3 ms·
> Marc Balmer comes across as some sort of Lua zealot. Lua is a great little language but c'mon. He is like a broken record. Why does he want this so badly? Qu
by vor_ 13y ago
> Marc Balmer comes across as some sort of Lua zealot. Lua is a great little language but c'mon. He is like a broken record. Why does he want this so badly?
Quoting bullet points from his FOSDEM 2013 slides:
• Modifying software written in C is hard for users
• Give users the power to modify and extend the system
• Let users explore the system in an easy way
• "Rapid Application Development" approach to driver/kernel development
• Modifying the system behavior
• Configuration of kernel subsystems
- pppppowerbook 13y agoThose are all terrible reasons to introduce complex code into the kernel. Making kernel development easy won't actually make it easy. The thing that makes it hard isn't C.
- synchronise 13y agoYou can compile NetBSD without kernel Lua support. Literally no reasons for complaint unless this breaks something, which it doesn't.
- voltagex_ 13y agoMaybe this argument is less useful for BSD users, but wouldn't most users be using the stock kernel? So it's up to the whims of the distributors whether to include it or not.
- Sanddancer 13y agoIt's pretty common in the BSD world to build your own kernel. The entire build system, imo, is much more streamlined and well thought out than linux, so relying on the distributor is not needed. Also, one of the missions of NetBSD is to get it working on everything, so as others have stated, being able to poke at the kernel in lua, both for developing drivers, and for pulling state out of the kernel for general debugging, is very useful.
- Touche 13y agoWhat complex code, lua?
- bch 13y agoI don't think these things (Lua zealot, above points) are mutually exclusive. I think this is mostly for prototyping though. I can't imagine running this in a production system. If it fosters exploration (via Lua, or the Lua interface to the kernel), it's a net win (no pun intended).