Merge pull request #3211 from mthom/setof_correction_and_opt

Correct and optimize setof/3 and bagof/3

Author: mthom

build/instructions_template.rs

@@ -643,11 +643,6 @@ enum SystemClauseType {
     UnattributedVar,
     #[strum_discriminants(strum(props(Arity = "4", Name = "$get_db_refs")))]
     GetDBRefs,
-    #[strum_discriminants(strum(props(
-        Arity = "2",
-        Name = "$keysort_with_constant_var_ordering"
-    )))]
-    KeySortWithConstantVarOrdering,
     #[strum_discriminants(strum(props(Arity = "0", Name = "$inference_limit_exceeded")))]
     InferenceLimitExceeded,
     #[strum_discriminants(strum(props(Arity = "1", Name = "$argv")))]

src/codegen.rs

@@ -639,7 +639,10 @@ impl CodeGenerator {
                 }
             },
             InlinedClauseType::IsRational(..) => match terms[0] {
-                Term::Literal(_, Literal::Rational(_)) => {
+                Term::Literal(
+                    _,
+                    Literal::Rational(_) | Literal::Fixnum(_) | Literal::Integer(_),
+                ) => {
                     instr!("$succeed")
                 }
                 Term::Var(ref vr, ref name) => {
@@ -679,10 +682,13 @@ impl CodeGenerator {
                 }
             },
             InlinedClauseType::IsNumber(..) => match terms[0] {
-                Term::Literal(_, Literal::F64(..))
-                | Term::Literal(_, Literal::Rational(_))
-                | Term::Literal(_, Literal::Integer(_))
-                | Term::Literal(_, Literal::Fixnum(_)) => {
+                Term::Literal(
+                    _,
+                    Literal::F64(..)
+                    | Literal::Rational(_)
+                    | Literal::Integer(_)
+                    | Literal::Fixnum(_),
+                ) => {
                     instr!("$succeed")
                 }
                 Term::Var(ref vr, ref name) => {
@@ -988,20 +994,27 @@ impl CodeGenerator {
                     self.marker.in_tail_position = false;
                     self.marker.reset_contents();
                 }
-                ClauseItem::FirstBranch(num_branches) => {
+                ClauseItem::FirstBranch {
+                    branch_num,
+                    num_branches,
+                } => {
                     branch_code_stack.add_new_branch_stack();
                     branch_code_stack.add_new_branch();
 
-                    self.marker.branch_stack.add_branch_stack(num_branches);
+                    self.marker
+                        .branch_stack
+                        .add_branch_stack(branch_num.clone(), num_branches);
                     self.marker.add_branch();
                 }
-                ClauseItem::NextBranch => {
+                ClauseItem::NextBranch { branch_num } => {
                     branch_code_stack.add_new_branch();
 
                     self.marker.add_branch();
-                    self.marker.branch_stack.incr_current_branch();
+                    self.marker
+                        .branch_stack
+                        .incr_current_branch(branch_num.clone());
                 }
-                ClauseItem::BranchEnd(depth) => {
+                ClauseItem::BranchEnd { depth } => {
                     if !clause_iter.in_tail_position() {
                         let subsumed_hits =
                             branch_code_stack.push_missing_vars(depth, &mut self.marker);

src/debray_allocator.rs

@@ -1,6 +1,6 @@
 use crate::allocator::*;
 use crate::codegen::SubsumedBranchHits;
-use crate::forms::{GenContext, Level};
+use crate::forms::{BranchNumber, GenContext, Level};
 use crate::instructions::*;
 use crate::machine::disjuncts::VarData;
 use crate::parser::ast::*;
@@ -24,32 +24,34 @@ pub struct BranchOccurrences {
     pub shallow_safety: BitSet<usize>, // unset means safe, set means unsafe (after the branch merge)
     pub deep_safety: BitSet<usize>,
     pub num_branches: usize,
-    pub current_branch: usize,
+    pub current_branch_idx: usize,
+    pub current_branch_num: BranchNumber,
     pub subsumed_hits: SubsumedBranchHits,
 }
 
 impl BranchOccurrences {
-    fn new(num_branches: usize) -> Self {
+    fn new(current_branch_num: BranchNumber, num_branches: usize) -> Self {
         Self {
             hits: BranchHits::with_hasher(FxBuildHasher::default()),
             shallow_safety: BitSet::default(),
             deep_safety: BitSet::default(),
             num_branches,
-            current_branch: 0,
+            current_branch_idx: 0,
+            current_branch_num,
             subsumed_hits: SubsumedBranchHits::with_hasher(FxBuildHasher::default()),
         }
     }
 
     pub(crate) fn add_branch_occurrence(&mut self, var_num: usize) {
-        debug_assert!(self.current_branch < self.num_branches);
+        debug_assert!(self.current_branch_idx < self.num_branches);
         let num_branches = self.num_branches;
 
         let entry = self
             .hits
             .entry(var_num)
             .or_insert_with(|| BitVec::repeat(false, num_branches));
 
-        entry.set(self.current_branch, true);
+        entry.set(self.current_branch_idx, true);
         self.subsumed_hits.insert(var_num);
     }
 }
@@ -76,29 +78,16 @@ impl DerefMut for BranchStack {
 }
 
 impl BranchStack {
-    fn branch_subsumes(&self, branch: &BranchDesignator, sub_branch: &BranchDesignator) -> bool {
-        if branch.branch_stack_num < sub_branch.branch_stack_num {
-            if branch.branch_stack_num == 0 {
-                true
-            } else {
-                let idx = branch.branch_stack_num - 1;
-                self[idx].current_branch == branch.branch_num
-            }
-        } else {
-            branch == sub_branch
-        }
-    }
-
     fn safety_unneeded_in_branch(
         &self,
         safety: &VarSafetyStatus,
         branch: &BranchDesignator,
     ) -> bool {
         match safety {
             VarSafetyStatus::Needed => false,
-            VarSafetyStatus::LocallyUnneeded(planter_branch) => {
-                self.branch_subsumes(planter_branch, branch)
-            }
+            VarSafetyStatus::LocallyUnneeded(planter_branch) => planter_branch
+                .branch_num
+                .has_as_subbranch(&branch.branch_num),
             VarSafetyStatus::GloballyUnneeded => true,
         }
     }
@@ -109,27 +98,24 @@ impl BranchStack {
         }
     }
 
-    pub(crate) fn add_branch_stack(&mut self, num_branches: usize) {
-        self.push(BranchOccurrences::new(num_branches));
+    pub(crate) fn add_branch_stack(&mut self, branch_num: BranchNumber, num_branches: usize) {
+        self.push(BranchOccurrences::new(branch_num, num_branches));
     }
 
     pub(crate) fn current_branch_designator(&self) -> BranchDesignator {
-        let branch_stack_num = self.len();
         let branch_num = self
             .last()
-            .map(|occurrences| occurrences.current_branch)
-            .unwrap_or(0);
+            .map(|occurrences| occurrences.current_branch_num.clone())
+            .unwrap_or_else(|| BranchNumber::default());
 
-        BranchDesignator {
-            branch_stack_num,
-            branch_num,
-        }
+        BranchDesignator { branch_num }
     }
 
     #[inline]
-    pub(crate) fn incr_current_branch(&mut self) {
+    pub(crate) fn incr_current_branch(&mut self, branch_num: BranchNumber) {
         let branch_occurrences = self.last_mut().unwrap();
-        branch_occurrences.current_branch += 1;
+        branch_occurrences.current_branch_idx += 1;
+        branch_occurrences.current_branch_num = branch_num;
     }
 
     #[inline]
@@ -235,12 +221,12 @@ impl DebrayAllocator {
                 VarAlloc::Perm(_, allocation) => {
                     let shallow_safety = VarSafetyStatus::needed_if(
                         shallow_safety.contains(var_num),
-                        branch_designator,
+                        &branch_designator,
                     );
 
                     let deep_safety = VarSafetyStatus::needed_if(
                         deep_safety.contains(var_num),
-                        branch_designator,
+                        &branch_designator,
                     );
 
                     if running_count < num_occurrences {
@@ -531,11 +517,11 @@ impl DebrayAllocator {
                     ..
                 },
             ) => {
-                *deep_safety = VarSafetyStatus::unneeded(branch_designator);
-                *shallow_safety = VarSafetyStatus::unneeded(branch_designator);
+                *deep_safety = VarSafetyStatus::unneeded(&branch_designator);
+                *shallow_safety = VarSafetyStatus::unneeded(&branch_designator);
             }
             VarAlloc::Temp { safety, .. } => {
-                *safety = VarSafetyStatus::unneeded(branch_designator);
+                *safety = VarSafetyStatus::unneeded(&branch_designator);
             }
             _ => {
                 unreachable!()
@@ -557,8 +543,8 @@ impl DebrayAllocator {
             ) => {
                 // GetVariable in head chunk is considered safe.
                 if lvl == Level::Deep {
-                    *deep_safety = VarSafetyStatus::unneeded(branch_designator);
-                    *shallow_safety = VarSafetyStatus::unneeded(branch_designator);
+                    *deep_safety = VarSafetyStatus::unneeded(&branch_designator);
+                    *shallow_safety = VarSafetyStatus::unneeded(&branch_designator);
                 } else if term_loc == GenContext::Head {
                     *shallow_safety = VarSafetyStatus::GloballyUnneeded;
                 } else if let Some(&temp_var_num) = self.shallow_temp_mappings.get(&self.arg_c) {
@@ -605,7 +591,7 @@ impl DebrayAllocator {
                 {
                     Target::argument_to_value(r, arg_c)
                 } else {
-                    *shallow_safety = VarSafetyStatus::unneeded(branch_designator);
+                    *shallow_safety = VarSafetyStatus::unneeded(&branch_designator);
                     Target::unsafe_argument_to_value(r, arg_c)
                 }
             }
@@ -640,7 +626,7 @@ impl DebrayAllocator {
                 {
                     Target::subterm_to_value(r)
                 } else {
-                    *deep_safety = VarSafetyStatus::unneeded(branch_designator);
+                    *deep_safety = VarSafetyStatus::unneeded(&branch_designator);
                     Target::unsafe_subterm_to_value(r)
                 }
             }
@@ -651,7 +637,7 @@ impl DebrayAllocator {
                 {
                     Target::subterm_to_value(r)
                 } else {
-                    *safety = VarSafetyStatus::unneeded(branch_designator);
+                    *safety = VarSafetyStatus::unneeded(&branch_designator);
                     Target::unsafe_subterm_to_value(r)
                 }
             }

src/forms.rs

@@ -18,9 +18,11 @@ use indexmap::{IndexMap, IndexSet};
 use ordered_float::OrderedFloat;
 
 use std::cell::Cell;
+use std::cmp::Ordering;
 use std::collections::VecDeque;
 use std::convert::TryFrom;
 use std::fmt;
+use std::hash::{Hash, Hasher};
 use std::ops::{AddAssign, Deref, DerefMut};
 use std::path::PathBuf;
 
@@ -42,12 +44,6 @@ impl AppendOrPrepend {
     }
 }
 
-#[derive(Debug, Clone, Copy)]
-pub enum VarComparison {
-    Indistinct,
-    Distinct,
-}
-
 #[derive(Debug, Clone, Copy, PartialEq, Eq, Hash)]
 pub enum Level {
     Deep,
@@ -128,10 +124,82 @@ impl ChunkType {
     }
 }
 
+#[derive(Debug, Clone)] //, PartialOrd, PartialEq, Eq, Hash)]
+pub(crate) struct BranchNumber {
+    pub(crate) branch_num: Rational,
+    pub(crate) delta: Rational,
+}
+
+impl Default for BranchNumber {
+    fn default() -> Self {
+        Self {
+            branch_num: Rational::from(0),
+            delta: Rational::from(1u64 << 31),
+        }
+    }
+}
+
+impl PartialEq<BranchNumber> for BranchNumber {
+    #[inline]
+    fn eq(&self, rhs: &BranchNumber) -> bool {
+        self.branch_num == rhs.branch_num
+    }
+}
+
+impl Eq for BranchNumber {}
+
+impl Hash for BranchNumber {
+    #[inline(always)]
+    fn hash<H: Hasher>(&self, hasher: &mut H) {
+        self.branch_num.hash(hasher)
+    }
+}
+
+impl PartialOrd<BranchNumber> for BranchNumber {
+    #[inline]
+    fn partial_cmp(&self, rhs: &BranchNumber) -> Option<Ordering> {
+        self.branch_num.partial_cmp(&rhs.branch_num)
+    }
+}
+
+impl BranchNumber {
+    pub(crate) fn has_as_subbranch(&self, other: &Self) -> bool {
+        other.delta <= self.delta
+            && other.branch_num >= self.branch_num
+            && other.branch_num < &self.branch_num + &self.delta
+    }
+
+    pub(crate) fn split(&self) -> BranchNumber {
+        BranchNumber {
+            branch_num: self.branch_num.clone() + &self.delta / Rational::from(2),
+            delta: &self.delta / Rational::from(4),
+        }
+    }
+
+    pub(crate) fn incr_by_delta(&self) -> BranchNumber {
+        BranchNumber {
+            branch_num: self.branch_num.clone() + &self.delta,
+            delta: self.delta.clone(),
+        }
+    }
+
+    pub(crate) fn halve_delta(&self) -> BranchNumber {
+        BranchNumber {
+            branch_num: self.branch_num.clone(),
+            delta: &self.delta / Rational::from(2),
+        }
+    }
+}
+
 #[derive(Debug)]
 pub enum ChunkedTerms {
-    Branch(Vec<VecDeque<ChunkedTerms>>),
-    Chunk { terms: VecDeque<QueryTerm> },
+    Branch {
+        branch_nums: Vec<BranchNumber>,
+        arms: Vec<VecDeque<ChunkedTerms>>,
+    },
+    Chunk {
+        terms: VecDeque<QueryTerm>,
+    },
 }
 
 #[derive(Debug)]
@@ -165,21 +233,22 @@ impl ChunkedTermVec {
     }
 
     pub fn reserve_branch(&mut self, capacity: usize) {
-        self.chunk_vec
-            .push_back(ChunkedTerms::Branch(Vec::with_capacity(capacity)));
+        self.chunk_vec.push_back(ChunkedTerms::Branch {
+            branch_nums: Vec::with_capacity(capacity),
+            arms: Vec::with_capacity(capacity),
+        });
     }
 
     #[inline]
     pub fn add_chunk(&mut self) {
-        let chunk = ChunkedTerms::Chunk {
+        self.chunk_vec.push_back(ChunkedTerms::Chunk {
             terms: VecDeque::from(vec![]),
-        };
-        self.chunk_vec.push_back(chunk);
+        });
     }
 
     pub fn push_chunk_term(&mut self, term: QueryTerm) {
         match self.chunk_vec.back_mut() {
-            Some(ChunkedTerms::Branch(_)) => {
+            Some(ChunkedTerms::Branch { .. }) => {
                 let chunk = ChunkedTerms::Chunk {
                     terms: VecDeque::from(vec![term]),
                 };