Refactor optimizer with aggressive collapsing

The optimizer has been refactored to separate its phases and introduce
more aggressive collapsing capabilities.

Phase 2 (Cracking) now focuses on stateless transformations, making it
easier to reason about and potentially parallelize.

Phase 2.5 (Aggressive Collapsing) has been introduced, enabling
optimizations like:
- Beta-reduction for lambda literals.
- Cracking and inlining for constant closures.
- Folding intrinsics for constant arithmetic.
- Short-circuiting conditional expressions.

These changes aim to improve performance by reducing redundant
computations and code bloat. The maximum number of optimization passes
has also been increased to 5 to allow for more complex transformations.
This commit is contained in:
Michael Schimmel
2026-02-21 19:42:09 +01:00
parent 0bbe35eeec
commit 56b8e8389b
5 changed files with 434 additions and 202 deletions
+316 -119
View File
@@ -3,17 +3,17 @@ use crate::ast::compiler::bound_nodes::{BoundKind, TypedNode, Address};
use crate::ast::types::{Value, StaticType}; use crate::ast::types::{Value, StaticType};
use crate::ast::vm::Closure; use crate::ast::vm::Closure;
use crate::ast::nodes::Node; use crate::ast::nodes::Node;
use std::collections::HashMap;
/// The Optimizer performs Phase 2: Partial Evaluation & Closure Cracking. /// The Optimizer performs Phase 2 (Cracking) and Phase 2.5 (Aggressive Collapsing).
pub struct Optimizer { pub struct Optimizer {
/// 0: None, 1: Cracking (Stateless transformation), 2: Aggressive (Collapsing)
pub level: u32, pub level: u32,
max_passes: usize, max_passes: usize,
} }
impl Optimizer { impl Optimizer {
pub fn new(level: u32) -> Self { pub fn new(level: u32) -> Self {
Self { level, max_passes: 3 } Self { level, max_passes: 5 }
} }
pub fn optimize(&self, node: TypedNode) -> TypedNode { pub fn optimize(&self, node: TypedNode) -> TypedNode {
@@ -24,6 +24,7 @@ impl Optimizer {
let mut current = node; let mut current = node;
for _ in 0..self.max_passes { for _ in 0..self.max_passes {
let next = self.visit_node(current.clone()); let next = self.visit_node(current.clone());
// TODO: Structural equality check to stop early
current = next; current = next;
} }
current current
@@ -31,84 +32,79 @@ impl Optimizer {
fn visit_node(&self, node: TypedNode) -> TypedNode { fn visit_node(&self, node: TypedNode) -> TypedNode {
let (new_kind, new_ty) = match node.kind { let (new_kind, new_ty) = match node.kind {
BoundKind::Call { callee, args } => { BoundKind::Call { callee, args } | BoundKind::TailCall { callee, args } => {
let callee = self.visit_node(*callee); let callee = self.visit_node(*callee);
let args = self.visit_node(*args); let args = self.visit_node(*args);
// --- Cracking & Collapsing --- if self.level >= 2 {
if let BoundKind::Constant(Value::Object(ref obj)) = callee.kind { // Case 1: Beta-Reduction for Lambda Literals
if let Some(closure) = obj.as_any().downcast_ref::<Closure>() { if let BoundKind::Lambda { params, body, .. } = &callee.kind
// Level 1+: Inlining Upvalues (Stateless transformation) && let Some(collapsed) = self.try_beta_reduce(params, &args, (**body).clone()) {
// Structure is PRESERVED (Call still exists), but code is specialized. return self.visit_node(collapsed);
let body = (*closure.function_node).clone(); }
let inlined_body = self.inline_upvalues(body, &closure.upvalues, 0);
// Case 2: Cracking and Inlining for Constant Closures
// Level 2: Aggressive Collapsing if let BoundKind::Constant(Value::Object(ref obj)) = callee.kind
if self.level >= 2 { && let Some(closure) = obj.as_any().downcast_ref::<Closure>() {
// 1. Collapse 0-parameter calls let mut sub = SubstitutionMap::new();
if let Some(0) = closure.positional_count { for (i, cell) in closure.upvalues.iter().enumerate() {
let is_empty_args = match &args.kind { sub.add_upvalue(i as u32, cell.borrow().clone());
BoundKind::Nop => true, }
BoundKind::Tuple { elements } => elements.is_empty(), let inlined_body = sub.inline((*closure.function_node).clone());
_ => false,
}; // Try to collapse the now-stateless lambda call
if is_empty_args { if let Some(collapsed) = self.try_beta_reduce(&closure.parameter_node, &args, inlined_body) {
return self.visit_node(inlined_body); return self.visit_node(collapsed);
}
} }
// 2. Beta-Reduction (TODO: Inlining parameters)
} }
// Stateless transformation (Cracking)
let cracked_lambda = Node {
identity: callee.identity.clone(),
ty: callee.ty.clone(),
kind: BoundKind::Lambda {
params: closure.parameter_node.clone(),
upvalues: vec![],
