109 lines
3.4 KiB
Rust
Raw Normal View History

2018-08-30 11:42:16 +02:00
use smallvec::{Array, SmallVec};
2019-12-22 17:42:04 -05:00
use std::ptr;
Overhaul `syntax::fold::Folder`. This commit changes `syntax::fold::Folder` from a functional style (where most methods take a `T` and produce a new `T`) to a more imperative style (where most methods take and modify a `&mut T`), and renames it `syntax::mut_visit::MutVisitor`. The first benefit is speed. The functional style does not require any reallocations, due to the use of `P::map` and `MoveMap::move_{,flat_}map`. However, every field in the AST must be overwritten; even those fields that are unchanged are overwritten with the same value. This causes a lot of unnecessary memory writes. The imperative style reduces instruction counts by 1--3% across a wide range of workloads, particularly incremental workloads. The second benefit is conciseness; the imperative style is usually more concise. E.g. compare the old functional style: ``` fn fold_abc(&mut self, abc: ABC) { ABC { a: fold_a(abc.a), b: fold_b(abc.b), c: abc.c, } } ``` with the imperative style: ``` fn visit_abc(&mut self, ABC { a, b, c: _ }: &mut ABC) { visit_a(a); visit_b(b); } ``` (The reductions get larger in more complex examples.) Overall, the patch removes over 200 lines of code -- even though the new code has more comments -- and a lot of the remaining lines have fewer characters. Some notes: - The old style used methods called `fold_*`. The new style mostly uses methods called `visit_*`, but there are a few methods that map a `T` to something other than a `T`, which are called `flat_map_*` (`T` maps to multiple `T`s) or `filter_map_*` (`T` maps to 0 or 1 `T`s). - `move_map.rs`/`MoveMap`/`move_map`/`move_flat_map` are renamed `map_in_place.rs`/`MapInPlace`/`map_in_place`/`flat_map_in_place` to reflect their slightly changed signatures. - Although this commit renames the `fold` module as `mut_visit`, it keeps it in the `fold.rs` file, so as not to confuse git. The next commit will rename the file.
2019-02-05 15:20:55 +11:00
pub trait MapInPlace<T>: Sized {
2019-12-22 17:42:04 -05:00
fn map_in_place<F>(&mut self, mut f: F)
where
F: FnMut(T) -> T,
{
Overhaul `syntax::fold::Folder`. This commit changes `syntax::fold::Folder` from a functional style (where most methods take a `T` and produce a new `T`) to a more imperative style (where most methods take and modify a `&mut T`), and renames it `syntax::mut_visit::MutVisitor`. The first benefit is speed. The functional style does not require any reallocations, due to the use of `P::map` and `MoveMap::move_{,flat_}map`. However, every field in the AST must be overwritten; even those fields that are unchanged are overwritten with the same value. This causes a lot of unnecessary memory writes. The imperative style reduces instruction counts by 1--3% across a wide range of workloads, particularly incremental workloads. The second benefit is conciseness; the imperative style is usually more concise. E.g. compare the old functional style: ``` fn fold_abc(&mut self, abc: ABC) { ABC { a: fold_a(abc.a), b: fold_b(abc.b), c: abc.c, } } ``` with the imperative style: ``` fn visit_abc(&mut self, ABC { a, b, c: _ }: &mut ABC) { visit_a(a); visit_b(b); } ``` (The reductions get larger in more complex examples.) Overall, the patch removes over 200 lines of code -- even though the new code has more comments -- and a lot of the remaining lines have fewer characters. Some notes: - The old style used methods called `fold_*`. The new style mostly uses methods called `visit_*`, but there are a few methods that map a `T` to something other than a `T`, which are called `flat_map_*` (`T` maps to multiple `T`s) or `filter_map_*` (`T` maps to 0 or 1 `T`s). - `move_map.rs`/`MoveMap`/`move_map`/`move_flat_map` are renamed `map_in_place.rs`/`MapInPlace`/`map_in_place`/`flat_map_in_place` to reflect their slightly changed signatures. - Although this commit renames the `fold` module as `mut_visit`, it keeps it in the `fold.rs` file, so as not to confuse git. The next commit will rename the file.
