mlua/src/value.rs

173 lines
5.2 KiB
Rust
Raw Normal View History

A lot of performance changes. Okay, so this is kind of a mega-commit of a lot of performance related changes to rlua, some of which are pretty complicated. There are some small improvements here and there, but most of the benefits of this change are from a few big changes. The simplest big change is that there is now `protect_lua` as well as `protect_lua_call`, which allows skipping a lightuserdata parameter and some stack manipulation in some cases. Second simplest is the change to use Vec instead of VecDeque for MultiValue, and to have MultiValue be used as a sort of "backwards-only" Vec so that ToLuaMulti / FromLuaMulti still work correctly. The most complex change, though, is a change to the way LuaRef works, so that LuaRef can optionally point into the Lua stack instead of only registry values. At state creation a set number of stack slots is reserved for the first N LuaRef types (currently 16), and space for these are also allocated separately allocated at callback time. There is a huge breaking change here, which is that now any LuaRef types MUST only be used with the Lua on which they were created, and CANNOT be used with any other Lua callback instance. This mostly will affect people using LuaRef types from inside a scope callback, but hopefully in those cases `Function::bind` will be a suitable replacement. On the plus side, the rules for LuaRef types are easier to state now. There is probably more easy-ish perf on the table here, but here's the preliminary results, based on my very limited benchmarks: create table time: [314.13 ns 315.71 ns 317.44 ns] change: [-36.154% -35.670% -35.205%] (p = 0.00 < 0.05) create array 10 time: [2.9731 us 2.9816 us 2.9901 us] change: [-16.996% -16.600% -16.196%] (p = 0.00 < 0.05) Performance has improved. create string table 10 time: [5.6904 us 5.7164 us 5.7411 us] change: [-53.536% -53.309% -53.079%] (p = 0.00 < 0.05) Performance has improved. call add function 3 10 time: [5.1134 us 5.1222 us 5.1320 us] change: [-4.1095% -3.6910% -3.1781%] (p = 0.00 < 0.05) Performance has improved. call callback add 2 10 time: [5.4408 us 5.4480 us 5.4560 us] change: [-6.4203% -5.7780% -5.0013%] (p = 0.00 < 0.05) Performance has improved. call callback append 10 time: [9.8243 us 9.8410 us 9.8586 us] change: [-26.937% -26.702% -26.469%] (p = 0.00 < 0.05) Performance has improved. create registry 10 time: [3.7005 us 3.7089 us 3.7174 us] change: [-8.4965% -8.1042% -7.6926%] (p = 0.00 < 0.05) Performance has improved. I think that a lot of these benchmarks are too "easy", and most API usage is going to be more like the 'create string table 10' benchmark, where there are a lot of handles and tables and strings, so I think that 25%-50% improvement is a good guess for most use cases.
2018-03-11 23:20:10 -04:00
use std::{slice, str, vec};
use std::iter::{self, FromIterator};
A lot of performance changes. Okay, so this is kind of a mega-commit of a lot of performance related changes to rlua, some of which are pretty complicated. There are some small improvements here and there, but most of the benefits of this change are from a few big changes. The simplest big change is that there is now `protect_lua` as well as `protect_lua_call`, which allows skipping a lightuserdata parameter and some stack manipulation in some cases. Second simplest is the change to use Vec instead of VecDeque for MultiValue, and to have MultiValue be used as a sort of "backwards-only" Vec so that ToLuaMulti / FromLuaMulti still work correctly. The most complex change, though, is a change to the way LuaRef works, so that LuaRef can optionally point into the Lua stack instead of only registry values. At state creation a set number of stack slots is reserved for the first N LuaRef types (currently 16), and space for these are also allocated separately allocated at callback time. There is a huge breaking change here, which is that now any LuaRef types MUST only be used with the Lua on which they were created, and CANNOT be used with any other Lua callback instance. This mostly will affect people using LuaRef types from inside a scope callback, but hopefully in those cases `Function::bind` will be a suitable replacement. On the plus side, the rules for LuaRef types are easier to state now. There is probably more easy-ish perf on the table here, but here's the preliminary results, based on my very limited benchmarks: create table time: [314.13 ns 315.71 ns 317.44 ns] change: [-36.154% -35.670% -35.205%] (p = 0.00 < 0.05) create array 10 time: [2.9731 us 2.9816 us 2.9901 us] change: [-16.996% -16.600% -16.196%] (p = 0.00 < 0.05) Performance has improved. create string table 10 time: [5.6904 us 5.7164 us 5.7411 us] change: [-53.536% -53.309% -53.079%] (p = 0.00 < 0.05) Performance has improved. call add function 3 10 time: [5.1134 us 5.1222 us 5.1320 us] change: [-4.1095% -3.6910% -3.1781%] (p = 0.00 < 0.05) Performance has improved. call callback add 2 10 time: [5.4408 us 5.4480 us 5.4560 us] change: [-6.4203% -5.7780% -5.0013%] (p = 0.00 < 0.05) Performance has improved. call callback append 10 time: [9.8243 us 9.8410 us 9.8586 us] change: [-26.937% -26.702% -26.469%] (p = 0.00 < 0.05) Performance has improved. create registry 10 time: [3.7005 us 3.7089 us 3.7174 us] change: [-8.4965% -8.1042% -7.6926%] (p = 0.00 < 0.05) Performance has improved. I think that a lot of these benchmarks are too "easy", and most API usage is going to be more like the 'create string table 10' benchmark, where there are a lot of handles and tables and strings, so I think that 25%-50% improvement is a good guess for most use cases.
2018-03-11 23:20:10 -04:00
use error::{Error, Result};
use types::{Integer, LightUserData, Number};
use string::String;
use table::Table;
use function::Function;
use thread::Thread;
use userdata::AnyUserData;
use lua::Lua;
/// A dynamically typed Lua value.
#[derive(Debug, Clone)]
pub enum Value<'lua> {
/// The Lua value `nil`.
Nil,
/// The Lua value `true` or `false`.
Boolean(bool),
/// A "light userdata" object, equivalent to a raw pointer.
LightUserData(LightUserData),
/// An integer number.
///
/// Any Lua number convertible to a `Integer` will be represented as this variant.
Integer(Integer),
/// A floating point number.
Number(Number),
/// An interned string, managed by Lua.
///
/// Unlike Rust strings, Lua strings may not be valid UTF-8.
String(String<'lua>),
/// Reference to a Lua table.
Table(Table<'lua>),
/// Reference to a Lua function (or closure).
Function(Function<'lua>),
/// Reference to a Lua thread (or coroutine).
Thread(Thread<'lua>),
/// Reference to a userdata object that holds a custom type which implements `UserData`.
/// Special builtin userdata types will be represented as other `Value` variants.
UserData(AnyUserData<'lua>),
/// `Error` is a special builtin userdata type. When received from Lua it is implicitly cloned.
Error(Error),
}
pub use self::Value::Nil;
impl<'lua> Value<'lua> {
pub(crate) fn type_name(&self) -> &'static str {
match *self {
Value::Nil => "nil",
Value::Boolean(_) => "boolean",
Value::LightUserData(_) => "light userdata",
Value::Integer(_) => "integer",
Value::Number(_) => "number",
Value::String(_) => "string",
Value::Table(_) => "table",
Value::Function(_) => "function",
Value::Thread(_) => "thread",
Value::UserData(_) | Value::Error(_) => "userdata",
}
}
}
/// Trait for types convertible to `Value`.
pub trait ToLua<'lua> {
/// Performs the conversion.
fn to_lua(self, lua: &'lua Lua) -> Result<Value<'lua>>;
}
/// Trait for types convertible from `Value`.
pub trait FromLua<'lua>: Sized {
/// Performs the conversion.
fn from_lua(lua_value: Value<'lua>, lua: &'lua Lua) -> Result<Self>;
}
/// Multiple Lua values used for both argument passing and also for multiple return values.
