Struct RuntimeData 

pub(crate) struct RuntimeData<'db> {
    pub db: &'db dyn Db,
    inference_vars: RwLock<Vec<InferenceVarData<'db>>>,
    sub_inference_var_pairs: Mutex<Set<(InferVarIndex, InferVarIndex)>>,
    ready_to_execute: Mutex<Vec<Arc<CheckTask>>>,
    waiting_on_inference_var: Mutex<Map<InferVarIndex, Vec<EqWaker>>>,
    complete: AtomicBool,
    next_task_id: AtomicU64,
    root_log: LogHandle<'db>,
}

Fields

db: &'db dyn Db

inference_vars: RwLock<Vec<InferenceVarData<'db>>>

Stores the data for each inference variable created thus far.

sub_inference_var_pairs: Mutex<Set<(InferVarIndex, InferVarIndex)>>

Pairs (a, b) of inference variables where a <: b is required. We insert into this set when we are relating two inference variables. If it is a new relation, then we know we must propagate bounds.

ready_to_execute: Mutex<Vec<Arc<CheckTask>>>

List of tasks that are ready to execute.

waiting_on_inference_var: Mutex<Map<InferVarIndex, Vec<EqWaker>>>

List of tasks that are blocked, keyed by the variable they are blocked on. When the data for InferVarIndex changes, the tasks will be awoken.

complete: AtomicBool

If true, inference state is frozen and will not change further.

next_task_id: AtomicU64

Integer indicating the next task id; each task gets a unique id.

root_log: LogHandle<'db>

Root log handle for this check. This handle is not used to record events, only to export the overall log. During the check, environments carry a log handle that is specific to the current task. This way when we log an event it is tied to the task that caused it.