Re: OT: JBoss & Linux Threads
On Sun, Dec 16, 2001 at 18:25:28 +0100, Holger Rauch wrote:
> On Sun, 16 Dec 2001, J.H.M. Dassen (Ray) wrote:
> > Some people's definition of that seems to include hybrid user and kernel
> > space threading, and in that case, the statement is correct.
>
> Is this a big disadvantage for Linux compared to other OSes?
I don't know enough about threads to have much trust in my own answer.
http://www.faqs.org/faqs/threads-faq/part1/ contains some material on the
difference between various types of threads. If one model of threading were
clearly superior to others, there wouldn't still be so many to choose from,
so I guess it comes down to weighing pros and cons in particular situations.
E.g., from the Blackdown Java FAQ, I gather that Linux not having hybrid
threading is mostly a problem if you need a very large number of threads.
> Doing a "/lib/libc-2.1.3.so" on my potato system shows linuxthreads-0.8 by
> Xavier Leroy
>
> So, I think potato *does* kernel threads, right?
Yes, provided you run a kernel that has the clone() syscall, but that was
introduced long ago, so potato's default kernel should have that already.
> > see e.g. http://www.informatik.hu-berlin.de/~mueller/pthreads/
>
> Is that implementation preferrable to the linuxthreads package that can be
> compiled into glibc? If so, for what reasons?
I'm out of my depth here. You may want to ask that on usenet; e.g. in a post
to comp.programming.threads and comp.os.linux.development.{apps,system} with
a followup-to comp.programming.threads.
HTH,
Ray
--
"My golden rule of computing is reboot your system every morning."
Jon C.A. DeKeles, Technical Director, ZDNet AnchorDesk
in http://www.zdnet.com/anchordesk/story/story_4100.html
Reply to: