zod validation (fixes #364) #367
Conversation
|
|
||
| const anyTypedQuery = await client.any(sql<Row>``); | ||
| expectTypeOf(anyTypedQuery).toEqualTypeOf<readonly Row[]>(); | ||
| const anyTyped = await client.any(sql.type(Row)``); |
There was a problem hiding this comment.
I don't think the old test should be deleted? Is slonik going to stop supporting sql<Row>`select 123`?? I would have thought this is just a new feature and should have corresponding new type tests. Right now this PR deletes many many tests that ensure slonik's types are correct for end users (since test/dts.ts also got deleted entirely).
There was a problem hiding this comment.
I don't see a path for supporting both. They don't mix will.
What happens if someone does:
sql<{id: number}>.type(
z.object({
id: z.string()
})
)`select 123`There was a problem hiding this comment.
Not a risk - that isn't valid ts syntax. The generic on sql is only when used as a function directly. sql.type is another, independent, generic function which returns a non-generic function. I'm very sure of this, and still pretty concerned about abandoning the existing (and totally satisfactory!) form slonik has now. If you're not convinced, can I have a few days to come up with a pull request which proves this out?
There was a problem hiding this comment.
If you're not convinced, can I have a few days to come up with a pull request which proves this out?
Absolutely!
gajus
force-pushed
the
gajus/add-zod-validation
branch
from
26003d3
to
20c5683
Compare
gajus
force-pushed
the
gajus/add-zod-validation
branch
from
20c5683
to
1d2b679
Compare
gajus
force-pushed
the
gajus/add-zod-validation
branch
from
1d2b679
to
aaaca65
Compare
|
Closing in favor of #369. |
No description provided.