ltree

[cemoktra]

We just played around with ltrees. sea-orm-cli currently generates an empty enum. Maybe for the special ltree case it shoud generate something like

#[derive(Debug, Clone, PartialEq, EnumIter, DeriveActiveEnum)]
#[sea_orm(rs_type = "String", db_type = "Enum", enum_name = "ltree")]
pub enum Ltree {
    Value(String)
}

Currently this ends up in error: Missing macro attribute, either string_valueornum_value should be specified, so there would need to be a way to extract the string out of the Value variant.

[billy1624]

Hey @cemoktra, I guess you need to implement ActiveEnum manually instead of deriving DeriveActiveEnum.
https://www.sea-ql.org/SeaORM/docs/generate-entity/enumeration#manual-implementation

[cemoktra]

Ok also possible, and what about mapping ltree to TEXT?

[cemoktra]

But I think I'll provide some changes for ltree

[billy1624]

I'm not sure how ltree works.

But for ActiveEnum,

• on insert, it will generate SQL 'enum_value' AS enum_type
• on select, it will generate SQL enum_column AS text

[cemoktra]

I started with the basic work on sqlx: launchbadge/sqlx#1696

[billy1624]

Thanks!! @cemoktra
Let me know if you need any help on it!

[cemoktra]

Yeah maybe. Currently sea-query does not depend on sqlx itself, but in order to use my new type it will

[cemoktra]

Do you have any opinion on how to integrate this into seaql?

[billy1624]

I think we can treat ltree as a simple string? Then, we don't need any additional SQLx dependency.
launchbadge/sqlx#1696 (comment)

[cemoktra]

Currently i get thread 'test::test' panicked at 'not implemented', sea-orm/src/driver/sqlx_postgres.rs:8:1 when trying to filter for an LTree with .eq, also .contains wants a String instead of PgLTree

.filter(test::Column::Ltree.eq(ltree))

Just doing a query works

[cemoktra]

Same for insert

[cemoktra]

Ok found the missing piece, sorry

[tyt2y3]

This is sad but I think our general response to "support a particular Postgres feature" will be "we will gladly accept PR" for now