body: Rc::new(inlined_body),
positional_count: closure.positional_count,
},
};
return Node {
identity: node.identity,
kind: BoundKind::Call {
callee: Box::new(cracked_lambda),
args: Box::new(args)
},
ty: node.ty,
};
}
}
// Level 2: Intrinsic Folding (Arithmetic on constants)
if self.level >= 2 {
if let Some(folded) = self.try_fold_intrinsic(&callee, &args) { if let Some(folded) = self.try_fold_intrinsic(&callee, &args) {
return folded; return folded;
} }
} }
// Level 1: Just Crack closures but keep the Call structure
if self.level >= 1
&& let BoundKind::Constant(Value::Object(ref obj)) = callee.kind
&& let Some(closure) = obj.as_any().downcast_ref::<Closure>() {
let mut sub = SubstitutionMap::new();
for (i, cell) in closure.upvalues.iter().enumerate() {
sub.add_upvalue(i as u32, cell.borrow().clone());
}
let inlined_body = sub.inline((*closure.function_node).clone());
let cracked_lambda = Node {
identity: callee.identity.clone(),
ty: callee.ty.clone(),
kind: BoundKind::Lambda {
params: closure.parameter_node.clone(),
upvalues: vec![], // CRACKED: Stateless
body: Rc::new(inlined_body),
positional_count: closure.positional_count,
},
};
return Node {
identity: node.identity,
kind: BoundKind::Call { callee: Box::new(cracked_lambda), args: Box::new(args) },
ty: node.ty,
};
}
(BoundKind::Call { callee: Box::new(callee), args: Box::new(args) }, node.ty) (BoundKind::Call { callee: Box::new(callee), args: Box::new(args) }, node.ty)
}, },
BoundKind::If { cond, then_br, else_br } => { BoundKind::If { cond, then_br, else_br } => {
let cond = self.visit_node(*cond); let cond = self.visit_node(*cond);
if self.level >= 2 { if self.level >= 2
if let BoundKind::Constant(ref val) = cond.kind { && let BoundKind::Constant(ref val) = cond.kind {
if val.is_truthy() { if val.is_truthy() {
return self.visit_node(*then_br); return self.visit_node(*then_br);
} else if let Some(else_node) = else_br { } else if let Some(else_node) = else_br {
return self.visit_node(*else_node); return self.visit_node(*else_node);
} else { } else {
return Node { return Node { identity: node.identity, kind: BoundKind::Nop, ty: StaticType::Void };
identity: node.identity,
kind: BoundKind::Nop,
ty: StaticType::Void,
};
} }
} }
}
let then_br = Box::new(self.visit_node(*then_br)); let then_br = Box::new(self.visit_node(*then_br));
let else_br = else_br.map(|e| Box::new(self.visit_node(*e))); let else_br = else_br.map(|e| Box::new(self.visit_node(*e)));
@@ -137,12 +133,13 @@ impl Optimizer {
(BoundKind::Block { exprs: new_exprs }, ty) (BoundKind::Block { exprs: new_exprs }, ty)
}, },
// Recursive Traversal
BoundKind::Lambda { params, upvalues, body, positional_count } => { BoundKind::Lambda { params, upvalues, body, positional_count } => {
let params = Rc::new(self.visit_node(params.as_ref().clone())); let params = Rc::new(self.visit_node(params.as_ref().clone()));
let body = Rc::new(self.visit_node(body.as_ref().clone())); let body = Rc::new(self.visit_node(body.as_ref().clone()));
(BoundKind::Lambda { params, upvalues, body, positional_count }, node.ty) (BoundKind::Lambda { params, upvalues, body, positional_count }, node.ty)
}, },
// Standard Recursive Traversal
BoundKind::DefLocal { name, slot, value, captured_by } => { BoundKind::DefLocal { name, slot, value, captured_by } => {
let value = Box::new(self.visit_node(*value)); let value = Box::new(self.visit_node(*value));
(BoundKind::DefLocal { name, slot, value, captured_by }, node.ty) (BoundKind::DefLocal { name, slot, value, captured_by }, node.ty)
@@ -167,108 +164,308 @@ impl Optimizer {
let bound_expanded = Box::new(self.visit_node(*bound_expanded)); let bound_expanded = Box::new(self.visit_node(*bound_expanded));
(BoundKind::Expansion { original_call, bound_expanded }, node.ty) (BoundKind::Expansion { original_call, bound_expanded }, node.ty)
}, },
k => (k, node.ty), k => (k, node.ty),
}; };
Node { Node { identity: node.identity, kind: new_kind, ty: new_ty }
identity: node.identity, }
kind: new_kind,
ty: new_ty, fn try_beta_reduce(&self, params: &TypedNode, args: &TypedNode, body: TypedNode) -> Option<TypedNode> {
// SAFETY: Only beta-reduce if the body does not contain local definitions.
if self.contains_def_local(&body) {
return None;
}
let mut sub = SubstitutionMap::new();
let mut arg_vals = Vec::new();
self.flatten_typed_tuple(args, &mut arg_vals);
let mut slot_index = 0;
self.map_params_to_args(params, &arg_vals, &mut slot_index, &mut sub);
// SAFETY: Only beta-reduce if ALL provided arguments were successfully inlined as constants.
