Incremental Merging

Author: dpl0a

src/eval/cache/incremental.rs

@@ -1,4 +1,5 @@
-/// A [Cache] implementation with incremental computation features.
+//! A [Cache] implementation with incremental computation features.
+use std::collections::{HashMap, HashSet};
 
 use super::{BlackholedError, Cache, CacheIndex, Closure, Environment, IdentKind};
 use crate::{
@@ -15,6 +16,12 @@ pub enum IncNodeState {
     Evaluated,
 }
 
+#[derive(Debug, Copy, Clone, Eq, PartialEq)]
+pub struct DependencyLink {
+    id: Ident,
+    idx: CacheIndex,
+}
+
 /// A node in the dependent computation graph stored in [IncCache].
 #[derive(Debug, Clone)]
 pub struct IncNode {
@@ -26,8 +33,10 @@ pub struct IncNode {
     bty: BindingType,
     // The state of the node.
     state: IncNodeState,
+    // Forward links to dependencies.
+    fwdlinks: Vec<DependencyLink>,
     // Backlinks to nodes depending on this node.
-    backlinks: Vec<CacheIndex>,
+    backlinks: Vec<DependencyLink>,
 }
 
 impl IncNode {
@@ -38,6 +47,7 @@ impl IncNode {
             kind,
             bty,
             state: IncNodeState::default(),
+            fwdlinks: Vec::new(),
             backlinks: Vec::new(),
         }
     }
@@ -60,7 +70,7 @@ impl IncCache {
         idx
     }
 
-    fn revnode_as_explicit_fun<'a, I>(node: &IncNode, args: I) -> IncNode
+    fn revnode_as_explicit_fun<'a, I>(node: &mut IncNode, args: I)
     where
         I: DoubleEndedIterator<Item = &'a Ident>,
     {
@@ -75,18 +85,141 @@ impl IncCache {
                 let as_function =
                     args.rfold(body, |built, id| RichTerm::from(Term::Fun(*id, built)));
 
-                IncNode::new(
-                    Closure {
-                        body: as_function,
-                        env,
-                    },
-                    node.kind,
-                    node.bty.clone(),
-                )
+                node.orig = Closure {
+                    body: as_function,
+                    env,
+                }
             }
-            _ => node.clone(),
+            _ => (),
+        }
+    }
+
+    fn update_backlinks(&mut self, idx: CacheIndex) {
+        let node = self.store.get(idx).unwrap().clone();
+        for i in node.fwdlinks {
+            let n = self.store.get_mut(i.idx).unwrap();
+            n.backlinks.push(DependencyLink { id: i.id, idx: idx });
+        }
+    }
+
+    fn propagate_dirty(&mut self, idx: CacheIndex) {
+        let mut node = self.store.get_mut(idx).unwrap();
+        node.cached = None;
+        node.state = IncNodeState::Suspended;
+
+        let mut visited = HashSet::new();
+        let mut stack = node.backlinks.clone();
+
+        visited.insert(idx);
+
+        while !stack.is_empty() {
+            let i = stack.pop().unwrap();
+            visited.insert(i.idx);
+            let mut current_node = self.store.get_mut(i.idx).unwrap();
+            current_node.cached = None;
+            current_node.state = IncNodeState::Suspended;
+            stack.extend(
+                current_node
+                    .backlinks
+                    .iter()
+                    .filter(|x| !visited.contains(&x.idx)),
+            )
+        }
+    }
+
+    fn propagate_dirty_vec(&mut self, indices: Vec<CacheIndex>) {
+        let mut visited = HashSet::new();
+        let mut stack = indices;
+
+        while !stack.is_empty() {
+            let i = stack.pop().unwrap();
+            visited.insert(i);
+            let mut current_node = self.store.get_mut(i).unwrap();
+            current_node.cached = None;
+            current_node.state = IncNodeState::Suspended;
+            println!("IDX: {:?} BLs: {:?}", i, current_node.backlinks);
+            stack.extend(
+                current_node
+                    .backlinks
+                    .iter()
+                    .map(|x| x.idx)
+                    .filter(|x| !visited.contains(&x)),
+            )
         }
     }
+
+    /* Do we need this when we can revert in place?
+
+    fn propagate_revert(&mut self, id: Ident, idx: CacheIndex) -> HashMap<Ident, CacheIndex> {
+        let mut nodes_reverted = HashMap::new();
+
+        let mut visited = HashSet::new();
+        let mut stack = vec![idx];
+
+        while !stack.is_empty() {
+            let i = stack.pop().unwrap();
+            visited.insert(i);
+
+            let idx_reverted = self.revert(&idx);
+            //FIXME: use the actual node's id
+            let node_id = Ident::from("TODO!");
+            nodes_reverted.insert(node_id, idx_reverted);
+
+            let current_node = self.store.get(i).unwrap();
+
+            stack.extend(
+                current_node
+                    .backlinks
+                    .iter()
+                    .map(|x| x.idx)
+                    .filter(|x| !visited.contains(x)),
+            )
+        }
+
+        nodes_reverted
+    } */
+
+    fn smart_clone(&mut self, v: Vec<CacheIndex>) -> HashMap<CacheIndex, CacheIndex> {
+        let mut new_indices = HashMap::new();
+
+        for i in v.iter() {
+            let current_node = self.store.get(*i).unwrap().clone();
+            new_indices.insert(*i, self.add_node(current_node));
+        }
+
+        for i in new_indices.values() {
+            let current_node = self.store.get_mut(*i).unwrap();
+
+            for dep in current_node.backlinks.iter_mut() {
+                dep.idx = if let Some(idx) = new_indices.get(&dep.idx) {
+                    *idx
+                } else {
+                    dep.idx
+                }
+            }
+
+            let mut to_be_updated = vec![];
+
+            for dep in current_node.fwdlinks.iter_mut() {
+                dep.idx = if let Some(idx) = new_indices.get(&dep.idx) {
+                    *idx
+                } else {
+                    to_be_updated.push(dep.clone());
+                    dep.idx
+                }
+            }
+
+            for dep in to_be_updated {
+                let target_node = self.store.get_mut(dep.idx).unwrap();
+                target_node.backlinks.push(DependencyLink {
+                    id: dep.id,
+                    idx: *i,
+                });
+            }
+        }
+
+        new_indices
+    }
 }
 