#[derive(Debug, Clone)]
A lot of performance changes. Okay, so this is kind of a mega-commit of a lot of performance related changes to rlua, some of which are pretty complicated. There are some small improvements here and there, but most of the benefits of this change are from a few big changes. The simplest big change is that there is now `protect_lua` as well as `protect_lua_call`, which allows skipping a lightuserdata parameter and some stack manipulation in some cases. Second simplest is the change to use Vec instead of VecDeque for MultiValue, and to have MultiValue be used as a sort of "backwards-only" Vec so that ToLuaMulti / FromLuaMulti still work correctly. The most complex change, though, is a change to the way LuaRef works, so that LuaRef can optionally point into the Lua stack instead of only registry values. At state creation a set number of stack slots is reserved for the first N LuaRef types (currently 16), and space for these are also allocated separately allocated at callback time. There is a huge breaking change here, which is that now any LuaRef types MUST only be used with the Lua on which they were created, and CANNOT be used with any other Lua callback instance. This mostly will affect people using LuaRef types from inside a scope callback, but hopefully in those cases `Function::bind` will be a suitable replacement. On the plus side, the rules for LuaRef types are easier to state now. There is probably more easy-ish perf on the table here, but here's the preliminary results, based on my very limited benchmarks: create table time: [314.13 ns 315.71 ns 317.44 ns] change: [-36.154% -35.670% -35.205%] (p = 0.00 < 0.05) create array 10 time: [2.9731 us 2.9816 us 2.9901 us] change: [-16.996% -16.600% -16.196%] (p = 0.00 < 0.05) Performance has improved. create string table 10 time: [5.6904 us 5.7164 us 5.7411 us] change: [-53.536% -53.309% -53.079%] (p = 0.00 < 0.05) Performance has improved. call add function 3 10 time: [5.1134 us 5.1222 us 5.1320 us] change: [-4.1095% -3.6910% -3.1781%] (p = 0.00 < 0.05) Performance has improved. call callback add 2 10 time: [5.4408 us 5.4480 us 5.4560 us] change: [-6.4203% -5.7780% -5.0013%] (p = 0.00 < 0.05) Performance has improved. call callback append 10 time: [9.8243 us 9.8410 us 9.8586 us] change: [-26.937% -26.702% -26.469%] (p = 0.00 < 0.05) Performance has improved. create registry 10 time: [3.7005 us 3.7089 us 3.7174 us] change: [-8.4965% -8.1042% -7.6926%] (p = 0.00 < 0.05) Performance has improved. I think that a lot of these benchmarks are too "easy", and most API usage is going to be more like the 'create string table 10' benchmark, where there are a lot of handles and tables and strings, so I think that 25%-50% improvement is a good guess for most use cases.
2018-03-11 23:20:10 -04:00
pub struct MultiValue<'lua>(Vec<Value<'lua>>);
impl<'lua> MultiValue<'lua> {
/// Creates an empty `MultiValue` containing no values.
pub fn new() -> MultiValue<'lua> {
A lot of performance changes. Okay, so this is kind of a mega-commit of a lot of performance related changes to rlua, some of which are pretty complicated. There are some small improvements here and there, but most of the benefits of this change are from a few big changes. The simplest big change is that there is now `protect_lua` as well as `protect_lua_call`, which allows skipping a lightuserdata parameter and some stack manipulation in some cases. Second simplest is the change to use Vec instead of VecDeque for MultiValue, and to have MultiValue be used as a sort of "backwards-only" Vec so that ToLuaMulti / FromLuaMulti still work correctly. The most complex change, though, is a change to the way LuaRef works, so that LuaRef can optionally point into the Lua stack instead of only registry values. At state creation a set number of stack slots is reserved for the first N LuaRef types (currently 16), and space for these are also allocated separately allocated at callback time. There is a huge breaking change here, which is that now any LuaRef types MUST only be used with the Lua on which they were created, and CANNOT be used with any other Lua callback instance. This mostly will affect people using LuaRef types from inside a scope callback, but hopefully in those cases `Function::bind` will be a suitable replacement. On the plus side, the