Hi,

My doctor thesis (sorry, written in Japanese) is work for removing GIL
(we call it GVL).  I concluded that with keeping compatibility with
current MRI, it is hard work to support parallel threads because single
thread execution will be slow down.

I recommend that Ruby should be encouraged that more course-grained
palatalization such as process level and so on. We will try to add this
feature after 2.0.

Thanks,
Koichi

(2012/10/02 0:12), Miguel Palhas wrote:
> Greetings
> 
> I'm currently an MSc student in Portugal, majoring in Parallel
> Computing. However, i'm been a Ruby fan for quite some time now, and
> currently have some Rails projects going on
> 
> At this point i'm about to start my master's thesis, but currently still
> no theme is defined, although there is something (non-related to this)
> being worked on
> 
> It was brought to my attention that ruby has some parallelism problems,
> for example, the existance of the GIL for most current implementations.
> However, it seems this is already being worked on (for Rubinius 2.0, i
> believe)
> This made me think, would there be any work to be done around this area,
> somehow related to this problematic, or other parallelism issues, that
> would be interesting for a master thesis?


-- 
// SASADA Koichi at atdot dot net