MRE for 1778

[junbl]

PR Info

• MRE for sea_orm_cli doesn't generate the correct relation for a foreign key in a different schema #1778

[tyt2y3]

Thanks for submitting. But can you update to the latest sea-orm-cli version and check the result? Preferably 0.12.rc-5

[junbl]

My bad, yep—I originally did it with 0.11 and regenerated for this example. Will try it with 0.12.

[junbl]

Reran with the latest version on master, which was 0.12.rc-4. Couldn't find a branch or anything for rc-5.

I also moved the files around into a separate module per schema (had mistakenly overwritten one mod/prelude with the other so the error looked the same), since that's the way my actual entities are organized, but I'm realizing I don't think that's standardized by sea-orm in any way I know of. Not sure if I got the idea of separating the schemata like that from an example of y'all's or if I just made that up.

If that is the intended organization (entity::schema::table::model), then I was thinking it should just have gone one level up (like super::super::src::src::id) but if it's not then that won't work in general. I guess the only like real danger with the current behavior is if you had a table with the same name in each schema, then the foreign key wouldn't give the compile error since it would resolve, just to the wrong table.

[tyt2y3]

https://crates.io/crates/sea-orm-cli/0.12.0-rc.5 rc.5 was just released a few days ago

[junbl]

Done!

[tyt2y3]

Thanks for helping out. A brief investigation says that we currently don't have a concept of "multiple schemas" in the codegen. And actually, inside SeaSchema, we are not capturing the key_column_usage.table_schema either. So we need to take care of this first.

However, a bigger question might be how actually we'd organize a multi-schema entity workspace.