feat(orm): add foreign key support

[m4tx]

This also contains a general ORM refactor:

• Auto enum was introduced instead of implying if a field is autoincrement based on its name
• save() now actually modifies the saved model with newly assigned autoincrement primary keys (if any)
• A separate SELECT query is performed after INSERT for MySQL to obtain the auto-generated values because MySQL doesn't support the RETURNING clause.

[seqre]

Seems like a good start

[m4tx]

I'm investigating the build failure, it builds successfully in current Rust beta and nightly... Wondering if rust-lang/rust#130654 might be the reason it's fixed.

If I won't get anywhere, I would consider waiting two more days for Rust 1.84 to hit stable and just merge this as-is. We don't really care about this project working on old stable compiler versions anyways... (yet!)

[m4tx]

Ah yes, seems like indeed what I've linked is the fix for the issue. There's a blogpost about this: https://blog.rust-lang.org/inside-rust/2023/12/22/trait-system-refactor-initiative.html

I think these trait impls are problematic:

impl<T: DatabaseField> ToDbFieldValue for Auto<T> {}
impl<T: DatabaseField> ToDbFieldValue for Option<Auto<T>> {}
impl<T: Model + Send + Sync> ToDbFieldValue for ForeignKey<T> {}
impl<T: Model + Send + Sync> ToDbFieldValue for Option<ForeignKey<T>> {}

(or some subset of them)

I don't really want to remove them though, since they make implementing new DatabaseField types much easier (and it's nice to have shiny new features if we're writing a shiny new framework anyway — modern codebase will be one of our advantages over other similar projects). Like I said in my previous comment, I'd just wait for Rust 1.84 to hit stable (9 Jan, so we're almost there) and just run CI on beta until then. @seqre what do you think?

[codecov]

Codecov Report

Attention: Patch coverage is 94.17040% with 117 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
flareon-cli/src/migration_generator.rs	92.93%	32 Missing ⚠️
flareon-macros/src/query.rs	9.52%	19 Missing ⚠️
flareon-codegen/src/model.rs	93.18%	15 Missing ⚠️
flareon/src/db.rs	89.14%	14 Missing ⚠️
flareon-codegen/src/expr.rs	93.93%	10 Missing ⚠️
flareon/src/db/query.rs	76.00%	6 Missing ⚠️
flareon-codegen/src/symbol_resolver.rs	98.87%	4 Missing ⚠️
flareon/src/db/migrations.rs	92.59%	4 Missing ⚠️
flareon/src/db/migrations/sorter.rs	73.33%	4 Missing ⚠️
flareon/src/db/relations.rs	95.45%	4 Missing ⚠️
... and 2 more

Flag	Coverage Δ
rust	86.90% <94.17%> (+2.91%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

Files with missing lines	Coverage Δ
flareon-macros/src/model.rs	95.76% <100.00%> (+1.53%)	⬆️
flareon-macros/tests/compile_tests.rs	100.00% <100.00%> (ø)
flareon/src/auth.rs	94.04% <100.00%> (ø)
flareon/src/db/impl_mysql.rs	100.00% <100.00%> (ø)
flareon/src/db/impl_postgres.rs	100.00% <100.00%> (ø)
flareon/src/db/impl_sqlite.rs	100.00% <100.00%> (ø)
flareon/src/db/sea_query_db.rs	100.00% <100.00%> (ø)
flareon/src/lib.rs	57.57% <ø> (ø)
flareon/src/utils/graph.rs	100.00% <100.00%> (ø)
flareon/tests/db.rs	99.40% <100.00%> (+0.74%)	⬆️
... and 12 more

[m4tx]

If I won't get anywhere, I would consider waiting two more days for Rust 1.84 to hit stable and just merge this as-is. We don't really care about this project working on old stable compiler versions anyways... (yet!)

1.84 has been released today, so we're back on stable.

[seqre]

I don't really want to remove them though, since they make implementing new DatabaseField types much easier (and it's nice to have shiny new features if we're writing a shiny new framework anyway — modern codebase will be one of our advantages over other similar projects). (...) @seqre what do you think?

I agree, since this is a new framework with no existing users, I see no point in supporting older compiler versions. Once it gains popularity, we can think about supporting N-X versions.

[seqre]

LGTM overall, some notes for possible improvements and nitpicks

[seqre]

Great work!