "asyncio"
  recursive_type_depth = 5
  previewFeatures      = ["postgresqlExtensions"]
  output               = "../../../pipelines/pipelines/prisma"
}

but reading the docs you linked in more detail, since we are using pyright, i set the value to recursive_type_depth = -1 and that dropped the generated size to 11MB which seems like it gives us plenty of headroom.

Thanks so much for the quick reply!

matt