rules for LuaRef types are easier to state now. There is probably more easy-ish perf on the table here, but here's the preliminary results, based on my very limited benchmarks: create table time: [314.13 ns 315.71 ns 317.44 ns] change: [-36.154% -35.670% -35.205%] (p = 0.00 < 0.05) create array 10 time: [2.9731 us 2.9816 us 2.9901 us] change: [-16.996% -16.600% -16.196%] (p = 0.00 < 0.05) Performance has improved. create string table 10 time: [5.6904 us 5.7164 us 5.7411 us] change: [-53.536% -53.309% -53.079%] (p = 0.00 < 0.05) Performance has improved. call add function 3 10 time: [5.1134 us 5.1222 us 5.1320 us] change: [-4.1095% -3.6910% -3.1781%] (p = 0.00 < 0.05) Performance has improved. call callback add 2 10 time: [5.4408 us 5.4480 us 5.4560 us] change: [-6.4203% -5.7780% -5.0013%] (p = 0.00 < 0.05) Performance has improved. call callback append 10 time: [9.8243 us 9.8410 us 9.8586 us] change: [-26.937% -26.702% -26.469%] (p = 0.00 < 0.05) Performance has improved. create registry 10 time: [3.7005 us 3.7089 us 3.7174 us] change: [-8.4965% -8.1042% -7.6926%] (p = 0.00 < 0.05) Performance has improved. I think that a lot of these benchmarks are too "easy", and most API usage is going to be more like the 'create string table 10' benchmark, where there are a lot of handles and tables and strings, so I think that 25%-50% improvement is a good guess for most use cases.
2018-03-11 23:20:10 -04:00
MultiValue(Vec::new())
}
}
impl<'lua> FromIterator<Value<'lua>> for MultiValue<'lua> {
fn from_iter<I: IntoIterator<Item = Value<'lua>>>(iter: I) -> Self {
A lot of performance changes. Okay, so this is kind of a mega-commit of a lot of performance related changes to rlua, some of which are pretty complicated. There are some small improvements here and there, but most of the benefits of this change are from a few big changes. The simplest big change is that there is now `protect_lua` as well as `protect_lua_call`, which allows skipping a lightuserdata parameter and some stack manipulation in some cases. Second simplest is the change to use Vec instead of VecDeque for MultiValue, and to have MultiValue be used as a sort of "backwards-only" Vec so that ToLuaMulti / FromLuaMulti still work correctly. The most complex change, though, is a change to the way LuaRef works, so that LuaRef can optionally point into the Lua stack instead of only registry values. At state creation a set number of stack slots is reserved for the first N LuaRef types (currently 16), and space for these are also allocated separately allocated at callback time. There is a huge breaking change here, which is that now any LuaRef types MUST only be used with the Lua on which they were created, and CANNOT be used with any other Lua callback instance. This mostly will affect people using LuaRef types from inside a scope callback, but hopefully in those cases `Function::bind` will be a suitable replacement. On the plus side, the rules for LuaRef types are easier to state now. There is probably more easy-ish perf on the table here, but here's the preliminary results, based on my very limited benchmarks: create table time: [314.13 ns 315.71 ns 317.44 ns] change: [-36.154% -35.670% -35.205%] (p = 0.00 < 0.05) create array 10 time: [2.9731 us 2.9816 us 2.9901 us] change: [-16.996% -16.600% -16.196%] (p = 0.00 < 0.05) Performance has improved. create string table 10 time: [5.6904 us 5.7164 us 5.7411 us] change: [-53.536% -53.309% -53.079%] (p = 0.00 < 0.05) Performance has improved. call add function 3 10 time: [5.1134 us 5.1222 us 5.1320 us] change: [-4.1095% -3.6910% -3.1781%] (p = 0.00 < 0.05) Performance has improved. call callback add 2 10 time: [5.4408 us 5.4480 us 5.4560 us] change: [-6.4203% -5.7780% -5.0013%] (p = 0.00 < 0.05) Performance has improved. call callback append 10 time: [9.8243 us 9.8410 us 9.8586 us] change: [-26.937% -26.702% -26.469%] (p = 0.00 < 0.05) Performance has improved. create registry 10 time: [3.7005 us 3.7089 us 3.7174 us] change: [-8.4965% -8.1042% -7.6926%] (p = 0.00 < 0.05) Performance has improved. I think that a lot of these benchmarks are too "easy", and most API usage is going to be more like the 'create string table 10' benchmark, where there are a lot of handles and tables and strings, so I think that 25%-50% improvement is a good guess for most use cases.