2019-02-05 15:20:55 +11:00
self.flat_map_in_place(|e| Some(f(e)))
2015-11-15 21:19:53 +01:00
}
Overhaul `syntax::fold::Folder`. This commit changes `syntax::fold::Folder` from a functional style (where most methods take a `T` and produce a new `T`) to a more imperative style (where most methods take and modify a `&mut T`), and renames it `syntax::mut_visit::MutVisitor`. The first benefit is speed. The functional style does not require any reallocations, due to the use of `P::map` and `MoveMap::move_{,flat_}map`. However, every field in the AST must be overwritten; even those fields that are unchanged are overwritten with the same value. This causes a lot of unnecessary memory writes. The imperative style reduces instruction counts by 1--3% across a wide range of workloads, particularly incremental workloads. The second benefit is conciseness; the imperative style is usually more concise. E.g. compare the old functional style: ``` fn fold_abc(&mut self, abc: ABC) { ABC { a: fold_a(abc.a), b: fold_b(abc.b), c: abc.c, } } ``` with the imperative style: ``` fn visit_abc(&mut self, ABC { a, b, c: _ }: &mut ABC) { visit_a(a); visit_b(b); } ``` (The reductions get larger in more complex examples.) Overall, the patch removes over 200 lines of code -- even though the new code has more comments -- and a lot of the remaining lines have fewer characters. Some notes: - The old style used methods called `fold_*`. The new style mostly uses methods called `visit_*`, but there are a few methods that map a `T` to something other than a `T`, which are called `flat_map_*` (`T` maps to multiple `T`s) or `filter_map_*` (`T` maps to 0 or 1 `T`s). - `move_map.rs`/`MoveMap`/`move_map`/`move_flat_map` are renamed `map_in_place.rs`/`MapInPlace`/`map_in_place`/`flat_map_in_place` to reflect their slightly changed signatures. - Although this commit renames the `fold` module as `mut_visit`, it keeps it in the `fold.rs` file, so as not to confuse git. The next commit will rename the file.
2019-02-05 15:20:55 +11:00
fn flat_map_in_place<F, I>(&mut self, f: F)
2019-12-22 17:42:04 -05:00
where
F: FnMut(T) -> I,
I: IntoIterator<Item = T>;
2015-11-15 21:19:53 +01:00
}
Overhaul `syntax::fold::Folder`. This commit changes `syntax::fold::Folder` from a functional style (where most methods take a `T` and produce a new `T`) to a more imperative style (where most methods take and modify a `&mut T`), and renames it `syntax::mut_visit::MutVisitor`. The first benefit is speed. The functional style does not require any reallocations, due to the use of `P::map` and `MoveMap::move_{,flat_}map`. However, every field in the AST must be overwritten; even those fields that are unchanged are overwritten with the same value. This causes a lot of unnecessary memory writes. The imperative style reduces instruction counts by 1--3% across a wide range of workloads, particularly incremental workloads. The second benefit is conciseness; the imperative style is usually more concise. E.g. compare the old functional style: ``` fn fold_abc(&mut self, abc: ABC) { ABC { a: fold_a(abc.a), b: fold_b(abc.b), c: abc.c, } } ``` with the imperative style: ``` fn visit_abc(&mut self, ABC { a, b, c: _ }: &mut ABC) { visit_a(a); visit_b(b); } ``` (The reductions get larger in more complex examples.) Overall, the patch removes over 200 lines of code -- even though the new code has more comments -- and a lot of the remaining lines have fewer characters. Some notes: - The old style used methods called `fold_*`. The new style mostly uses methods called `visit_*`, but there are a few methods that map a `T` to something other than a `T`, which are called `flat_map_*` (`T` maps to multiple `T`s) or `filter_map_*` (`T` maps to 0 or 1 `T`s). - `move_map.rs`/`MoveMap`/`move_map`/`move_flat_map` are renamed `map_in_place.rs`/`MapInPlace`/`map_in_place`/`flat_map_in_place` to reflect their slightly changed signatures. - Although this commit renames the `fold` module as `mut_visit`, it keeps it in the `fold.rs` file, so as not to confuse git. The next commit will rename the file.
