I say that having a Lua-based project in openFrameworks for over 10 years: I keep the Lua version up to date with the latest stable release, but make it possible for people to use LuaJIT. It definitely helps with realtime graphics on embedded systems, RPi for example.
On Oct 2, 2024, at 6:13 PM, pd-list-request@lists.iem.at wrote:
Message: 3Date: Wed, 2 Oct 2024 18:13:32 +0200From: Christof Ressi <info@christofressi.com>Subject: [PD] Re: Lua in pdTo: pd-list@lists.iem.atMessage-ID: <b5a3d5d6-e548-43db-8863-8c3bd92812d1@christofressi.com>Content-Type: multipart/alternative; boundary="------------sin7OcxzKLJn12uvlllSAVsz"LuaJIT is awesome, but is it a good idea to run a JIT on the audio thread? Compiling code is certainly not realtime safe and I think we have no control over /when /individual functions are being compiled. So while LuaJIT would give better performance, I assume it would have a larger worst-case latency. One the other hand, LuaJIT also has a faster bytecode interpreter (written in hand-optimized assembly). We could allow people to select the JIT compiler or the bytecode interpreter as a runtime option.Side note: LuaJIT also supports a subset of Lua 5.2.