From: Christoph Kappel <unexist@...>
Date: 2009-12-05T20:58:05+09:00
Subject: [ruby-core:27066] Re: Ruby causes nearly as much cpu wakeups  as firefox

------=_Part_624_1092948281.1260014278745
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable


Why is there a scheduling required? As far as I understand the thread model=
 of 19 a thread can only access it's own vm
instance and not vms of other threads. Also when there is no other thread h=
olding a vm, why still the polling? Threads
can be awakened on signal receive as well, so dispatching signals can't be =
a reason for this.

I really have decent problems with the odd threading in 19 and I can't find=
 any docs about it. The solution can't be forcing
someone to read the whole sources?

---- On Sat, 05 Dec 2009 02:21:51 +0100 Hongli Lai <hongli@plan99.net> wrot=
e ----=20

 > Christoph Kappel wrote:=20
 > > Thanks for the reply, is there a proper reason for this polling thread=
? Actually I don't like the idea of a thread=20
 > > that wastes resources just to check that nothing is to do and an addit=
ional delay of thread context switches=20
 > > and signal processing.=20
 > =20
 > That thread is responsible for timesliced scheduling of Ruby threads. Ho=
wever in=20
 > single-threaded Ruby apps, this doesn't happen, and when Ruby is doing n=
othing it=20
 > really is doing nothing.=20
 > =20
 > Not sure about 1.9.
 >
------=_Part_624_1092948281.1260014278745--