2018-03-11 23:20:10 -04:00
MultiValue::from_vec(Vec::from_iter(iter))
}
}
impl<'lua> IntoIterator for MultiValue<'lua> {
type Item = Value<'lua>;
A lot of performance changes. Okay, so this is kind of a mega-commit of a lot of performance related changes to rlua, some of which are pretty complicated. There are some small improvements here and there, but most of the benefits of this change are from a few big changes. The simplest big change is that there is now `protect_lua` as well as `protect_lua_call`, which allows skipping a lightuserdata parameter and some stack manipulation in some cases. Second simplest is the change to use Vec instead of VecDeque for MultiValue, and to have MultiValue be used as a sort of "backwards-only" Vec so that ToLuaMulti / FromLuaMulti still work correctly. The most complex change, though, is a change to the way LuaRef works, so that LuaRef can optionally point into the Lua stack instead of only registry values. At state creation a set number of stack slots is reserved for the first N LuaRef types (currently 16), and space for these are also allocated separately allocated at callback time. There is a huge breaking change here, which is that now any LuaRef types MUST only be used with the Lua on which they were created, and CANNOT be used with any other Lua callback instance. This mostly will affect people using LuaRef types from inside a scope callback, but hopefully in those cases `Function::bind` will be a suitable replacement. On the plus side, the rules for LuaRef types are easier to state now. There is probably more easy-ish perf on the table here, but here's the preliminary results, based on my very limited benchmarks: create table time: [314.13 ns 315.71 ns 317.44 ns] change: [-36.154% -35.670% -35.205%] (p = 0.00 < 0.05) create array 10 time: [2.9731 us 2.9816 us 2.9901 us] change: [-16.996% -16.600% -16.196%] (p = 0.00 < 0.05) Performance has improved. create string table 10 time: [5.6904 us 5.7164 us 5.7411 us] change: [-53.536% -53.309% -53.079%] (p = 0.00 < 0.05) Performance has improved. call add function 3 10 time: [5.1134 us 5.1222 us 5.1320 us] change: [-4.1095% -3.6910% -3.1781%] (p = 0.00 < 0.05) Performance has improved. call callback add 2 10 time: [5.4408 us 5.4480 us 5.4560 us] change: [-6.4203% -5.7780% -5.0013%] (p = 0.00 < 0.05) Performance has improved. call callback append 10 time: [9.8243 us 9.8410 us 9.8586 us] change: [-26.937% -26.702% -26.469%] (p = 0.00 < 0.05) Performance has improved. create registry 10 time: [3.7005 us 3.7089 us 3.7174 us] change: [-8.4965% -8.1042% -7.6926%] (p = 0.00 < 0.05) Performance has improved. I think that a lot of these benchmarks are too "easy", and most API usage is going to be more like the 'create string table 10' benchmark, where there are a lot of handles and tables and strings, so I think that 25%-50% improvement is a good guess for most use cases.
