360: Permit renaming constructors, getters and setters r=nikomatsakis a=MihailMihov
The goal is to add an option `constructor_name` to `#[salsa::input]`, `#[salsa::interned]` and `#[salsa::tracked]` that allows changing the name of the generated constructor. After that add attributes `get` and `set` to the fields which allow overriding the names of the getters and setters. In the end the following snippet should compile and work:
```rust
#[salsa::interned(constructor = from_string)]
struct MyInterned {
#[get(text)] #[set(set_text)] #[return_ref]
field: String,
}
impl MyInterned {
pub fn new(db: &dyn Db, s: impl Display) -> MyInterned {
MyInterned::from_string(db, s.to_string())
}
pub fn field(self, db: &dyn Db) -> &str {
self.text(db)
}
pub fn set_field(self, db: &mut dyn Db, id: String) {
self.set_text(&mut db, id)
}
}
```
resolves#332
Co-authored-by: Mihail Mihov <mmihov.personal@gmail.com>
353: start documenting plumbing r=nikomatsakis a=nikomatsakis
Feedback desired! I am trying to document an overview of the new salsa 2022 plumbing. I'd love for folks to [read these docs and tell me if they make sense](https://deploy-preview-353--salsa-rs.netlify.app/plumbing.html).
Co-authored-by: Niko Matsakis <niko@alum.mit.edu>
352: Add options to tracked funcitons for lru capacity r=nikomatsakis a=XFFXFF
fixes#344
Now we can write something like the following to set the lru capacity of tracked functions
```rust
#[salsa::tracked(lru=32)]
fn my_tracked_fn(db: &dyn crate::Db, ...) { }
```
some details:
* lru should not be combined with specify. We will report an error if people do #[salsa::tracked(lru = 32, specify)]
* set 0 as default capacity to disable LRU (Because I think doing this would make the code simpler when implementing `create_ingredients` of tracked functions).
* old salsa support to change lru capacity at runtime, [as noted here](https://salsa-rs.github.io/salsa/rfcs/RFC0004-LRU.html?highlight=change#reference-guide), but we do not support this now
Co-authored-by: XFFXFF <1247714429@qq.com>
It turns out that we have some outputs (accumulators) for which
it only makes sense to have a `DependencyIndex` (they don't have
individual keys to identify).
We also track whether reset is required at the ingredient level.
For tracked struct fields, we were not using `push_mut`,
and I think that was an oversight.
The plan is to do a "dependent ingredient" interlinking pass
once the database is constructed.
I realized I can do this better:
* require that outputs are DatabaseKeyIndex, fewer unwraps,
more clearly justified
* when we validate result of tracked fn, also validate its outputs
(this is incompletely implemented, would ideally be separated
into its own commit, but I'm short for time)
The last step will allow us to keep the memoized results for
assigned values and means we don't have to eagerly clear them.
If we see an "assigned value" that is not verified in the current
revision, it can simply be considered dirty.
We can still delete them when entities are deleted, but they're
less special.
Rename QueryInputs to QueryEdges.
Modify its fields to track both inputs and outputs.
The size of the struct doesn't actually change,
as the separator comes out of padding.
Rename QueryInputs to QueryEdges and modify its fields
to track both inputs and outputs. The size of the struct
doesn't actually change, the separator comes out of padding.