 impl Cache for IncCache {
@@ -204,9 +337,12 @@ impl Cache for IncCache {
             kind: node.kind,
             bty: node.bty.clone(),
             state: node.state,
+            fwdlinks: node.fwdlinks.clone(),
             backlinks: node.backlinks.clone(),
         };
 
+        // TODO: Should this push the dependencies?
+
         self.add_node(new_node)
     }
 
@@ -224,13 +360,19 @@ impl Cache for IncCache {
             BindingType::Revertible(ref deps) => match deps {
                 FieldDeps::Unknown => new_cached.env.extend(rec_env.iter().cloned()),
                 FieldDeps::Known(deps) if deps.is_empty() => (),
-                FieldDeps::Known(deps) => new_cached
-                    .env
-                    .extend(rec_env.iter().filter(|(id, _)| deps.contains(id)).cloned()),
+                FieldDeps::Known(deps) => {
+                    let deps = rec_env.iter().filter(|(id, _)| deps.contains(id)).cloned();
+                    node.fwdlinks = deps
+                        .clone()
+                        .map(|(id, idx)| DependencyLink { id, idx })
+                        .collect();
+                    new_cached.env.extend(deps);
+                }
             },
         }
 
         node.cached = Some(new_cached);
+        self.update_backlinks(*idx);
     }
 
     fn saturate<'a, I: DoubleEndedIterator<Item = &'a Ident> + Clone>(
@@ -239,7 +381,7 @@ impl Cache for IncCache {
         env: &mut Environment,
         fields: I,
     ) -> RichTerm {
-        let node = self.store.get(idx).unwrap();
+        let node = self.store.get_mut(idx).unwrap();
 
         let mut deps_filter: Box<dyn FnMut(&&Ident) -> bool> = match node.bty.clone() {
             BindingType::Revertible(FieldDeps::Known(deps)) => {
@@ -249,13 +391,10 @@ impl Cache for IncCache {
             BindingType::Normal => Box::new(|_: &&Ident| false),
         };
 