2018-03-11 23:20:10 -04:00
type IntoIter = iter::Rev<vec::IntoIter<Value<'lua>>>;
fn into_iter(self) -> Self::IntoIter {
A lot of performance changes. Okay, so this is kind of a mega-commit of a lot of performance related changes to rlua, some of which are pretty complicated. There are some small improvements here and there, but most of the benefits of this change are from a few big changes. The simplest big change is that there is now `protect_lua` as well as `protect_lua_call`, which allows skipping a lightuserdata parameter and some stack manipulation in some cases. Second simplest is the change to use Vec instead of VecDeque for MultiValue, and to have MultiValue be used as a sort of "backwards-only" Vec so that ToLuaMulti / FromLuaMulti still work correctly. The most complex change, though, is a change to the way LuaRef works, so that LuaRef can optionally point into the Lua stack instead of only registry values. At state creation a set number of stack slots is reserved for the first N LuaRef types (currently 16), and space for these are also allocated separately allocated at callback time. There is a huge breaking change here, which is that now any LuaRef types MUST only be used with the Lua on which they were created, and CANNOT be used with any other Lua callback instance. This mostly will affect people using LuaRef types from inside a scope callback, but hopefully in those cases `Function::bind` will be a suitable replacement. On the plus side, the rules for LuaRef types are easier to state now. There is probably more easy-ish perf on the table here, but here's the preliminary results, based on my very limited benchmarks: create table time: [314.13 ns 315.71 ns 317.44 ns] change: [-36.154% -35.670% -35.205%] (p = 0.00 < 0.05) create array 10 time: [2.9731 us 2.9816 us 2.9901 us] change: [-16.996% -16.600% -16.196%] (p = 0.00 < 0.05) Performance has improved. create string table 10 time: [5.6904 us 5.7164 us 5.7411 us] change: [-53.536% -53.309% -53.079%] (p = 0.00 < 0.05) Performance has improved. call add function 3 10 time: [5.1134 us 5.1222 us 5.1320 us] change: [-4.1095% -3.6910% -3.1781%] (p = 0.00 < 0.05) Performance has improved. call callback add 2 10 time: [5.4408 us 5.4480 us 5.4560 us] change: [-6.4203% -5.7780% -5.0013%] (p = 0.00 < 0.05) Performance has improved. call callback append 10 time: [9.8243 us 9.8410 us 9.8586 us] change: [-26.937% -26.702% -26.469%] (p = 0.00 < 0.05) Performance has improved. create registry 10 time: [3.7005 us 3.7089 us 3.7174 us] change: [-8.4965% -8.1042% -7.6926%] (p = 0.00 < 0.05) Performance has improved. I think that a lot of these benchmarks are too "easy", and most API usage is going to be more like the 'create string table 10' benchmark, where there are a lot of handles and tables and strings, so I think that 25%-50% improvement is a good guess for most use cases.
2018-03-11 23:20:10 -04:00
self.0.into_iter().rev()
}
}
A lot of performance changes. Okay, so this is kind of a mega-commit of a lot of performance related changes to rlua, some of which are pretty complicated. There are some small improvements here and there, but most of the benefits of this change are from a few big changes. The simplest big change is that there is now `protect_lua` as well as `protect_lua_call`, which allows skipping a lightuserdata parameter and some stack manipulation in some cases. Second simplest is the change to use Vec instead of VecDeque for MultiValue, and to have MultiValue be used as a sort of "backwards-only" Vec so that ToLuaMulti / FromLuaMulti still work correctly. The most complex change, though, is a change to the way LuaRef works, so that LuaRef can optionally point into the Lua stack instead of only registry values. At state creation a set number of stack slots is reserved for the first N LuaRef types (currently 16), and space for these are also allocated separately allocated at callback time. There is a huge breaking change here, which is that now any LuaRef types MUST only be used with the Lua on which they were created, and CANNOT be used with any other Lua callback instance. This mostly will affect people using LuaRef types from inside a scope callback, but hopefully in those cases `Function::bind` will be a suitable replacement. On the plus side, the rules for LuaRef types are easier to state now. There is probably more easy-ish perf on the table here, but here's the preliminary results, based on my very limited benchmarks: create table time: [314.13 ns 315.71 ns 317.44 ns] change: [-36.154% -35.670% -35.205%] (p = 0.00 < 0.05) create array 10 time: [2.9731 us 2.9816 us 2.9901 us] change: [-16.996% -16.600% -16.196%] (p = 0.00 < 0.05) Performance has improved. create string table 10 time: [5.6904 us 5.7164 us 5.7411 us] change: [-53.536% -53.309% -53.079%] (p = 0.00 < 0.05) Performance has improved. call add function 3 10 time: [5.1134 us 5.1222 us 5.1320 us] change: [-4.1095% -3.6910% -3.1781%] (p = 0.00 < 0.05) Performance has improved. call callback add 2 10 time: [5.4408 us 5.4480 us 5.4560 us] change: [-6.4203% -5.7780% -5.0013%] (p = 0.00 < 0.05) Performance has improved. call callback append 10 time: [9.8243 us 9.8410 us 9.8586 us] change: [-26.937% -26.702% -26.469%] (p = 0.00 < 0.05) Performance has improved. create registry 10 time: [3.7005 us 3.7089 us 3.7174 us] change: [-8.4965% -8.1042% -7.6926%] (p = 0.00 < 0.05) Performance has improved. I think that a lot of these benchmarks are too "easy", and most API usage is going to be more like the 'create string table 10' benchmark, where there are a lot of handles and tables and strings, so I think that 25%-50% improvement is a good guess for most use cases.
2018-03-11 23:20:10 -04:00
impl<'a, 'lua> IntoIterator for &'a MultiValue<'lua> {
type Item = &'a Value<'lua>;
type IntoIter = iter::Rev<slice::Iter<'a, Value<'lua>>>;
A lot of performance changes. Okay, so this is kind of a mega-commit of a lot of performance related changes to rlua, some of which are pretty complicated. There are some small improvements here and there, but most of the benefits of this change are from a few big changes. The simplest big change is that there is now `protect_lua` as well as `protect_lua_call`, which allows skipping a lightuserdata parameter and some stack manipulation in some cases. Second simplest is the change to use Vec instead of VecDeque for MultiValue, and to have MultiValue be used as a sort of "backwards-only" Vec so that ToLuaMulti / FromLuaMulti still work correctly. The most complex change, though, is a change to the way LuaRef works, so that LuaRef can optionally point into the Lua stack instead of only registry values. At state creation a set number of stack slots is reserved for the first N LuaRef types (currently 16), and space for these are also allocated separately allocated at callback time. There is a huge breaking change here, which is that now any LuaRef types MUST only be used with the Lua on which they were created, and CANNOT be used with any other Lua callback instance. This mostly will affect people using LuaRef types from inside a scope callback, but hopefully in those cases `Function::bind` will be a suitable replacement. On the plus side, the rules for LuaRef types are easier to state now. There is probably more easy-ish perf on the table here, but here's the preliminary results, based on my very limited benchmarks: create table time: [314.13 ns 315.71 ns 317.44 ns] change: [-36.154% -35.670% -35.205%] (p = 0.00 < 0.05) create array 10 time: [2.9731 us 2.9816 us 2.9901 us] change: [-16.996% -16.600% -16.196%] (p = 0.00 < 0.05) Performance has improved. create string table 10 time: [5.6904 us 5.7164 us 5.7411 us] change: [-53.536% -53.309% -53.079%] (p = 0.00 < 0.05) Performance has improved. call add function 3 10 time: [5.1134 us 5.1222 us 5.1320 us] change: [-4.1095% -3.6910% -3.1781%] (p = 0.00 < 0.05) Performance has improved. call callback add 2 10 time: [5.4408 us 5.4480 us 5.4560 us] change: [-6.4203% -5.7780% -5.0013%] (p = 0.00 < 0.05) Performance has improved. call callback append 10 time: [9.8243 us 9.8410 us 9.8586 us] change: [-26.937% -26.702% -26.469%] (p = 0.00 < 0.05) Performance has improved. create registry 10 time: [3.7005 us 3.7089 us 3.7174 us] change: [-8.4965% -8.1042% -7.6926%] (p = 0.00 < 0.05) Performance has improved. I think that a lot of these benchmarks are too "easy", and most API usage is going to be more like the 'create string table 10' benchmark, where there are a lot of handles and tables and strings, so I think that 25%-50% improvement is a good guess for most use cases.
