3 ms·

The closing keynote at RustConf is generally someone who is new to Rust or has never used Rust but can bring in an interesting perspective. Last year we had Joe

The closing keynote at RustConf is generally someone who is new to Rust or has never used Rust but can bring in an interesting perspective. Last year we had Joe Duffy, who I don't think has done much Rust aside from perhaps basic experimentation, but has a lot of relevant experience from running the Midori (Microsoft's safe systems language) project. (also, the author isn't a "guy")

Shoothe 9y ago
It's interesting because when I read about how Rust can rewrite iterators code into more performant direct loops I thought about Joe's blog posts that mentioned the same efforts for .NET. Also Span<T>. It seems .NET will have similar abstractions like Rust (of course not all can be migrated).
- pygy_ 9y ago
  LuaJIT does the same, down to an absurd level. From the luafun library readme (which uses iterators under the hood): https://github.com/luafun/luafun https://github.com/luafun/luafun ———— > Let's see an example: > -- Functional style > require "fun" () > -- calculate sum(x for x^2 in 1..n) > n = 100 > print(reduce(operator.add, 0, map(function(x) return x^2 end, range(n)))) 328350 > -- Object-oriented style > local fun = require "fun" > -- calculate sum(x for x^2 in 1..n) > print(fun.range(n):map(function(x) return x^2 end):reduce(operator.add, 0)) 328350 > Lua Fun takes full advantage of the innovative tracing JIT compiler to achieve transcendental performance on nested functional expressions. Functional compositions and high-order functions can be translated into efficient machine code. Can you believe it? Just try to run the example above with `luajit -jdump` and see what happens: -- skip some initilization code -- ->LOOP: 0bcaffd0 movaps xmm5, xmm7 0bcaffd3 movaps xmm7, xmm1 0bcaffd6 addsd xmm7, xmm5 0bcaffda ucomisd xmm7, xmm0 0bcaffde jnb 0x0bca0024 ->5 0bcaffe4 movaps xmm5, xmm7 0bcaffe7 mulsd xmm5, xmm5 0bcaffeb addsd xmm6, xmm5 0bcaffef jmp 0x0bcaffd0 ->LOOP ---- TRACE 1 stop -> loop > The functional chain above was translated by LuaJIT to (!) one machine loop containing just 10 CPU assembly instructions without CALL.

Shoothe 9y ago

It's interesting because when I read about how Rust can rewrite iterators code into more performant direct loops I thought about Joe's blog posts that mentioned the same efforts for .NET. Also Span<T>. It seems .NET will have similar abstractions like Rust (of course not all can be migrated).

pygy_ 9y ago
LuaJIT does the same, down to an absurd level. From the luafun library readme (which uses iterators under the hood): https://github.com/luafun/luafun https://github.com/luafun/luafun ———— > Let's see an example: > -- Functional style > require "fun" () > -- calculate sum(x for x^2 in 1..n) > n = 100 > print(reduce(operator.add, 0, map(function(x) return x^2 end, range(n)))) 328350 > -- Object-oriented style > local fun = require "fun" > -- calculate sum(x for x^2 in 1..n) > print(fun.range(n):map(function(x) return x^2 end):reduce(operator.add, 0)) 328350 > Lua Fun takes full advantage of the innovative tracing JIT compiler to achieve transcendental performance on nested functional expressions. Functional compositions and high-order functions can be translated into efficient machine code. Can you believe it? Just try to run the example above with `luajit -jdump` and see what happens: -- skip some initilization code -- ->LOOP: 0bcaffd0 movaps xmm5, xmm7 0bcaffd3 movaps xmm7, xmm1 0bcaffd6 addsd xmm7, xmm5 0bcaffda ucomisd xmm7, xmm0 0bcaffde jnb 0x0bca0024 ->5 0bcaffe4 movaps xmm5, xmm7 0bcaffe7 mulsd xmm5, xmm5 0bcaffeb addsd xmm6, xmm5 0bcaffef jmp 0x0bcaffd0 ->LOOP ---- TRACE 1 stop -> loop > The functional chain above was translated by LuaJIT to (!) one machine loop containing just 10 CPU assembly instructions without CALL.