Re: [boost] [Boost-users] Attn: We need GSoC 2014 mentors for Boost!

Subject: Re: [boost] [Boost-users] Attn: We need GSoC 2014 mentors for Boost!
From: Hartmut Kaiser (hartmut.kaiser_at_[hidden])
Date: 2014-02-10 13:56:55

> > > Boost.Thread / Work-Stealing-Thread-Pool: This is something already
> > > implemented as part of HPX (https://github.com/STEllAR-GROUP/hpx),
> > > we have several work-stealing implementations, FWIW. Feel free to use.
> >
> > Again thanks. The intent is to have this in Boost.
> 
> Would HPX's implementation be easily portable to Boost do you think?

There are many things in HPX which could be extracted to be part of Boost.
All of it is Boost licensed and very much aligned with Boost coding style
and quality requirements. Thus I wouldn't see any formal problems. Things
which come to mind (in no particular order) are:

- Fully move-enabled and serializable implementations of tuple, any, bind,
protect, function
- Highly efficient binary portable serialization archives (including array
and zero copy optimizations)
- Small utilities such as invoke, void_guard, deferred_call, one_shot_bind,
high_resolution_timer, decay, etc.
- Adaptation of other proposed Boost code like a portable stack-backtrace, a
plugin library, a generic cache library, etc.
- A (user level) thread scheduling subsystem allowing to mix and match
different policies, like different work-stealing mechanisms (or none at
all), lifo/fifo queues, flat/hierarchical schedulers, etc.

At the same time, the parts of HPX have been developed together and might be
interlined and it might not be trivial (albeit not impossible) to rip out
only part of it without pulling in half of other things. 

Regards Hartmut
---------------
http://boost-spirit.com
http://stellar.cct.lsu.edu