News for hackers. Stuff that matters.

almk277

Re:Lua support in the NetBSD kernel on Thursday October 17, 2013 @05:01AM (#6564283)
It's just legacy. C as infrastructure is good, C as a language is bad. There's a whole industry and millions of lines of code in visual basic, but I never met somebody who says that VB is good.

Is there a way to say "do this under spinlock"?

Is there a way to handle errors not mixing it with main logic? Not wrapping nearly every line with if/goto?

Is there a way to create a generic collection? No, linux list.h is not that: it requires manual initialization, it does not work with elements, but with their special parts, it's macroses requires logically unneeded information (such as 'member' in list_for_each_entry), and so on. And this is the best you can do in C.

C is too unobvious, too. Google for "wait4 backdoor".

C works, of course. It just requires too much boilerplate code and too much programmer resources.

Re:Lua support in the NetBSD kernel on Thursday October 17, 2013 @03:41AM (#6564100)
It's good somebody even tries to break the old myth about C as an essential kernel part. C is _not_ good for drivers: it lacks most of the appropriate abstractions, just as facilities to create them. Once we switch from asm to C - why not go further?

Speed matters, of course, but I suppose that even script speed is enough for many drivers (and, just in case, we have luaJIT).

Don't sure if lua is a right choice, but it's an interesting experiment - and this is what NetBSD is good for.