2019-02-05 15:20:55 +11:00
impl<T> MapInPlace<T> for Vec<T> {
fn flat_map_in_place<F, I>(&mut self, mut f: F)
2019-12-22 17:42:04 -05:00
where
F: FnMut(T) -> I,
I: IntoIterator<Item = T>,
2015-11-15 21:19:53 +01:00
{
let mut read_i = 0;
let mut write_i = 0;
unsafe {
let mut old_len = self.len();
self.set_len(0); // make sure we just leak elements in case of panic
while read_i < old_len {
// move the read_i'th item out of the vector and map it
// to an iterator
let e = ptr::read(self.get_unchecked(read_i));
let iter = f(e).into_iter();
2015-11-15 21:19:53 +01:00
read_i += 1;
for e in iter {
2015-11-15 21:19:53 +01:00
if write_i < read_i {
ptr::write(self.get_unchecked_mut(write_i), e);
write_i += 1;
} else {
// If this is reached we ran out of space
// in the middle of the vector.
// However, the vector is in a valid state here,
// so we just do a somewhat inefficient insert.
self.set_len(old_len);
self.insert(write_i, e);
old_len = self.len();
self.set_len(0);
read_i += 1;
write_i += 1;
}
}
}
// write_i tracks the number of actually written new items.
self.set_len(write_i);
}
}
}
Overhaul `syntax::fold::Folder`. This commit changes `syntax::fold::Folder` from a functional style (where most methods take a `T` and produce a new `T`) to a more imperative style (where most methods take and modify a `&mut T`), and renames it `syntax::mut_visit::MutVisitor`. The first benefit is speed. The functional style does not require any reallocations, due to the use of `P::map` and `MoveMap::move_{,flat_}map`. However, every field in the AST must be overwritten; even those fields that are unchanged are overwritten with the same value. This causes a lot of unnecessary memory writes. The imperative style reduces instruction counts by 1--3% across a wide range of workloads, particularly incremental workloads. The second benefit is conciseness; the imperative style is usually more concise. E.g. compare the old functional style: ``` fn fold_abc(&mut self, abc: ABC) { ABC { a: fold_a(abc.a), b: fold_b(abc.b), c: abc.c, } } ``` with the imperative style: ``` fn visit_abc(&mut self, ABC { a, b, c: _ }: &mut ABC) { visit_a(a); visit_b(b); } ``` (The reductions get larger in more complex examples.) Overall, the patch removes over 200 lines of code -- even though the new code has more comments -- and a lot of the remaining lines have fewer characters. Some notes: - The old style used methods called `fold_*`. The new style mostly uses methods called `visit_*`, but there are a few methods that map a `T` to something other than a `T`, which are called `flat_map_*` (`T` maps to multiple `T`s) or `filter_map_*` (`T` maps to 0 or 1 `T`s). - `move_map.rs`/`MoveMap`/`move_map`/`move_flat_map` are renamed `map_in_place.rs`/`MapInPlace`/`map_in_place`/`flat_map_in_place` to reflect their slightly changed signatures. - Although this commit renames the `fold` module as `mut_visit`, it keeps it in the `fold.rs` file, so as not to confuse git. The next commit will rename the file.
2019-02-05 15:20:55 +11:00
impl<T, A: Array<Item = T>> MapInPlace<T> for SmallVec<A> {
fn flat_map_in_place<F, I>(&mut self, mut f: F)
2019-12-22 17:42:04 -05:00
where
F: FnMut(T) -> I,
I: IntoIterator<Item = T>,
{
let mut read_i = 0;
let mut write_i = 0;
unsafe {
let mut old_len = self.len();
self.set_len(0); // make sure we just leak elements in case of panic
while read_i < old_len {
// move the read_i'th item out of the vector and map it
// to an iterator
let e = ptr::read(self.get_unchecked(read_i));
let iter = f(e).into_iter();
read_i += 1;
for e in iter {
if write_i < read_i {
ptr::write(self.get_unchecked_mut(write_i), e);
write_i += 1;
} else {
// If this is reached we ran out of space
// in the middle of the vector.
// However, the vector is in a valid state here,
// so we just do a somewhat inefficient insert.
self.set_len(old_len);
self.insert(write_i, e);
old_len = self.len();
self.set_len(0);
read_i += 1;
write_i += 1;
}
}
}
// write_i tracks the number of actually written new items.
self.set_len(write_i);
}
}
}