// If we leave any Local(i) references, the VM will crash because the frame is gone.
if sub.locals.len() < arg_vals.len() {
return None;
}
if sub.is_empty() && !arg_vals.is_empty() {
return None;
}
Some(sub.inline(body))
}
fn flatten_typed_tuple(&self, node: &TypedNode, into: &mut Vec<TypedNode>) {
if let BoundKind::Tuple { elements } = &node.kind {
for el in elements {
self.flatten_typed_tuple(el, into);
}
} else if !matches!(node.kind, BoundKind::Nop) {
into.push(node.clone());
}
}
fn map_params_to_args(&self, pattern: &TypedNode, args: &[TypedNode], offset: &mut usize, sub: &mut SubstitutionMap) {
match &pattern.kind {
BoundKind::Parameter { slot, .. } => {
if let Some(arg) = args.get(*offset)
&& let BoundKind::Constant(val) = &arg.kind {
sub.add_local(*slot, val.clone());
}
*offset += 1;
}
BoundKind::Tuple { elements } => {
for el in elements {
self.map_params_to_args(el, args, offset, sub);
}
}
_ => {}
} }
} }
fn try_fold_intrinsic(&self, callee: &TypedNode, args: &TypedNode) -> Option<TypedNode> { fn try_fold_intrinsic(&self, callee: &TypedNode, args: &TypedNode) -> Option<TypedNode> {
// Simple constant folding for + and - on Ints if let BoundKind::Get { name, .. } = &callee.kind
if let BoundKind::Get { name, .. } = &callee.kind { && let BoundKind::Tuple { elements } = &args.kind
if let BoundKind::Tuple { elements } = &args.kind { && elements.len() == 2
if elements.len() == 2 { {
if let (BoundKind::Constant(Value::Int(a)), BoundKind::Constant(Value::Int(b))) = (&elements[0].kind, &elements[1].kind) { let val_a = self.get_const_int(&elements[0]);
let res = match &*name.name { let val_b = self.get_const_int(&elements[1]);
"+" => Some(Value::Int(a + b)),
"-" => Some(Value::Int(a - b)), if let (Some(a), Some(b)) = (val_a, val_b) {
"*" => Some(Value::Int(a * b)), let res = match &*name.name {
"/" if *b != 0 => Some(Value::Int(a / b)), "+" => Some(Value::Int(a + b)),
_ => None "-" => Some(Value::Int(a - b)),
}; "*" => Some(Value::Int(a * b)),
if let Some(val) = res { "/" if b != 0 => Some(Value::Int(a / b)),
return Some(Node { _ => None
identity: callee.identity.clone(), };
ty: StaticType::Int, if let Some(val) = res {
kind: BoundKind::Constant(val), return Some(Node {
}); identity: callee.identity.clone(),
} ty: StaticType::Int,
} kind: BoundKind::Constant(val),
});
} }
} }
} }
None None
} }
fn inline_upvalues(&self, node: TypedNode, upvalues: &[Rc<std::cell::RefCell<Value>>], depth: usize) -> TypedNode { fn get_const_int(&self, node: &TypedNode) -> Option<i64> {
match &node.kind {
BoundKind::Constant(Value::Int(i)) => Some(*i),
BoundKind::Tuple { elements } if elements.len() == 1 => self.get_const_int(&elements[0]),
_ => None
}
}
fn contains_def_local(&self, node: &TypedNode) -> bool {
match &node.kind {
BoundKind::DefLocal { .. } => true,
BoundKind::If { cond, then_br, else_br } => {
self.contains_def_local(cond) || self.contains_def_local(then_br) || else_br.as_ref().is_some_and(|e| self.contains_def_local(e))
}
BoundKind::Block { exprs } => exprs.iter().any(|e| self.contains_def_local(e)),
BoundKind::Call { callee, args } | BoundKind::TailCall { callee, args } => self.contains_def_local(callee) || self.contains_def_local(args),
BoundKind::Lambda { .. } => false, // Nested lambda definitions are safe (they have their own frame)
BoundKind::Tuple { elements } => elements.iter().any(|e| self.contains_def_local(e)),
BoundKind::Record { fields } => fields.iter().any(|(k, v)| self.contains_def_local(k) || self.contains_def_local(v)),
BoundKind::Set { value, .. } => self.contains_def_local(value),
_ => false,
}
}
}
/// Helper for transitively inlining values and cleaning up capture lists.
struct SubstitutionMap {
/// Mapping Address::Local(slot) -> Value
locals: HashMap<u32, Value>,
/// Mapping Address::Upvalue(idx) -> Value
upvalues: HashMap<u32, Value>,
}
impl SubstitutionMap {
fn new() -> Self {
Self { locals: HashMap::new(), upvalues: HashMap::new() }
}
fn add_local(&mut self, slot: u32, val: Value) {
self.locals.insert(slot, val);
}
fn add_upvalue(&mut self, idx: u32, val: Value) {
self.upvalues.insert(idx, val);
}
fn is_empty(&self) -> bool {
self.locals.is_empty() && self.upvalues.is_empty()
}
fn inline(&self, node: TypedNode) -> TypedNode {
self.inline_recursive(node, 0, &HashMap::new())
}
fn inline_recursive(&self, node: TypedNode, depth: usize, upvalue_subs: &HashMap<u32, Value>) -> TypedNode {
let (new_kind, new_ty) = match node.kind {
BoundKind::Get { addr, name } => {
match addr {
Address::Local(slot) if depth == 0 => {
if let Some(val) = self.locals.get(&slot) {
(BoundKind::Constant(val.clone()), val.static_type())
} else { (BoundKind::Get { addr, name }, node.ty) }
}
Address::Upvalue(idx) if depth == 0 => {
if let Some(val) = self.upvalues.get(&idx) {
(BoundKind::Constant(val.clone()), val.static_type())
} else { (BoundKind::Get { addr, name }, node.ty) }
}
Address::Upvalue(idx) if depth > 0 => {
if let Some(val) = upvalue_subs.get(&idx) {
(BoundKind::Constant(val.clone()), val.static_type())
} else { (BoundKind::Get { addr, name }, node.ty) }
}
_ => (BoundKind::Get { addr, name }, node.ty)
}
},
BoundKind::Lambda { params, upvalues: original_upvalues, body, positional_count } => {
// Phase 3: Un-Capturing
// If this lambda captures variables that we just inlined, we MUST remove them
// from its capture list to avoid "Stack underflow capture" errors in the VM.
