Repository navigation
Joined table variable names should include source table name for disambiguation #560
Description
Hi, great library!
When using an SQL JOIN statement, we can end up with tables that have the same column name in the same variable interpolation. Currently, this results in names such as Name and Name_2. For example, given the following schema:
CREATE EXTENSION IF NOT EXISTS pgcrypto;
CREATE TABLE users (
id UUID gen_random_uuid (),
name TEXT
)
CREATE TABLE posts (
id UUID gen_random_uuid (),
user_id UUID REFERENCES users (id),
name TEXT
)And the following query:
-- name: GetPostsByUserByName :one
SELECT * FROM posts
LEFT JOIN users ON posts.user_id = users.id
WHERE posts.name = $1 AND users.name = $2We end up with the following parameter type:
type GetPostsByUserByNameParams struct {
Name uuid.UUID
Name_2 uuid.UUID
}This is obviously not ideal, and I think we should do what SQL does for disambiguation; use the table names. It could look something like this (using the existing algorithm for de-pluralising table names):
type GetPostsByUserByNameParams struct {
PostName uuid.UUID
UserName uuid.UUID
}Alternatively, if that is too heavy handed, I tried using the rename functionality to override the variable name, but it doesn't seem to work as expected; the following accomplished nothing:
rename:
posts.name: "PostName"
users.name: "UserName"Any thoughts appreciated!
Activity
kyleconroy commentedon Jun 22, 2020
The renaming functionality is poorly documented. It actually doesn't work for parameters.
But I have a solution for you: named parameters. You can used either
sqlc.arg()or@.I do agree that we should be able to generate these better names automatically, but it's not something I'm focusing on right now.
johanbrandhorst commentedon Jun 22, 2020
Fantastic, thanks for the workaround!