Well I think leaf cases are the usual thing when using FFI, unless
you're wrapping some Ruby-aware code which re-enters the interpreter.
I'd vote for an "attach_blocking_function" or something like that.

On 7 ene, 22:10, Wayne Meissner <wmeiss...@gmail.com> wrote:

> Support for this is technically there, however it is not expressable
> using attach_function.
>
> e.g.
>
> lib = DynamicLibrary.open("libc.so", DynamicLibrary.RTLD_LAZY|
> DynamicLibrary.RTLD_GLOBAL)
> addr = lib.find_function("sleep")
>
> # :blocking => true tells FFI the function will block, so it releases the GIL
> fn = Function.new(:int, [ :int ], addr, { :blocking => true })
> module Sleep
>   fn.attach(self, "libc_sleep")
> end
>
> Or something like that.  There is a spec that tests the functionality,
> which you might want to look at in case I got something wrong above.
>
> I'm not sure I want to make this easier.  On CRuby, it is really easy
> to do nasty things if you're not careful with this, e.g. if a blocking
> function calls a ruby callback, very bad things will happen.  So you
> really only want to use it for leaf functions.
>
> The real solution of course, is to upgrade to JRuby - it handles all
> this blocking stuff automagically.
>
> 2010/1/8 rogerdpack <rogerpack2...@gmail.com>:>> several of these calls in parallel native threads) and I just found

> >> out about Ruby's "rb_thread_blocking_region" C call.

>

> > It might be nice if FFI would optionally wrap its calls in an
> > rb_thread_blocking_region (just on 1.9, I guess).
> > -r