Garbage Collection in Python - current status of issue?

[skaller]

Aahz Maruch wrote:
> skaller  <skaller at maxtal.com.au> wrote:
> >
> >Finally, there is an even worse problem with threading, particularly
> >on Linux. Boehm has to stop the world to do garbage collection.  On
> >Solaris, this is no problem, since Solaris threads are grouped within a
> >process. On linux, there are no threads: all threads are processes, and
> >stopping the world has to stop the whole system.
> 
> This is the first time I've heard that claim about threads on Linux; can
> you point me at some resources that discuss the issue in general
> (without assuming that you're in the middle of writing a program)?

Sorry, no. AFAIK, the linux kernel handles all threads of control
the same way. The distinction between a thread and process is
a matter of shared address spaces, which is independent.
In other words, the distinction is higher level than the kernel,
typically handled by user space C libraries. This means Linux
supports a much richer thread/process architecture than, say, Solaris,
where the kernel supplies native threads local to and distinct from
processes, but the Solaris solution is more robust.

Someone more expert in the area should correct me if I'm wrong,
which is possible [my information comes from the time I played with
getting the boehm collector to work with Python, in the days when
pThreads where being phased out and LinuxThreads replacing them --
so my information could be out of date]

-- 
John Skaller, mailto:skaller at maxtal.com.au
1/10 Toxteth Rd Glebe NSW 2037 Australia
homepage: http://www.maxtal.com.au/~skaller
downloads: http://www.triode.net.au/~skaller