Re: FFI, cURL and function callback

  • From: Justin Cormack <justin@xxxxxxxxxxxxxxxxxxxxx>
  • To: luajit@xxxxxxxxxxxxx
  • Date: Fri, 29 Mar 2013 12:35:08 +0000

On Thu, Mar 28, 2013 at 2:08 PM, William Adams <william_a_adams@xxxxxxx>wrote:

>
> "Is ljsyscall mature enough"?  I'd say it's at least as mature as any
> other interface that might serve the same purpose.  I think it's a very
> good starting point.  Primary for me has been the conding methodology
> (always have a return value for success as well as the actual value).  It
> also takes the approach of allowing you to pass in buffers where
> appropriate, or allocate if you don't.  If this is not the library, then I
> think it's close enough that with enough focused attention it could become
> THE library.  Again, barring any other library that I don't currently know
> about.
>
>
There is a lot I want to do with ljsyscall, but the interfaces that you
need for sockets etc should not change significantly and incompatibly
(unless people want changes). Mainly adding new stuff now (eg raw socket
support recently) and and there is still stuff to cleanup. I am thinking of
adding BSD/Mac OS support which will mean splitting out Linux specific
stuff, but it will still be raw support with no abstraction layer. Clearly
if you want to support Windows as well in a general library there will be a
fair amount of abstraction to do anyway.

Most of the work in a general library is going to be in the higher level
stuff anyway, and designing the APIs so that different types of stream
compose and substitute easily.

Justin

Other related posts:

  1. Home
  2. luajit
  3. Archive
  4. 03-2013