Re: [ANN] opeth 0.0-1

[Satoru Kawahara <s1311350@...>]

I read the bytecode on spring, last year and found that there is room for optimizations.

This is becaus I tried to optimize the BYTECODE.

And I don't think the importance of the consisitency.

On Sat, Jan 14, 2017 at 12:52 AM, Xianfu Pan <pxfgod@gmail.com> wrote:

why not introduce those features directly to LUA5.3? I mean a feature that lua reads the `source-code.lua` and compiles it to `source-code.luac`, and outputs the `optimized-source-code.lua` meanwhile. (after dead-code elimination and constant folding ...). 

The most importance is to keep the **consistency of debug information**  between `optimized-source-code.lua`and `source-code.luac` as well as equivalence.

In other words, an optimizer of lua source-code.(after opt, I compile the code with official lua5.3; it runs faster and it'll keeps the consistency naturally.)

Are you considering about this?   Kawahara sann.

2017-01-13 21:46 GMT+08:00 Satoru Kawahara <s1311350@coins.tsukuba.ac.jp>:

This is great! I have been wanting to get function inlining in Ravi
and maybe I can port your implementation. I want it to be part of the
compiler so I will need to convert your implementation to C.

Please let me know if you have any documentation on how this works.

Thank you, but it hasn't been documented yet.

--

Kawahara Satoru

College of Information Science

University of Tsukuba

--

Kawahara Satoru

College of Information Science

University of Tsukuba