let mut new_upvalues = Vec::new();
let mut next_inner_subs = HashMap::new();
let mut mapping = Vec::new(); // Old Upvalue Index -> New Upvalue Index (or None)
for (old_idx, capture_addr) in original_upvalues.iter().enumerate() {
let mut inlined_val = None;
if depth == 0 {
match capture_addr {
Address::Local(slot) => {
if let Some(val) = self.locals.get(slot) { inlined_val = Some(val.clone()); }
}
Address::Upvalue(idx) => {
if let Some(val) = self.upvalues.get(idx) { inlined_val = Some(val.clone()); }
}
_ => {}
}
} else if let Address::Upvalue(old_parent_idx) = capture_addr
&& let Some(val) = upvalue_subs.get(old_parent_idx) {
inlined_val = Some(val.clone());
}
if let Some(val) = inlined_val {
next_inner_subs.insert(old_idx as u32, val);
mapping.push(None);
} else {
mapping.push(Some(new_upvalues.len() as u32));
new_upvalues.push(*capture_addr);
}
}
// If any upvalues were removed, we must re-index the Get(Upvalue) nodes in the body.
let body_node = self.inline_recursive(body.as_ref().clone(), depth + 1, &next_inner_subs);
let reindexed_body = if new_upvalues.len() != original_upvalues.len() {
self.reindex_upvalues(body_node, &mapping)
} else {
body_node
};
(BoundKind::Lambda { params, upvalues: new_upvalues, body: Rc::new(reindexed_body), positional_count }, node.ty)
},
// Boilerplate Recursive Traversal
BoundKind::If { cond, then_br, else_br } => {
let cond = Box::new(self.inline_recursive(*cond, depth, upvalue_subs));
let then_br = Box::new(self.inline_recursive(*then_br, depth, upvalue_subs));
let else_br = else_br.map(|e| Box::new(self.inline_recursive(*e, depth, upvalue_subs)));
(BoundKind::If { cond, then_br, else_br }, node.ty)
},
BoundKind::Block { exprs } => {
let exprs = exprs.into_iter().map(|e| self.inline_recursive(e, depth, upvalue_subs)).collect();
(BoundKind::Block { exprs }, node.ty)
},
BoundKind::Call { callee, args } | BoundKind::TailCall { callee, args } => {
let callee = Box::new(self.inline_recursive(*callee, depth, upvalue_subs));
let args = Box::new(self.inline_recursive(*args, depth, upvalue_subs));
(BoundKind::Call { callee, args }, node.ty)
},
BoundKind::DefLocal { name, slot, value, captured_by } => {
let value = Box::new(self.inline_recursive(*value, depth, upvalue_subs));
(BoundKind::DefLocal { name, slot, value, captured_by }, node.ty)
},
BoundKind::Set { addr, value } => {
let value = Box::new(self.inline_recursive(*value, depth, upvalue_subs));
(BoundKind::Set { addr, value }, node.ty)
},
BoundKind::Tuple { elements } => {
let elements = elements.into_iter().map(|e| self.inline_recursive(e, depth, upvalue_subs)).collect();
(BoundKind::Tuple { elements }, node.ty)
},
BoundKind::Record { fields } => {
let fields = fields.into_iter().map(|(k, v)| (self.inline_recursive(k, depth, upvalue_subs), self.inline_recursive(v, depth, upvalue_subs))).collect();
(BoundKind::Record { fields }, node.ty)
},
k => (k, node.ty),
};
Node { identity: node.identity, kind: new_kind, ty: new_ty }
}
/// Re-maps Get(Upvalue(old_idx)) to Get(Upvalue(new_idx)) based on a mapping table.
fn reindex_upvalues(&self, node: TypedNode, mapping: &[Option<u32>]) -> TypedNode {
let (new_kind, new_ty) = match node.kind { let (new_kind, new_ty) = match node.kind {
BoundKind::Get { addr: Address::Upvalue(idx), name } => { BoundKind::Get { addr: Address::Upvalue(idx), name } => {
if depth == 0 { if let Some(Some(new_idx)) = mapping.get(idx as usize) {
if let Some(cell) = upvalues.get(idx as usize) { (BoundKind::Get { addr: Address::Upvalue(*new_idx), name }, node.ty)
let val = cell.borrow().clone();
let ty = val.static_type();
(BoundKind::Constant(val), ty)
} else {
(BoundKind::Get { addr: Address::Upvalue(idx), name }, node.ty)
}
} else { } else {
(BoundKind::Get { addr: Address::Upvalue(idx), name }, node.ty) (BoundKind::Get { addr: Address::Upvalue(idx), name }, node.ty)
} }
}, },
// Nested Lambda: Increment depth (we only re-index upvalues of the current scope)
BoundKind::Lambda { params, upvalues, body, positional_count } => {
// IMPORTANT: In MyC, nested upvalues refer to the capture list of the nested lambda.
