Re: Turning Lua into C++ (was: alleviate the load of the GC)

  • From: Theo Schlossnagle <jesus@xxxxxxxxxx>
  • To: luajit@xxxxxxxxxxxxx
  • Date: Wed, 2 Sep 2015 11:19:43 -0400

It's important to us for LuaJIT to remain focused on performance and
stability. We have had issues with the GC in the past and been moderately
successful moving specific heavy tasks into FFI managed heap... It would be
nice if there was a toolkit for doing that so we all didn't reinvent the
wheel... I see that toolkit not changing LuaJIT from a features or language
PoV.

Improving the architecture support (and mapping to new instructions as new
chips arrive), maintaining strong stability guarantees and making
improvements (or overhauling) the GC. The last is, however, an epic task.

LuaJIT meets our current needs well now, I'm not interested in it morphing
into something that no longer meets our needs.

On Wed, Sep 2, 2015 at 11:13 AM, David Edelsohn <dje.gcc@xxxxxxxxx> wrote:

On Wed, Sep 2, 2015 at 11:11 AM, Alex Szpakowski <aszpakowski@xxxxxxxxx>
wrote:
it seems some people are seeing this transition simply as an opportunity
to shape an actively used fork of Lua to suit their own whims in lieu of
forking or changing PUC Lua, without much regard to the actual
(well-established) role of LuaJIT. That line of thinking is actively
harmful to LuaJIT’s health, in my opinion.

+1

- David




--

Theo Schlossnagle

http://omniti.com/is/theo-schlossnagle

Other related posts: