-
-
Notifications
You must be signed in to change notification settings - Fork 643
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[FEATURE]: Allow to pass empty values to inArray #1295
[FEATURE]: Allow to pass empty values to inArray #1295
Comments
I believe such a change would lead to an easier developer experience. That being said, similar changes to other aspects of the API have been closed by the team with no action. See #1078 |
I would expect |
I'm not part of the team. I'm just trying to help cleanup the issues. There are two options that have been discussed:
function isTuple<T extends any[]>(array: T): T is [T, ...T[]] {
return array.length > 0;
} And handle the case accordingly in userland.
The down side of the first is that the user has to deal with it (like they are right now, just now the types help them figure it out they need to do it), and handle the case when the array is empty, like this for example: ...
.where(isTuple(myArray) ? inArray(column, myArray) : sql`false`)
... The down side of the second one is that it introduces hidden behavior, we just don't know down the line what other issues it will start. |
I think it's reasonable for a user to assume the following functionality:
In JS: |
Adding a +1 for this; throwing an error when passed an empty array is not useful nor expected, nor does it raise a typescript error which especially sucks. |
I've been using inArray extensively in my recent project and handling this behavior is my single biggest pain point with drizzle. Throwing an error because the array is empty is not expected at all, and I've spent a fair amount of time fixing production code just because of this "feature". I honestly don't see why (as a benefit to the developer) the empty array is not handled internally by drizzle and that it is expected that this should be handled externally outside drizzle instead. |
This could be a solution: https://moderndash.io/docs/arrayminlength |
Hello guys ! Do not hesitate to add your thumb-ups ;) |
Recently migrated an app from Prisma to Drizzle, and this has been a big problem. Major ORMs handle this case gracefully, like with Prisma: await prisma.user.findMany({ where: { id: { in: [] }}})
// => just return empty result With Rails: User.where(id: [])
# => converted to WHERE 1=0 and return empty result I'll try to explain how bad this can get. This innocuous query: await db.query.users.findMany({
where: (users, { inArray, or }) =>
or(
inArray(users.a, arrayA),
inArray(users.b, arrayB),
),
}) has to be rewritten with a guard logic like this: let where
if (arrayA.length === 0 && arrayB.length === 0) {
return []
}
if (arrayA.length > 0 && arrayB.length === 0) {
where = inArray(users.a, arrayA)
}
if (arrayA.length === 0 && arrayB.length > 0) {
where = inArray(users.b, arrayB)
}
if (arrayA.length > 0 && arrayB.length > 0) {
where = or(inArray(users.a, arrayA), inArray(users.b, arrayB))
}
await db.query.users.findMany({ where }) combinatorial explosion awaits. :) |
I agree this should be fixed, since it's a big barrier to adoption of drizzle for those moving from other ORMs. It's one of those edge cases where matching SQL is not worth the tradeoff. For those who need a temporary solution, I would recommend re-exporting
import {
inArray as drizzleInArray,
notInArray as drizzleNotInArray,
sql,
} from "drizzle-orm"
/**
* Test whether the first parameter, a column or expression,
* has a value from a list passed as the second argument.
*
* ## Examples
*
* ```ts
* // Select cars made by Ford or GM.
* db.select().from(cars)
* .where(inArray(cars.make, ['Ford', 'GM']))
* ```
*
* @see notInArray for the inverse of this test
*/
const inArray: typeof drizzleInArray = (
...args: Parameters<typeof inArray>
): ReturnType<typeof inArray> => {
const [column, values] = args;
// https://github.com/drizzle-team/drizzle-orm/issues/1295
if (Array.isArray(values) && values.length === 0) {
return sql`false`;
}
return drizzleInArray(...args);
};
/**
* Test whether the first parameter, a column or expression,
* has a value that is not present in a list passed as the
* second argument.
*
* ## Examples
*
* ```ts
* // Select cars made by any company except Ford or GM.
* db.select().from(cars)
* .where(notInArray(cars.make, ['Ford', 'GM']))
* ```
*
* @see inArray for the inverse of this test
*/
const notInArray: typeof drizzleNotInArray = (
...args: Parameters<typeof drizzleNotInArray>
): ReturnType<typeof drizzleNotInArray> => {
const [column, values] = args;
// https://github.com/drizzle-team/drizzle-orm/issues/1295
if (Array.isArray(values) && values.length === 0) {
return sql`true`;
}
return drizzleNotInArray(...args);
};
export { inArray, notInArray }; Then in your file where you export export * from "./re-exports"; |
Oh this monkey patch is a lifesaver! Thanks a ton @Hansenq ! |
@AndriiSherman reviewed the MR. |
@AndriiSherman @RemiPeruto I am looking to update some code to account for this. one example looks like this: if (idsToDelete.length) {
await db.delete(someTable).where(inArray(someTable.id, idsToDelete));
} With this PR, does it make sense to keep the code as is to avoid a useless query? or will drizzle abort a DB call and just resolve the promise if the sql evaluates to nothing. it would be nice to be able to remove the sql`false` |
@jakeleventhal looking at the MR, it does nothing to cancel the query. It just changes a part of the generated SQL to false. To be able to not make a DB call, drizzle would have to be able to prove that the query will do nothing. From what I understood, this is not a design goal of drizzle. In your case the SQL would look simply like DELETE FROM sometable WHERE false which will always fail, but the query might be more complicated like: await db.delete(someTable).where(or(inArray(someTable.id, idsToDelete), true)); which would look DELETE FROM sometable WHERE false OR true and suddenly the query would have to be executed |
Describe what you want
Passing empty array to
inArray
throws:I think it would be more ergonomic to allow empty array as it would avoid additional checks outside of query builder.
Also, throwing an error causes unexpected runtime error if the array changes dynamically.
That's actually what have happened to me with following code:
The text was updated successfully, but these errors were encountered: