[burnt-resistor]

Go compiles massive codebases in that time. V can recompile itself probably 2-3x in that time.

I don't take any new language seriously unless it's memory safe, free of UB, able to interoperate with what already exists including optional shared libraries (because static linking the world every time in everything is memory and disk wasteful), and assists formal proofs of correctness. Otherwise, what already exists seems preferable for serious use and hobbies can remain fun distractions.

[binary132]

If a language can’t use shared objects at all, it’s really not much use, is it? Almost all languages make use of at least the posix syscall interfaces provided by the OS, and some platforms don’t even allow you to roll your own syscalls, iirc.

In my eyes it’s more important that FFI be easy, automatic, and as efficient as possible. Go imposes a significant cost on FFI, for example, and many languages have typesystems that are very unfriendly to C ABI or basically require swig.

One thing I really appreciate about Lua is that I can write Lua interfaces for my classes and methods quite easily, and even use Lua as a garbage-collected allocator for native types. Automating the generation of Lua interfaces can easily be done natively with metaprogramming, without involving dependencies like swig. It’s so good that instead of feeling like you’re figuring it out on the Lua side, it’s as if Lua puts the host language first, Lua doesn’t even need to be the owner of the process. It is almost ideal other than the clumsy interface between the stack-like C side and tables, and the dynamic parameter lists.

For so many languages FFI is an afterthought at best.

[38]

troll spotted