// We DON'T re-index inside nested lambdas because their upvalue list is separate.
(BoundKind::Lambda { params, upvalues, body, positional_count }, node.ty)
},
// Recursive Traversal
BoundKind::If { cond, then_br, else_br } => { BoundKind::If { cond, then_br, else_br } => {
let cond = Box::new(self.inline_upvalues(*cond, upvalues, depth)); let cond = Box::new(self.reindex_upvalues(*cond, mapping));
let then_br = Box::new(self.inline_upvalues(*then_br, upvalues, depth)); let then_br = Box::new(self.reindex_upvalues(*then_br, mapping));
let else_br = else_br.map(|e| Box::new(self.inline_upvalues(*e, upvalues, depth))); let else_br = else_br.map(|e| Box::new(self.reindex_upvalues(*e, mapping)));
(BoundKind::If { cond, then_br, else_br }, node.ty) (BoundKind::If { cond, then_br, else_br }, node.ty)
}, },
BoundKind::Block { exprs } => { BoundKind::Block { exprs } => {
let exprs = exprs.into_iter().map(|e| self.inline_upvalues(e, upvalues, depth)).collect(); let exprs = exprs.into_iter().map(|e| self.reindex_upvalues(e, mapping)).collect();
(BoundKind::Block { exprs }, node.ty) (BoundKind::Block { exprs }, node.ty)
}, },
BoundKind::Call { callee, args } => { BoundKind::Call { callee, args } | BoundKind::TailCall { callee, args } => {
let callee = Box::new(self.inline_upvalues(*callee, upvalues, depth)); let callee = Box::new(self.reindex_upvalues(*callee, mapping));
let args = Box::new(self.inline_upvalues(*args, upvalues, depth)); let args = Box::new(self.reindex_upvalues(*args, mapping));
(BoundKind::Call { callee, args }, node.ty) (BoundKind::Call { callee, args }, node.ty)
}, },
BoundKind::TailCall { callee, args } => {
let callee = Box::new(self.inline_upvalues(*callee, upvalues, depth));
let args = Box::new(self.inline_upvalues(*args, upvalues, depth));
(BoundKind::TailCall { callee, args }, node.ty)
},
BoundKind::DefLocal { name, slot, value, captured_by } => { BoundKind::DefLocal { name, slot, value, captured_by } => {
let value = Box::new(self.inline_upvalues(*value, upvalues, depth)); let value = Box::new(self.reindex_upvalues(*value, mapping));
(BoundKind::DefLocal { name, slot, value, captured_by }, node.ty) (BoundKind::DefLocal { name, slot, value, captured_by }, node.ty)
}, },
BoundKind::Set { addr, value } => { BoundKind::Set { addr, value } => {
let value = Box::new(self.inline_upvalues(*value, upvalues, depth)); let value = Box::new(self.reindex_upvalues(*value, mapping));
(BoundKind::Set { addr, value }, node.ty) (BoundKind::Set { addr, value }, node.ty)
}, },
BoundKind::Tuple { elements } => { BoundKind::Tuple { elements } => {
let elements = elements.into_iter().map(|e| self.inline_upvalues(e, upvalues, depth)).collect(); let elements = elements.into_iter().map(|e| self.reindex_upvalues(e, mapping)).collect();
(BoundKind::Tuple { elements }, node.ty) (BoundKind::Tuple { elements }, node.ty)
}, },
BoundKind::Record { fields } => { BoundKind::Record { fields } => {
let fields = fields.into_iter().map(|(k, v)| (self.inline_upvalues(k, upvalues, depth), self.inline_upvalues(v, upvalues, depth))).collect(); let fields = fields.into_iter().map(|(k, v)| (self.reindex_upvalues(k, mapping), self.reindex_upvalues(v, mapping))).collect();
(BoundKind::Record { fields }, node.ty) (BoundKind::Record { fields }, node.ty)
}, },
BoundKind::Lambda { params, upvalues: lambda_upvalues, body, positional_count } => {
let body = Rc::new(self.inline_upvalues(body.as_ref().clone(), upvalues, depth + 1));
(BoundKind::Lambda { params, upvalues: lambda_upvalues, body, positional_count }, node.ty)
},
k => (k, node.ty), k => (k, node.ty),
}; };
Node { Node { identity: node.identity, kind: new_kind, ty: new_ty }
identity: node.identity,
kind: new_kind,
ty: new_ty,
}
} }
} }
+98 -72
View File
@@ -1,22 +1,22 @@
use std::rc::Rc; use crate::ast::compiler::binder::Binder;
use crate::ast::compiler::{TypeChecker, TypedNode};
use crate::ast::nodes::{Node, Symbol, UntypedKind};
use crate::ast::parser::Parser;
use crate::ast::types::{Object, StaticType, Value};
use crate::ast::vm::{TracingObserver, VM};
use std::cell::RefCell; use std::cell::RefCell;
use std::collections::HashMap; use std::collections::HashMap;
use crate::ast::types::{Value, StaticType, Object}; use std::rc::Rc;
use crate::ast::nodes::{Node, UntypedKind, Symbol};
use crate::ast::parser::Parser;
use crate::ast::compiler::binder::Binder;
use crate::ast::compiler::{TypedNode, TypeChecker};
use crate::ast::vm::{VM, TracingObserver};
use crate::ast::compiler::tco::TCO; use crate::ast::compiler::bound_nodes::{Address, BoundNode};
use crate::ast::compiler::lambda_collector::LambdaCollector;
use crate::ast::compiler::dumper::Dumper; use crate::ast::compiler::dumper::Dumper;
use crate::ast::compiler::macros::{MacroExpander, MacroRegistry, MacroEvaluator}; use crate::ast::compiler::lambda_collector::LambdaCollector;
use crate::ast::compiler::specializer::{Specializer, MonoCache, FunctionRegistry}; use crate::ast::compiler::macros::{MacroEvaluator, MacroExpander, MacroRegistry};
use crate::ast::compiler::optimizer::Optimizer; use crate::ast::compiler::optimizer::Optimizer;
use crate::ast::compiler::specializer::{FunctionRegistry, MonoCache, Specializer};
use crate::ast::compiler::tco::TCO;
use crate::ast::rtl; use crate::ast::rtl;
use crate::ast::rtl::intrinsics; use crate::ast::rtl::intrinsics;
use crate::ast::compiler::bound_nodes::{Address, BoundNode};
pub struct Environment { pub struct Environment {
pub global_names: Rc<RefCell<HashMap<Symbol, u32>>>, pub global_names: Rc<RefCell<HashMap<Symbol, u32>>>,
@@ -50,19 +50,24 @@ struct RuntimeMacroEvaluator {
} }
impl MacroEvaluator for RuntimeMacroEvaluator { impl MacroEvaluator for RuntimeMacroEvaluator {
fn evaluate(&self, node: &Node<UntypedKind>, bindings: &HashMap<Rc<str>, Node<UntypedKind>>) -> Result<Value, String> { fn evaluate(
&self,
node: &Node<UntypedKind>,
bindings: &HashMap<Rc<str>, Node<UntypedKind>>,
) -> Result<Value, String> {
// 1. Check if it's a simple parameter substitution // 1. Check if it's a simple parameter substitution
if let UntypedKind::Identifier(sym) = &node.kind if let UntypedKind::Identifier(sym) = &node.kind
&& let Some(arg_node) = bindings.get(&sym.name) { && let Some(arg_node) = bindings.get(&sym.name)
return Ok(Value::Object(Rc::new(arg_node.clone()) as Rc<dyn Object>)); {
return Ok(Value::Object(Rc::new(arg_node.clone()) as Rc<dyn Object>));
} }
// 2. Full evaluation for complex compile-time expressions // 2. Full evaluation for complex compile-time expressions
let bound_ast = Binder::bind_root(self.global_names.clone(), node)?; let bound_ast = Binder::bind_root(self.global_names.clone(), node)?;
let checker = TypeChecker::new(self.global_types.clone()); let checker = TypeChecker::new(self.global_types.clone());
let typed_ast = checker.check(bound_ast, &[])?; let typed_ast = checker.check(bound_ast, &[])?;
let mut vm = VM::new(self.global_values.clone()); let mut vm = VM::new(self.global_values.clone());
vm.run(&typed_ast) vm.run(&typed_ast)
} }
@@ -102,7 +107,12 @@ impl Environment {
MacroExpander::new(MacroRegistry::new(), evaluator) MacroExpander::new(MacroRegistry::new(), evaluator)
} }
pub fn register_native(&self, name: &str, ty: StaticType, func: impl Fn(Vec<Value>) -> Value + 'static) { pub fn register_native(
&self,
name: &str,
ty: StaticType,
func: impl Fn(Vec<Value>) -> Value + 'static,
) {
let mut names = self.global_names.borrow_mut(); let mut names = self.global_names.borrow_mut();
let mut types = self.global_types.borrow_mut(); let mut types = self.global_types.borrow_mut();
let mut values = self.global_values.borrow_mut(); let mut values = self.global_values.borrow_mut();
@@ -143,7 +153,10 @@ impl Environment {
// 2. Check for trailing tokens // 2. Check for trailing tokens
if !parser.at_eof() { if !parser.at_eof() {
return Err("Unexpected trailing expressions in script. Use (do ...) for sequences.".to_string()); return Err(
"Unexpected trailing expressions in script. Use (do ...) for sequences."
.to_string(),
);
} }
// 3. Expand Macros // 3. Expand Macros
@@ -179,64 +192,71 @@ impl Environment {
let registry = Rc::new(EnvFunctionRegistry { let registry = Rc::new(EnvFunctionRegistry {
registry: self.function_registry.clone(), registry: self.function_registry.clone(),
}); });
let rtl_lookup = Rc::new(|name: &str, args: &[StaticType]| intrinsics::lookup(name, args)); let rtl_lookup = Rc::new(|name: &str, args: &[StaticType]| intrinsics::lookup(name, args));
let func_reg = self.function_registry.clone(); let func_reg = self.function_registry.clone();
let mono_cache = self.monomorph_cache.clone(); let mono_cache = self.monomorph_cache.clone();
let global_values = self.global_values.clone(); let global_values = self.global_values.clone();
let global_types = self.global_types.clone(); let global_types = self.global_types.clone();
let opt_level = self.optimization_level; let opt_level = self.optimization_level;
let compiler = Rc::new(move |func_template: BoundNode, arg_types: &[StaticType]| -> Result<(Value, StaticType), String> {
// 1. Re-TypeCheck the template with concrete argument types
let checker = TypeChecker::new(global_types.clone());
let retyped_ast = checker.check(func_template, arg_types)?;
// 2. Specialize (Recursive)
let sub_registry = Rc::new(EnvFunctionRegistry { registry: func_reg.clone() });
let sub_rtl_lookup = Rc::new(|name: &str, args: &[StaticType]| intrinsics::lookup(name, args));
let sub_specializer = Specializer::new(
Some(sub_registry),
None,
Some(sub_rtl_lookup),
Some(mono_cache.clone())
);
let specialized_ast = sub_specializer.specialize(retyped_ast);
// 3. Optimize (Phase 2: Cracking & Folding)
let optimizer = Optimizer::new(opt_level);
let optimized_ast = optimizer.optimize(specialized_ast);
// 4. TCO
let tco_ast = TCO::optimize(optimized_ast);
// 5. Compile to Value (VM)
let mut vm = VM::new(global_values.clone());
let compiled_val = match vm.run(&tco_ast) {
Ok(v) => v,
Err(e) => return Err(format!("VM Error during specialization: {}", e)),
};
// 6. Determine correct return type from the newly inferred function signature let compiler = Rc::new(
let ret_type = if let StaticType::Function(sig) = &tco_ast.ty { move |func_template: BoundNode,
sig.ret.clone() arg_types: &[StaticType]|
} else { -> Result<(Value, StaticType), String> {
StaticType::Any // 1. Re-TypeCheck the template with concrete argument types
}; let checker = TypeChecker::new(global_types.clone());
let retyped_ast = checker.check(func_template, arg_types)?;
// 2. Specialize (Recursive)
let sub_registry = Rc::new(EnvFunctionRegistry {
registry: func_reg.clone(),
});
let sub_rtl_lookup =
Rc::new(|name: &str, args: &[StaticType]| intrinsics::lookup(name, args));
let sub_specializer = Specializer::new(
Some(sub_registry),
None,
Some(sub_rtl_lookup),
Some(mono_cache.clone()),
);
let specialized_ast = sub_specializer.specialize(retyped_ast);
// 3. Optimize (Phase 2: Cracking & Folding)
let optimizer = Optimizer::new(opt_level);
let optimized_ast = optimizer.optimize(specialized_ast);
// 4. TCO
let tco_ast = TCO::optimize(optimized_ast);
// 5. Compile to Value (VM)
let mut vm = VM::new(global_values.clone());
let compiled_val = match vm.run(&tco_ast) {
Ok(v) => v,
Err(e) => return Err(format!("VM Error during specialization: {}", e)),
};
// 6. Determine correct return type from the newly inferred function signature
let ret_type = if let StaticType::Function(sig) = &tco_ast.ty {
sig.ret.clone()
} else {
StaticType::Any
};
Ok((compiled_val, ret_type))
},
);
Ok((compiled_val, ret_type))
});
let specializer = Specializer::new( let specializer = Specializer::new(
Some(registry), Some(registry),
Some(compiler), Some(compiler),
Some(rtl_lookup), Some(rtl_lookup),
Some(self.monomorph_cache.clone()), Some(self.monomorph_cache.clone()),
); );
specializer.specialize(node) specializer.specialize(node)
} }
@@ -244,10 +264,10 @@ impl Environment {
pub fn run(&self, node: &TypedNode) -> Result<Value, String> { pub fn run(&self, node: &TypedNode) -> Result<Value, String> {
let mut vm = VM::new(self.global_values.clone()); let mut vm = VM::new(self.global_values.clone());
let mut result = vm.run(node)?; let mut result = vm.run(node)?;
// Handle potential script body closure // Handle potential script body closure
if let Value::Object(obj) = &result if let Value::Object(obj) = &result
&& let Some(closure) = obj.as_any().downcast_ref::<crate::ast::vm::Closure>() && let Some(closure) = obj.as_any().downcast_ref::<crate::ast::vm::Closure>()
{ {
result = vm.run(&closure.function_node)?; result = vm.run(&closure.function_node)?;
} }
@@ -258,7 +278,10 @@ impl Environment {
if let Some(closure) = next_obj.as_any().downcast_ref::<crate::ast::vm::Closure>() { if let Some(closure) = next_obj.as_any().downcast_ref::<crate::ast::vm::Closure>() {
result = vm.run_with_args(closure, next_args)?; result = vm.run_with_args(closure, next_args)?;
} else { } else {
return Err(format!("Tail call target is not a closure: {}", next_obj.type_name())); return Err(format!(
"Tail call target is not a closure: {}",
next_obj.type_name()
));
} }
} }
Ok(result) Ok(result)
@@ -286,10 +309,10 @@ impl Environment {