-        let node_as_function = self.add_node(IncCache::revnode_as_explicit_fun(
-            node,
-            fields.clone().filter(&mut deps_filter),
-        ));
+        IncCache::revnode_as_explicit_fun(node, fields.clone().filter(&mut deps_filter));
 
         let fresh_var = Ident::fresh();
-        env.insert(fresh_var, node_as_function);
+        env.insert(fresh_var, idx);
 
         let as_function_closurized = RichTerm::from(Term::Var(fresh_var));
         let args = fields.filter_map(|id| deps_filter(&id).then(|| RichTerm::from(Term::Var(*id))));
@@ -264,4 +403,12 @@ impl Cache for IncCache {
             RichTerm::from(Term::App(partial_app, arg))
         })
     }
+
+    fn smart_clone(&mut self, v: Vec<CacheIndex>) -> HashMap<CacheIndex, CacheIndex> {
+        self.smart_clone(v)
+    }
+
+    fn propagate_dirty(&mut self, indices: Vec<CacheIndex>) {
+        self.propagate_dirty_vec(indices);
+    }
 }

src/eval/cache/lazy.rs

@@ -4,8 +4,11 @@ use crate::{
     identifier::Ident,
     term::{record::FieldDeps, BindingType, RichTerm, Term},
 };
-use std::cell::{Ref, RefCell, RefMut};
 use std::rc::{Rc, Weak};
+use std::{
+    cell::{Ref, RefCell, RefMut},
+    collections::HashMap,
+};
 
 /// The state of a thunk.
 ///
@@ -355,7 +358,7 @@ impl ThunkData {
 /// inside a record may be invalidated by merging, and thus need to store the unaltered original
 /// expression. Those aspects are handled and discussed in more detail in
 /// [InnerThunkData].
-#[derive(Clone, Debug, PartialEq)]
+#[derive(Clone, Debug)]
 pub struct Thunk {
     data: Rc<RefCell<ThunkData>>,
     ident_kind: IdentKind,
@@ -560,6 +563,21 @@ impl Thunk {
         self.data.borrow().deps()
     }
 }
+
+impl PartialEq for Thunk {
+    fn eq(&self, other: &Self) -> bool {
+        self.data.as_ptr() == other.data.as_ptr() && self.ident_kind == other.ident_kind
+    }
+}
+
+impl Eq for Thunk {}
+
+impl std::hash::Hash for Thunk {
+    fn hash<H: std::hash::Hasher>(&self, state: &mut H) {
+        let raw_ptr = self.data.as_ptr();
+        (self.ident_kind, raw_ptr).hash(state)
+    }
+}
 /// A thunk update frame.
 ///
 /// A thunk update frame is put on the stack whenever a variable is entered, such that once this
@@ -692,4 +710,15 @@ impl Cache for CBNCache {
     ) -> Result<Self::UpdateIndex, BlackholedError> {
         idx.mk_update_frame()
     }
+
+    fn smart_clone(
+        &mut self,
+        v: Vec<CacheIndex>,
+    ) -> std::collections::HashMap<CacheIndex, CacheIndex> {
+        v.into_iter()
+            .map(|idx| (idx.clone(), self.revert(&idx)))
+            .collect()
+    }
+
+    fn propagate_dirty(&mut self, indices: Vec<CacheIndex>) {}
 }

src/eval/cache/mod.rs

@@ -1,3 +1,5 @@
+use std::collections::HashMap;
+
 /// The Nickel generic evaluation cache. This module abstracts away the details for managing
 /// suspended computations and their memoization strategies.
 ///
@@ -11,15 +13,15 @@ use crate::{
 };
 
 pub mod lazy;
-//pub mod incremental;
+// pub mod incremental;
 
 /// An index to a specific item stored in the cache
 pub type CacheIndex = lazy::Thunk;
-//pub type CacheIndex = usize;
+// pub type CacheIndex = usize;
 
 /// The current Cache implementation
 pub type CacheImpl = lazy::CBNCache;
-//pub type CacheImpl = incremental::IncCache;
+// pub type CacheImpl = incremental::IncCache;
 
 /// A black-holed node was accessed, which would lead to infinite recursion.
 #[derive(Clone, Copy, Debug, Eq, PartialEq)]
@@ -94,4 +96,8 @@ pub trait Cache: Clone {
         &mut self,
         idx: &mut CacheIndex,
     ) -> Result<Self::UpdateIndex, BlackholedError>;
+
+    fn smart_clone(&mut self, v: Vec<CacheIndex>) -> HashMap<CacheIndex, CacheIndex>;
+
+    fn propagate_dirty(&mut self, indices: Vec<CacheIndex>);
 }