2018-03-11 23:20:10 -04:00
fn into_iter(self) -> Self::IntoIter {
(&self.0).into_iter().rev()
}
}
A lot of performance changes. Okay, so this is kind of a mega-commit of a lot of performance related changes to rlua, some of which are pretty complicated. There are some small improvements here and there, but most of the benefits of this change are from a few big changes. The simplest big change is that there is now `protect_lua` as well as `protect_lua_call`, which allows skipping a lightuserdata parameter and some stack manipulation in some cases. Second simplest is the change to use Vec instead of VecDeque for MultiValue, and to have MultiValue be used as a sort of "backwards-only" Vec so that ToLuaMulti / FromLuaMulti still work correctly. The most complex change, though, is a change to the way LuaRef works, so that LuaRef can optionally point into the Lua stack instead of only registry values. At state creation a set number of stack slots is reserved for the first N LuaRef types (currently 16), and space for these are also allocated separately allocated at callback time. There is a huge breaking change here, which is that now any LuaRef types MUST only be used with the Lua on which they were created, and CANNOT be used with any other Lua callback instance. This mostly will affect people using LuaRef types from inside a scope callback, but hopefully in those cases `Function::bind` will be a suitable replacement. On the plus side, the rules for LuaRef types are easier to state now. There is probably more easy-ish perf on the table here, but here's the preliminary results, based on my very limited benchmarks: create table time: [314.13 ns 315.71 ns 317.44 ns] change: [-36.154% -35.670% -35.205%] (p = 0.00 < 0.05) create array 10 time: [2.9731 us 2.9816 us 2.9901 us] change: [-16.996% -16.600% -16.196%] (p = 0.00 < 0.05) Performance has improved. create string table 10 time: [5.6904 us 5.7164 us 5.7411 us] change: [-53.536% -53.309% -53.079%] (p = 0.00 < 0.05) Performance has improved. call add function 3 10 time: [5.1134 us 5.1222 us 5.1320 us] change: [-4.1095% -3.6910% -3.1781%] (p = 0.00 < 0.05) Performance has improved. call callback add 2 10 time: [5.4408 us 5.4480 us 5.4560 us] change: [-6.4203% -5.7780% -5.0013%] (p = 0.00 < 0.05) Performance has improved. call callback append 10 time: [9.8243 us 9.8410 us 9.8586 us] change: [-26.937% -26.702% -26.469%] (p = 0.00 < 0.05) Performance has improved. create registry 10 time: [3.7005 us 3.7089 us 3.7174 us] change: [-8.4965% -8.1042% -7.6926%] (p = 0.00 < 0.05) Performance has improved. I think that a lot of these benchmarks are too "easy", and most API usage is going to be more like the 'create string table 10' benchmark, where there are a lot of handles and tables and strings, so I think that 25%-50% improvement is a good guess for most use cases.
2018-03-11 23:20:10 -04:00
impl<'lua> MultiValue<'lua> {
pub fn from_vec(mut v: Vec<Value<'lua>>) -> MultiValue<'lua> {
v.reverse();
MultiValue(v)
}
pub fn into_vec(self) -> Vec<Value<'lua>> {
let mut v = self.0;
v.reverse();
v
}
pub fn from_vec_rev(v: Vec<Value<'lua>>) -> MultiValue<'lua> {
MultiValue(v)
}
pub fn into_vec_rev(self) -> Vec<Value<'lua>> {
self.0
}
pub fn reserve(&mut self, size: usize) {
self.0.reserve(size);
}
pub fn push_front(&mut self, value: Value<'lua>) {
self.0.push(value);
}
pub fn pop_front(&mut self) -> Option<Value<'lua>> {
self.0.pop()
}
pub fn len(&self) -> usize {
self.0.len()
}
pub fn iter(&self) -> iter::Rev<slice::Iter<Value<'lua>>> {
self.0.iter().rev()
}
}
/// Trait for types convertible to any number of Lua values.
///
/// This is a generalization of `ToLua`, allowing any number of resulting Lua values instead of just
/// one. Any type that implements `ToLua` will automatically implement this trait.
pub trait ToLuaMulti<'lua> {
/// Performs the conversion.
fn to_lua_multi(self, lua: &'lua Lua) -> Result<MultiValue<'lua>>;
}
/// Trait for types that can be created from an arbitrary number of Lua values.
///
/// This is a generalization of `FromLua`, allowing an arbitrary number of Lua values to participate
/// in the conversion. Any type that implements `FromLua` will automatically implement this trait.
pub trait FromLuaMulti<'lua>: Sized {
/// Performs the conversion.
///
/// In case `values` contains more values than needed to perform the conversion, the excess
/// values should be ignored. This reflects the semantics of Lua when calling a function or
/// assigning values. Similarly, if not enough values are given, conversions should assume that
/// any missing values are nil.
fn from_lua_multi(values: MultiValue<'lua>, lua: &'lua Lua) -> Result<Self>;
}