let mut vm = VM::new(self.global_values.clone()); let mut vm = VM::new(self.global_values.clone());
let mut observer = TracingObserver::new(); let mut observer = TracingObserver::new();
let mut result = vm.run_with_observer(&mut observer, &linked); let mut result = vm.run_with_observer(&mut observer, &linked);
// If result is a closure (script entry), execute the body too // If result is a closure (script entry), execute the body too
if let Ok(Value::Object(obj)) = &result if let Ok(Value::Object(obj)) = &result
&& let Some(closure) = obj.as_any().downcast_ref::<crate::ast::vm::Closure>() && let Some(closure) = obj.as_any().downcast_ref::<crate::ast::vm::Closure>()
{ {
result = vm.run_with_observer(&mut observer, &closure.function_node); result = vm.run_with_observer(&mut observer, &closure.function_node);
} }
@@ -300,7 +323,10 @@ impl Environment {
if let Some(closure) = next_obj.as_any().downcast_ref::<crate::ast::vm::Closure>() { if let Some(closure) = next_obj.as_any().downcast_ref::<crate::ast::vm::Closure>() {
result = vm.run_with_args_observed(&mut observer, closure, next_args); result = vm.run_with_args_observed(&mut observer, closure, next_args);
} else { } else {
result = Err(format!("Tail call target is not a closure: {}", next_obj.type_name())); result = Err(format!(
"Tail call target is not a closure: {}",
next_obj.type_name()
));
break; break;
} }
} }
+5 -3
View File
@@ -82,9 +82,11 @@ mod tests {
#[test] #[test]
fn test_examples() { fn test_examples() {
let results = crate::utils::tester::run_functional_tests(); for level in 0..=2 {
for res in results { let results = crate::utils::tester::run_functional_tests_with_level(level);
assert!(res.success, "Example {} failed: {}", res.name, res.message); for res in results {
assert!(res.success, "Example {} failed at level {}: {}", res.name, level, res.message);
}
} }
} }
+7 -5
View File
@@ -95,12 +95,13 @@ impl CompilerApp {
// Reset environment for a fresh run // Reset environment for a fresh run
self.env = Environment::new(); self.env = Environment::new();
self.env.optimization_level = 2;
let res = (|| -> Result<(myc::ast::types::Value, std::time::Duration), String> { let res = (|| -> Result<(myc::ast::types::Value, std::time::Duration), String> {
let start_run = std::time::Instant::now(); let start_run = std::time::Instant::now();
let result = if self.debug_enabled { let result = if self.debug_enabled {
let (res, logs) = self.env.run_debug(&self.source_code)?; let (res, logs) = self.env.run_debug(&self.source_code)?;
// Batch-truncate logs for GUI performance // Batch-truncate logs for GUI performance
let mut display_logs = logs; let mut display_logs = logs;
if display_logs.len() > 1000 { if display_logs.len() > 1000 {
@@ -110,9 +111,10 @@ impl CompilerApp {
for line in &mut display_logs { for line in &mut display_logs {
if line.len() > 255 if line.len() > 255
&& let Some((idx, _)) = line.char_indices().nth(255) { && let Some((idx, _)) = line.char_indices().nth(255)
line.truncate(idx); {
line.push_str("..."); line.truncate(idx);
line.push_str("...");
} }
} }
@@ -131,7 +133,7 @@ impl CompilerApp {
Ok((val, d_run)) => { Ok((val, d_run)) => {
let d_total = start_total.elapsed(); let d_total = start_total.elapsed();
let now = chrono::Local::now().format("%H:%M:%S").to_string(); let now = chrono::Local::now().format("%H:%M:%S").to_string();
let mut stats = format!( let mut stats = format!(
"Execution Successful.\nResult: {}\n\nTotal Duration: {:?}\nFinished at: {}", "Execution Successful.\nResult: {}\n\nTotal Duration: {:?}\nFinished at: {}",
val, d_total, now, val, d_total, now,
+8 -3
View File
@@ -18,12 +18,17 @@ pub struct BenchmarkResult {
} }
pub fn run_functional_tests() -> Vec<TestResult> { pub fn run_functional_tests() -> Vec<TestResult> {
run_functional_tests_with_level(0)
}
pub fn run_functional_tests_with_level(level: u32) -> Vec<TestResult> {
let mut results = Vec::new(); let mut results = Vec::new();
let entries = fs::read_dir("examples").unwrap(); let entries = fs::read_dir("examples").unwrap();
let output_re = Regex::new(r";; Output: (.*)").unwrap(); let output_re = Regex::new(r";; Output: (.*)").unwrap();
for entry in entries.filter_map(|e| e.ok()) { for entry in entries.filter_map(|e| e.ok()) {
let env = Environment::new(); // Fresh environment per test file let mut env = Environment::new(); // Fresh environment per test file
env.optimization_level = level;
let path = entry.path(); let path = entry.path();
if path.extension().is_some_and(|ext| ext == "myc") { if path.extension().is_some_and(|ext| ext == "myc") {
let content = fs::read_to_string(&path).unwrap(); let content = fs::read_to_string(&path).unwrap();
@@ -47,14 +52,14 @@ pub fn run_functional_tests() -> Vec<TestResult> {
results.push(TestResult { results.push(TestResult {
name, name,
success: false, success: false,
message: format!("Expected {}, got {}", expected, val_str), message: format!("Level {}: Expected {}, got {}", level, expected, val_str),
}); });
} }
} }
Err(e) => results.push(TestResult { Err(e) => results.push(TestResult {
name, name,
success: false, success: false,
message: format!("Error: {}", e), message: format!("Level {}: Error: {}", level, e),
}), }),
} }
} }