SeaQL/sea-orm

Description I believe that EntityLoader::load()::with() does not work when trying to additionally select self referencing relations. Below is a hopefully working simple setup which I adapted from my existing project here. It should showcase the issue, but I haven't fully implemented it within a test project. Steps to Reproduce human.rs use sea_orm::entity::prelude::*; #[sea_orm::model] #[derive(Clone, Debug, PartialEq, Eq, DeriveEntityModel)] #[sea_orm(table_name = "humans")] pub struct Model { #[sea_orm(primary_key)] id: i32, #[sea_orm(has_many)] pub hats: HasMany<super::hats::Entity>, #[sea_orm( self_ref, via = "jt_parent_to_child", from = "Humans", to = "Child" )] pub children: HasMany<Entity>, #[sea_orm(self_ref, via = "jt_parent_to_child", reverse)] pub parents: HasMany<Entity>, } impl ActiveModelBehavior for ActiveModel {} jt_parent_child.rs use sea_orm::entity::prelude::*; #[sea_orm::model] #[derive(Clone, Deb

Motivation With a many to many relation like the following, if you want to filter on tag column like note::Entity::load().with(tag::Entity).filter(tag::Column::Name.eq(query)) the query errs with "Query Error: error returned from database: missing FROM-clause entry for table \"tag\" at line 3603". To avoid this, I cannot use EntityLoader but the old EntityTrait::find. Note #[sea_orm::model] #[derive(Clone, Debug, PartialEq, Eq, DeriveEntityModel, Serialize, Deserialize)] #[sea_orm(table_name = "note")] pub struct Model { #[sea_orm(primary_key, auto_increment = false)] pub id: Uuid, #[sea_orm(column_type = "Text")] pub text: String, #[sea_orm(has_many, via = "tags_notes")] pub tags: HasMany<super::tag::Entity>, } Tag #[sea_orm::model] #[derive(Clone, Debug, PartialEq, Eq, DeriveEntityModel, Serialize, Deserialize)] #[sea_orm(table_name = "tag")] pub struct Model { #[sea_orm(primary_key, auto_increment = false)] pub id: Uuid, #[sea_orm(column_type =

Description This is a continuation of #3160, which was closed due to thinking that things were closed by SeaQL/sea-query#1093. This sea-query fixed the issues happening for optional values, but didn't remedy the deserialization problems for non-optional ones. When deserializing an ActiveValue using from_json, if a TimeDateTimeWithTimezone field doesn't have a corresponding value in the JSON payload, somewhere along the way it defaults to a stringified version of the unix epoch, which causes the following error down the line: Json Error: invalid type: string "'1970-01-01 00:00:00.000000 +00:00'", expected an `OffsetDateTime` Steps to Reproduce This is a minimal example of the exact problem I'm having: use sea_orm::entity::prelude::*; use serde::{Deserialize, Serialize}; #[sea_orm::model] #[derive(Clone, Debug, PartialEq, Serialize, Deserialize, DeriveEntityModel)] #[sea_orm(table_name = "example")] #[serde(rename_all = "UPPERCASE")]

Discussed in #3170 Originally posted by kapalla August 13, 2026 A simple model like this #[sea_orm::model] #[derive(Clone, Debug, PartialEq, Eq, DeriveEntityModel)] #[sea_orm(table_name = "cake")] pub struct Model { #[sea_orm(primary_key)] pub id: i32, pub name: String, } causes a clippy error with the -D clippy::derive-partial-eq-without-eq flag or the clippy::nursery group. sea-orm/sea-orm-macros/src/derives/model_ex.rs Lines 100 to 102 in 13be0db if meta.path.is_ident("Eq") { // skip } else if meta.path.is_ident("DeriveEntityModel") {

Motivation The macro for creating models/entities is very useful, but quite limited when it comes to declaring more complex indexes. Some problems include: there is no way to create a partial index there is no way to create a composite index without it being unique (unique_key) the order of columns inside composite indexes matters I don't expect the macro to support more advanced use cases, but there should be a way to specify them manually. But it's problematic with entity-first approach and schema syncing - even if you create the index yourself afterwards, the index would be dropped on the next schema sync, and the index would need to be created every time from scratch. Proposed Solutions don't drop index on schema sync, or add an option for it. add a possibility to specify custom indexes to synchronize